2015-10-01T00:36:17Z karswell quit (Read error: Connection reset by peer)
2015-10-01T00:36:30Z karswell joined #sbcl
2015-10-01T00:48:20Z Quadrescence joined #sbcl
2015-10-01T00:54:54Z stassats: started working on disassembly
2015-10-01T00:55:05Z stassats: so that it'd be easier to add/fix the rest of the stuff
2015-10-01T01:02:10Z stassats: and i'm doing it through slime, which makes it much faster
2015-10-01T01:18:33Z kanru` quit (Ping timeout: 265 seconds)
2015-10-01T01:41:11Z joshe joined #sbcl
2015-10-01T01:59:19Z rszeno joined #sbcl
2015-10-01T02:01:33Z rpg joined #sbcl
2015-10-01T02:16:54Z stassats quit (Ping timeout: 255 seconds)
2015-10-01T02:34:09Z nicdev is now known as nicdev_
2015-10-01T03:01:48Z rpg quit (Quit: rpg)
2015-10-01T03:03:27Z psy_ joined #sbcl
2015-10-01T03:11:41Z nikki93 joined #sbcl
2015-10-01T03:19:46Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-01T03:37:16Z psy_ quit (Ping timeout: 265 seconds)
2015-10-01T03:39:04Z stassats joined #sbcl
2015-10-01T03:42:09Z nikki93 joined #sbcl
2015-10-01T03:45:07Z nikki93_ joined #sbcl
2015-10-01T03:46:26Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-01T03:48:20Z _iwc joined #sbcl
2015-10-01T03:52:23Z nikki93_ quit (Remote host closed the connection)
2015-10-01T04:02:13Z _iwc quit (K-Lined)
2015-10-01T04:04:06Z loke: Is it possible to have a non-normalised RATIONAL? Like 4/2 or 4/4?
2015-10-01T04:04:19Z loke: Or even 3/1
2015-10-01T04:04:22Z stassats: nope
2015-10-01T04:04:33Z Bike: the normalization is part of the standard
2015-10-01T04:04:48Z loke: Bike: I know that, which is why I'm asking here and not in #lisp :-)
2015-10-01T04:05:09Z gingerale joined #sbcl
2015-10-01T04:06:08Z loke: I'm just wondering if there is a way in which such a number can be achieved at all, or if these is something in the internal representation that makes it impossible? (what I really would like to know is how SBCL ensures that it can never happen)
2015-10-01T04:06:23Z stassats: it never creates such a thing
2015-10-01T04:06:27Z stassats: i can create it
2015-10-01T04:06:53Z loke: stassats: I see. Is there a specific place in the code where the normalisation happens that I can look at instead of annoying you guys? :-)
2015-10-01T04:07:38Z stassats: anywhere it uses build-ratio
2015-10-01T04:08:05Z loke: build-ratio, which package is it in?
2015-10-01T04:08:15Z stassats: you can figure that on your own
2015-10-01T04:08:26Z loke: fair enough
2015-10-01T04:11:26Z Fare joined #sbcl
2015-10-01T04:15:14Z loke: Thanks for the pointers. I'm starting to understand how rationals are implemented now.
2015-10-01T04:25:24Z nikki93 joined #sbcl
2015-10-01T04:36:10Z Fare quit (Ping timeout: 246 seconds)
2015-10-01T04:44:28Z gingerale quit (Remote host closed the connection)
2015-10-01T04:45:07Z psy_ joined #sbcl
2015-10-01T04:53:54Z rszeno quit (Quit: Leaving.)
2015-10-01T05:12:30Z psy_ quit (Read error: Connection reset by peer)
2015-10-01T05:13:32Z ChanServ has set mode +o Xof
2015-10-01T05:14:45Z Xof is now known as Krystof
2015-10-01T05:36:27Z igajsin1 quit (Quit: Leaving.)
2015-10-01T06:01:52Z Cymew quit (Ping timeout: 272 seconds)
2015-10-01T06:03:30Z Shinmera joined #sbcl
2015-10-01T06:16:13Z karswell quit (Remote host closed the connection)
2015-10-01T06:16:21Z karswell joined #sbcl
2015-10-01T06:32:31Z kanru` joined #sbcl
2015-10-01T07:10:46Z ASau quit (Ping timeout: 240 seconds)
2015-10-01T07:27:04Z Cymew joined #sbcl
2015-10-01T08:03:35Z yvm quit (Ping timeout: 265 seconds)
2015-10-01T08:05:24Z yvm joined #sbcl
2015-10-01T08:47:30Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-01T08:59:26Z sammwch joined #sbcl
2015-10-01T09:01:31Z sammwch quit (Excess Flood)
2015-10-01T09:01:39Z sammwch joined #sbcl
2015-10-01T09:01:46Z heddwch quit (Read error: Connection reset by peer)
2015-10-01T09:02:05Z sammwch is now known as heddwch
2015-10-01T09:09:18Z kanru` quit (Read error: Connection reset by peer)
2015-10-01T09:22:40Z DGASAU joined #sbcl
2015-10-01T09:23:36Z Ettore joined #sbcl
2015-10-01T09:36:53Z nikki93 quit (Ping timeout: 268 seconds)
2015-10-01T09:37:50Z nikki93 joined #sbcl
2015-10-01T09:42:11Z stassats quit (Ping timeout: 265 seconds)
2015-10-01T11:53:50Z hlavaty quit (Remote host closed the connection)
2015-10-01T11:54:07Z hlavaty joined #sbcl
2015-10-01T12:19:29Z nikki93 quit (Remote host closed the connection)
2015-10-01T13:04:42Z psilord quit (Quit: Leaving.)
2015-10-01T13:12:30Z eudoxia joined #sbcl
2015-10-01T13:26:21Z fridim_ joined #sbcl
2015-10-01T13:53:11Z Fare joined #sbcl
2015-10-01T14:01:15Z scymtym_ joined #sbcl
2015-10-01T14:15:09Z flip214 quit (Quit: leaving)
2015-10-01T14:18:01Z flip214 joined #sbcl
2015-10-01T14:18:01Z flip214 quit (Changing host)
2015-10-01T14:18:01Z flip214 joined #sbcl
2015-10-01T14:38:57Z psilord joined #sbcl
2015-10-01T14:39:30Z Cymew quit (Ping timeout: 240 seconds)
2015-10-01T14:45:19Z Fare quit (Ping timeout: 240 seconds)
2015-10-01T15:06:12Z nikki93 joined #sbcl
2015-10-01T15:10:26Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-01T15:26:59Z reb: HEAD sbcl drops into ldb while running :infinite-error-protection
2015-10-01T15:29:53Z gingerale joined #sbcl
2015-10-01T15:33:21Z dougk_: reb: probably my fault, i'll take a look.
2015-10-01T15:44:43Z scymtym_: dougk_: i was going to report this, but didn't get to it yet. if it helps, this is the first batch of commits exhibiting the problem: https://ci.cor-lab.org/job/sbcl-master/1208/
2015-10-01T15:48:27Z eudoxia_ joined #sbcl
2015-10-01T15:49:11Z eudoxia quit (Read error: Connection reset by peer)
2015-10-01T15:49:20Z eudoxia_ quit (Client Quit)
2015-10-01T16:22:46Z dougk_: thanks folks. Stupid mistake, fix pushed.
2015-10-01T16:36:58Z nikki93 joined #sbcl
2015-10-01T16:39:18Z kanru` joined #sbcl
2015-10-01T16:40:55Z nyef joined #sbcl
2015-10-01T16:41:13Z nikki93 quit (Ping timeout: 246 seconds)
2015-10-01T16:51:33Z kanru` quit (Remote host closed the connection)
2015-10-01T16:53:39Z kanru` joined #sbcl
2015-10-01T16:58:04Z psy_ joined #sbcl
2015-10-01T17:14:39Z kanru` quit (Ping timeout: 240 seconds)
2015-10-01T17:20:48Z kanru` joined #sbcl
2015-10-01T17:24:34Z reb: dougk_: thanks!
2015-10-01T18:25:39Z nikki93 joined #sbcl
2015-10-01T18:30:15Z nikki93 quit (Ping timeout: 255 seconds)
2015-10-01T18:44:51Z gingerale quit (Remote host closed the connection)
2015-10-01T18:49:55Z Fare joined #sbcl
2015-10-01T18:59:58Z scymtym__ joined #sbcl
2015-10-01T19:01:22Z scymtym_ quit (Ping timeout: 246 seconds)
2015-10-01T19:17:49Z attila_lendvai joined #sbcl
2015-10-01T19:32:57Z stassats joined #sbcl
2015-10-01T19:39:49Z anthracite joined #sbcl
2015-10-01T19:46:00Z anthracite quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2015-10-01T19:48:47Z Fare quit (Ping timeout: 264 seconds)
2015-10-01T19:49:13Z Bicyclidine joined #sbcl
2015-10-01T20:11:04Z Fare joined #sbcl
2015-10-01T20:12:37Z jsgrant joined #sbcl
2015-10-01T20:14:26Z nikki93 joined #sbcl
2015-10-01T20:19:23Z nikki93 quit (Ping timeout: 264 seconds)
2015-10-01T20:20:59Z anthracite joined #sbcl
2015-10-01T20:28:53Z tifa joined #sbcl
2015-10-01T20:42:25Z lispne-w quit (Remote host closed the connection)
2015-10-01T20:53:18Z logand joined #sbcl
2015-10-01T21:05:32Z anthracite quit (Quit: Textual IRC Client: www.textualapp.com)
2015-10-01T21:17:18Z nikki93 joined #sbcl
2015-10-01T21:23:27Z fridim_ quit (Ping timeout: 244 seconds)
2015-10-01T21:33:43Z ASau joined #sbcl
2015-10-01T21:35:48Z logand quit (Ping timeout: 264 seconds)
2015-10-01T21:48:26Z psy_ quit (Ping timeout: 260 seconds)
2015-10-01T21:53:25Z psy_ joined #sbcl
2015-10-01T21:59:42Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-01T22:07:36Z kanru` quit (Ping timeout: 264 seconds)
2015-10-01T22:30:44Z psilord quit (Quit: Leaving.)
2015-10-01T23:00:59Z nikki93 quit (Remote host closed the connection)
2015-10-01T23:07:54Z jsgrant quit (Ping timeout: 255 seconds)
2015-10-01T23:11:12Z kanru` joined #sbcl
2015-10-01T23:14:03Z Bicyclidine quit (Quit: leaving)
2015-10-01T23:16:04Z Fare quit (Quit: Leaving)
2015-10-01T23:19:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-01T23:27:32Z Ettore quit (Quit: Leaving.)
2015-10-01T23:31:07Z psilord joined #sbcl
2015-10-01T23:44:14Z hydan joined #sbcl
2015-10-02T00:01:52Z nikki93 joined #sbcl
2015-10-02T00:20:48Z ASau quit (Ping timeout: 264 seconds)
2015-10-02T00:28:36Z nicdev_ is now known as sipo
2015-10-02T00:40:36Z hydan quit (Ping timeout: 255 seconds)
2015-10-02T00:45:07Z hydan joined #sbcl
2015-10-02T01:01:05Z hydan quit (Remote host closed the connection)
2015-10-02T01:06:56Z hydan joined #sbcl
2015-10-02T01:20:20Z nikki93 quit (Ping timeout: 265 seconds)
2015-10-02T01:23:27Z nikki93 joined #sbcl
2015-10-02T01:52:43Z scymtym__ quit (Ping timeout: 265 seconds)
2015-10-02T02:07:19Z kanru` quit (Ping timeout: 246 seconds)
2015-10-02T02:20:13Z psy_ quit (Ping timeout: 268 seconds)
2015-10-02T02:43:37Z psy joined #sbcl
2015-10-02T02:44:03Z psy quit (Max SendQ exceeded)
2015-10-02T02:44:36Z psy joined #sbcl
2015-10-02T02:50:45Z nikki93 quit (Read error: No route to host)
2015-10-02T02:51:07Z nikki93 joined #sbcl
2015-10-02T02:56:39Z stassats quit (Ping timeout: 240 seconds)
2015-10-02T03:37:22Z nikki93 quit (Remote host closed the connection)
2015-10-02T03:52:53Z zacts|pi joined #sbcl
2015-10-02T03:53:30Z psy quit (Disconnected by services)
2015-10-02T03:53:56Z psy_ joined #sbcl
2015-10-02T04:01:42Z nikki93 joined #sbcl
2015-10-02T04:03:12Z zacts|pi quit (Ping timeout: 246 seconds)
2015-10-02T04:06:10Z zacts|pi joined #sbcl
2015-10-02T04:41:39Z ASau joined #sbcl
2015-10-02T04:46:01Z stassats joined #sbcl
2015-10-02T04:46:13Z hydan quit (Ping timeout: 256 seconds)
2015-10-02T04:50:45Z stassats quit (Ping timeout: 256 seconds)
2015-10-02T05:09:39Z smokeink joined #sbcl
2015-10-02T05:10:07Z smokeink: hello! could this be a sbcl bug? http://pastecode.ru/8983c/ my locale is LANG=zh_TW.utf8
2015-10-02T05:14:26Z nikki93 quit (Read error: No route to host)
2015-10-02T05:14:44Z nikki93 joined #sbcl
2015-10-02T05:26:30Z zacts|pi quit
2015-10-02T05:56:45Z logand joined #sbcl
2015-10-02T05:58:39Z Krystof: if you're talking about the decoding error: you might have files that have names which are not valid utf-8 bytestrings
2015-10-02T05:59:19Z Krystof: there's a wishlist bug logged somewhere to use a permissive utf-8 encoding for some things
2015-10-02T06:04:26Z smokeink: hmm not sure what was happening, that error is gone now after i ran (ql:register-local-projects)
2015-10-02T06:14:58Z nikki93 quit (Ping timeout: 246 seconds)
2015-10-02T06:16:26Z nikki93 joined #sbcl
2015-10-02T06:18:01Z ASau quit (Ping timeout: 250 seconds)
2015-10-02T06:19:25Z Quadrescence joined #sbcl
2015-10-02T06:22:46Z nikki93_ joined #sbcl
2015-10-02T06:24:14Z smokeink quit (Ping timeout: 272 seconds)
2015-10-02T06:24:53Z smokeink joined #sbcl
2015-10-02T06:25:02Z nikki93 quit (Ping timeout: 268 seconds)
2015-10-02T07:10:50Z Shinmera joined #sbcl
2015-10-02T07:19:00Z fridim_ joined #sbcl
2015-10-02T07:22:04Z scymtym__ joined #sbcl
2015-10-02T07:29:50Z logand quit (Ping timeout: 240 seconds)
2015-10-02T08:13:23Z smokeink quit (Quit: Angelic v4.4 - http://angelic.flexnet.org)
2015-10-02T08:20:54Z Cymew joined #sbcl
2015-10-02T08:40:28Z nikki93_ quit (Remote host closed the connection)
2015-10-02T09:00:37Z heddwch quit (Read error: Connection reset by peer)
2015-10-02T09:00:59Z attila_lendvai joined #sbcl
2015-10-02T09:00:59Z attila_lendvai quit (Changing host)
2015-10-02T09:00:59Z attila_lendvai joined #sbcl
2015-10-02T09:01:10Z heddwch joined #sbcl
2015-10-02T09:14:21Z DGASAU` joined #sbcl
2015-10-02T09:15:11Z karswell` joined #sbcl
2015-10-02T09:17:09Z christop1 joined #sbcl
2015-10-02T09:17:22Z zacts` joined #sbcl
2015-10-02T09:17:45Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-02T09:17:47Z kanru quit (Ping timeout: 250 seconds)
2015-10-02T09:17:49Z zacts quit (Ping timeout: 250 seconds)
2015-10-02T09:17:50Z christoph_debian quit (Ping timeout: 250 seconds)
2015-10-02T09:17:51Z DGASAU quit (Ping timeout: 250 seconds)
2015-10-02T09:17:51Z karswell quit (Ping timeout: 250 seconds)
2015-10-02T09:17:51Z DeadTrickster quit (Ping timeout: 250 seconds)
2015-10-02T09:20:21Z DeadTrickster joined #sbcl
2015-10-02T09:40:29Z loke quit (Remote host closed the connection)
2015-10-02T09:43:29Z loke joined #sbcl
2015-10-02T10:23:49Z Posterdati: please help, I've got this error loading slime: redefining EMACS-INSPECT (#<SB-PCL:SYSTEM-CLASS COMMON-LISP:T>) in DEFMETHOD
2015-10-02T10:24:23Z Posterdati: emacs 24 + sbcl 1.2.15 + slime 2.14 + quicklisp
2015-10-02T10:24:33Z |3b|: doesn't sound like an error
2015-10-02T10:25:22Z Posterdati: yes but it ask me to enter y or no from debugger
2015-10-02T10:25:47Z Posterdati: it is a bit annoying
2015-10-02T10:26:07Z |3b|: are you sure it is from that? what is the prompt when it asks y/n?
2015-10-02T10:26:13Z |3b| is guessing slime/swank version mismatch
2015-10-02T10:26:42Z Posterdati: how can I get swank version?
2015-10-02T10:28:04Z |3b|: if that's the problem it would be in the prompt for that y/n
2015-10-02T10:28:54Z |3b|: but check the filename it loads at the top of *inferior-lisp* and make sure it matches where you expect swank to be
2015-10-02T10:29:16Z Posterdati: "End of file during parsing" while loading the history. Continue? (y or n)
2015-10-02T10:29:35Z Posterdati: what is this?!?!?
2015-10-02T10:29:43Z |3b|: ah, that sounds like your slime history is corrupted
2015-10-02T10:29:54Z Posterdati: ok
2015-10-02T10:30:09Z Posterdati: in quicklisp?
2015-10-02T10:30:55Z |3b|: try renaming ~/.slime-history.eld
2015-10-02T10:32:01Z |3b|: if that doesn't exist or if moving it out of the way doesn't help, ask #lisp, since it doesn't sound #sbcl specific
2015-10-02T10:32:04Z Posterdati: deleted!
2015-10-02T10:32:11Z Posterdati: it was one byte long
2015-10-02T10:32:22Z Posterdati: thanks
2015-10-02T10:33:18Z Posterdati: ah
2015-10-02T10:33:23Z Posterdati: I downloaded latest sbcl
2015-10-02T10:33:26Z Posterdati: 1.2.16
2015-10-02T10:33:33Z Posterdati: and installed in /usr/local
2015-10-02T10:33:42Z Posterdati: as stated in the INSTALL doc
2015-10-02T10:34:06Z Posterdati: was the source copied to in /usr/local/lib/sbcl/src ?
2015-10-02T10:35:20Z Posterdati: is there a way to automate the sbcl installation from the repl?
2015-10-02T10:36:33Z Shinmera: You could just run http://plaster.tymoon.eu/view/85
2015-10-02T10:38:24Z Posterdati: does it compile sbcl from source?
2015-10-02T10:38:44Z Posterdati: and installs sources too?
2015-10-02T10:38:57Z Posterdati: just to use M + .
2015-10-02T10:38:58Z Shinmera: You could answer those questions in a few seconds of looking at it yourself.
2015-10-02T10:39:19Z Posterdati: ok
2015-10-02T10:39:47Z |3b|: if you downloaded a binary, i don't think it includes source. if you built it yourself, it just leaves the source wherever you built it (and look for it there if you go into debugger)
2015-10-02T10:40:19Z |3b|: you can tell it to look for the source somewhere specific if you move it or download it separately, something to do with setting some logical pathname translation
2015-10-02T10:40:39Z |3b| doesn't know the details of that though, since i just build from source and leave it where i built it
2015-10-02T10:41:24Z Shinmera: The trick to that is in the script as well.
2015-10-02T10:41:40Z Posterdati: ok thanks
2015-10-02T10:43:34Z DGASAU` is now known as DGASAU
2015-10-02T10:52:52Z attila_lendvai: there's this proposal, too: https://bugs.launchpad.net/sbcl/+bug/666086 (getting rid of run-sbcl.sh using better defaults)
2015-10-02T11:00:16Z Posterdati: Shinmera: the git version is not the official binary download, it's a beta... Any dangers?
2015-10-02T11:04:10Z Shinmera: You can download manually, check out some commit/tag/branch and launch the script then if you want to make sure.
2015-10-02T11:04:18Z Shinmera: *clone manually
2015-10-02T11:04:43Z Shinmera: Just make sure that the path fits what the script expects. See the BASE_DIR and SOURCE_DIR variables for that.
2015-10-02T11:21:06Z karswell` quit (Read error: Connection reset by peer)
2015-10-02T11:21:18Z karswell` joined #sbcl
2015-10-02T11:30:00Z scymtym: Shinmera: are you aware of SB-EXT:SET-SBCL-SOURCE-LOCATION?
2015-10-02T11:31:33Z Shinmera: scymtym: I am now!
2015-10-02T11:36:51Z stassats joined #sbcl
2015-10-02T12:07:38Z foo` joined #sbcl
2015-10-02T12:07:39Z foo` quit (Remote host closed the connection)
2015-10-02T13:28:17Z stassats: thinking of merging by the end of the week
2015-10-02T13:28:25Z stassats: many things work already
2015-10-02T13:29:29Z stassats: the hope that somebody else has arm64 hardware to test is quite slim
2015-10-02T13:30:42Z nyef: ... does qemu count?
2015-10-02T13:30:58Z stassats: if you're willing to use it
2015-10-02T13:33:03Z eudoxia joined #sbcl
2015-10-02T13:40:46Z chris2: i have a vm
2015-10-02T13:45:40Z chris2: stassats: if you have a bootstrap binary, i can try a native build
2015-10-02T13:45:56Z chris2: hm, perhaps it works with armv7 even
2015-10-02T13:52:50Z chris2: ok, armv7 1.2.14 would run on the box
2015-10-02T14:45:15Z Krystof: Hetzner has ~€10/month ARM64 servers (or €25/month ARM64 servers with worthwhile hard disks)
2015-10-02T14:46:03Z stassats: i probably rather meant testing while using it, not deliberate testing
2015-10-02T14:50:45Z Krystof: yeah
2015-10-02T14:50:59Z attila_lendvai: maybe I was wrong buying my own odroid xu4? they use them for hosting... so, then apparently I have an arm64 box
2015-10-02T14:53:20Z Krystof: heh
2015-10-02T14:53:32Z Krystof: I was just about to say, yes, $79 for an xu4
2015-10-02T14:53:49Z Krystof: not that I really need more arm boards around the house
2015-10-02T14:54:06Z Krystof: (really I need to Do Stuff rather than Grab More Random Hardware)
2015-10-02T14:54:37Z Shinmera: But buying is fun!
2015-10-02T14:55:45Z Krystof: I find it intensely stressful :-)
2015-10-02T14:55:59Z smokeink joined #sbcl
2015-10-02T14:59:13Z attila_lendvai: the amount of time I've wasted on reading up on arm server boards with low consumption... I don't even want to consider, especially that my home server is still not running live and useful stuff
2015-10-02T15:00:01Z smokeink: hello, i've got a bug to report http://pastecode.ru/c70a1/
2015-10-02T15:01:40Z stassats: can you report it properly?
2015-10-02T15:01:51Z stassats: also without involving uiop
2015-10-02T15:01:52Z smokeink: no
2015-10-02T15:02:08Z smokeink: first you'd have to define 'properly'
2015-10-02T15:02:11Z attila_lendvai: smokeink: stassats meant to ask you if you could add it to https://bugs.launchpad.net/sbcl/
2015-10-02T15:03:01Z stassats: if you can't, then don't wonder why it got forgotten
2015-10-02T15:03:12Z attila_lendvai: smokeink: you could (trace sb-ext:save-lisp-and-die) and then extract a direct call without uiop
2015-10-02T15:03:38Z stassats: wouldn't it just map one to one?
2015-10-02T15:03:39Z smokeink: ok
2015-10-02T15:08:28Z |3b|: smokeink: did you try a short path on e: and a long path on C: ?
2015-10-02T15:10:54Z stassats: is there enough space?
2015-10-02T15:11:18Z stassats: permissions?
2015-10-02T15:12:21Z |3b|: and long directory exists, but file doesn't exist or isn't locked by any running process if it does?
2015-10-02T15:17:46Z stassats: |3b|: if that's true, sbcl should handle that better
2015-10-02T15:20:13Z smokeink: hmmmm
2015-10-02T15:20:28Z smokeink: the problem doesn't seem to be free space or permissions or long pathname
2015-10-02T15:21:05Z smokeink: it's only when i write to a network drive that the problem arises
2015-10-02T15:21:50Z stassats: that's better
2015-10-02T15:21:56Z smokeink: eudoxia:\ is a network drive
2015-10-02T15:21:59Z smokeink: e : \
2015-10-02T15:22:42Z smokeink: it's a virtualbox virtual network drive to share stuff with the host
2015-10-02T15:25:46Z |3b| can save-lisp-and-die to a mapped drive from a linux share on win 8.1with 1.2.15.23
2015-10-02T15:27:12Z |3b|: and from 1.2.1 from windows cmd.exe
2015-10-02T15:27:17Z stassats: smokeink: how about creating a simple file?
2015-10-02T15:27:23Z smokeink: https://bugs.launchpad.net/sbcl/+bug/1502201
2015-10-02T15:27:24Z |3b|: can you write to that file from other programs?
2015-10-02T15:27:29Z smokeink: yes
2015-10-02T15:27:46Z stassats: not from other programs
2015-10-02T15:27:48Z smokeink: i created a simple text file from within sbcl, no issue
2015-10-02T15:27:48Z stassats: from sbcl
2015-10-02T15:27:51Z stassats: ok
2015-10-02T15:28:04Z smokeink: lemme try again on a longer path
2015-10-02T15:31:40Z smokeink: for some reason it hanged when i attempted to output a textfile on a longer path
2015-10-02T15:31:57Z smokeink is restarting the vm
2015-10-02T15:33:44Z smokeink: and btw when it attempts to write the core file it creates the .exe file and then it fails when attempting to write. after that the file can't be deleted because it's 'in use'
2015-10-02T15:34:43Z smokeink: i can write text files now
2015-10-02T15:34:47Z smokeink: from within sbcl
2015-10-02T15:35:11Z smokeink: maybe it crashed cuz i didn't properly close the stream
2015-10-02T15:35:17Z smokeink: had a typo
2015-10-02T15:37:31Z smokeink: writing cores still doesn't work: error writing to core file: Invalid argument
2015-10-02T15:37:47Z smokeink: the folder where i attempt to write it exists
2015-10-02T15:37:59Z smokeink: available space 900 gb
2015-10-02T15:38:05Z |3b|: how big is the file it creates when it errors?
2015-10-02T15:38:25Z smokeink: 1.71mb
2015-10-02T15:39:31Z smokeink: i have to go to sleep now, if there's a need i can give you teamviewer access next time, to examine it more in depth
2015-10-02T15:40:03Z smokeink: see you
2015-10-02T15:40:09Z smokeink quit (Quit: $)
2015-10-02T15:40:09Z stassats: is it 32 or 64?
2015-10-02T15:41:14Z stassats: 64
2015-10-02T15:43:23Z stassats: i can save cores fine as well
2015-10-02T15:45:04Z stassats: with that exact same path
2015-10-02T15:46:12Z |3b| wonders if it is specific to virtualbox drives
2015-10-02T15:46:28Z stassats: on virtualbox
2015-10-02T15:46:47Z |3b|: ah
2015-10-02T15:47:07Z |3b|: maybe xp 64 then?
2015-10-02T15:47:15Z stassats: 8.1 here
2015-10-02T15:52:25Z eudoxia_ joined #sbcl
2015-10-02T15:53:11Z eudoxia_ quit (Client Quit)
2015-10-02T15:56:46Z eudoxia quit (Ping timeout: 272 seconds)
2015-10-02T16:01:13Z hlavaty quit (Read error: Connection reset by peer)
2015-10-02T16:07:53Z gingerale joined #sbcl
2015-10-02T16:11:29Z chris2: stassats: is your sbcl/arm64 supposed to build out of the box?
2015-10-02T16:11:50Z stassats: when it will get into the mainline, sure
2015-10-02T16:11:58Z chris2: right now i mean :)
2015-10-02T16:12:36Z chris2: Krystof: attila_lendvai: xu4 is armv7
2015-10-02T16:14:09Z attila_lendvai: chris2: that's what I thought originally, thanks for the clarification
2015-10-02T16:14:59Z chris2: i dont know any bare metal amd64 hosting
2015-10-02T16:15:03Z chris2: *arm64
2015-10-02T16:16:12Z chris2: stassats: with our tree, early-vm.lisp fails with The variable LINKAGE-TABLE-SPACE-START is unbound.
2015-10-02T16:16:55Z stassats: that's expected
2015-10-02T16:16:58Z stassats: just don't build it
2015-10-02T16:24:56Z Krystof: hm, my mistake
2015-10-02T16:25:51Z stassats: (didn't want to commit a fix for linkage-table-space-start because it's outside of the arm64 tree)
2015-10-02T16:29:31Z zacts` quit (Quit: ERC (IRC client for Emacs 24.5.1))
2015-10-02T16:32:06Z gingerale quit (Remote host closed the connection)
2015-10-02T16:38:58Z gingerale joined #sbcl
2015-10-02T16:41:10Z logand joined #sbcl
2015-10-02T16:47:01Z stassats quit (Ping timeout: 256 seconds)
2015-10-02T16:53:12Z stassats joined #sbcl
2015-10-02T17:00:12Z Ettore joined #sbcl
2015-10-02T17:11:19Z logand quit (Ping timeout: 240 seconds)
2015-10-02T17:13:49Z scymtym__ quit (Ping timeout: 246 seconds)
2015-10-02T17:24:51Z Ettore quit (Quit: Leaving.)
2015-10-02T17:29:31Z nikki93 joined #sbcl
2015-10-02T17:42:22Z nikki93 quit (Remote host closed the connection)
2015-10-02T17:57:53Z ASau joined #sbcl
2015-10-02T18:15:06Z gingerale quit (Ping timeout: 240 seconds)
2015-10-02T19:04:19Z Bicyclidine joined #sbcl
2015-10-02T19:06:37Z scymtym_ joined #sbcl
2015-10-02T19:29:34Z karswell` quit (Read error: Connection reset by peer)
2015-10-02T19:29:50Z karswell` joined #sbcl
2015-10-02T20:00:35Z karswell` quit (Read error: Connection reset by peer)
2015-10-02T20:07:17Z S161No3 joined #sbcl
2015-10-02T20:08:13Z S161No3 quit (Quit: Page closed)
2015-10-02T20:32:51Z prxq joined #sbcl
2015-10-02T21:01:58Z Cymew quit (Quit: Konversation terminated!)
2015-10-02T21:41:12Z eudoxia joined #sbcl
2015-10-02T21:44:23Z eudoxia quit (Client Quit)
2015-10-02T21:53:01Z fridim_ quit (Ping timeout: 256 seconds)
2015-10-02T22:05:44Z Ettore joined #sbcl
2015-10-02T22:05:49Z Ettore quit (Read error: Connection reset by peer)
2015-10-02T22:06:11Z Ettore joined #sbcl
2015-10-02T22:06:23Z Ettore quit (Read error: Connection reset by peer)
2015-10-02T22:06:40Z Ettore joined #sbcl
2015-10-02T22:20:52Z karswell` joined #sbcl
2015-10-02T22:30:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-02T22:37:47Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-02T22:55:09Z prxq quit (Remote host closed the connection)
2015-10-02T22:58:20Z Ettore quit (Quit: Leaving.)
2015-10-02T23:44:25Z stassats quit (Ping timeout: 246 seconds)
2015-10-03T00:03:15Z scymtym_ quit (Ping timeout: 255 seconds)
2015-10-03T00:12:35Z nikki93 joined #sbcl
2015-10-03T00:15:43Z rszeno joined #sbcl
2015-10-03T00:42:34Z karswell` is now known as karswell
2015-10-03T01:16:17Z nikki93 quit (Remote host closed the connection)
2015-10-03T01:42:47Z Bicyclidine quit (Quit: adabada)
2015-10-03T02:06:32Z loke quit (Ping timeout: 250 seconds)
2015-10-03T02:06:58Z rszeno quit (Ping timeout: 250 seconds)
2015-10-03T02:08:24Z loke joined #sbcl
2015-10-03T02:12:13Z rszeno joined #sbcl
2015-10-03T02:25:10Z dougk2 joined #sbcl
2015-10-03T04:14:47Z smokeink joined #sbcl
2015-10-03T04:55:48Z nikki93 joined #sbcl
2015-10-03T05:03:23Z dougk2 quit (Ping timeout: 252 seconds)
2015-10-03T05:21:21Z rszeno quit (Quit: Leaving.)
2015-10-03T05:35:57Z nikki93 quit (Ping timeout: 250 seconds)
2015-10-03T06:19:19Z gingerale joined #sbcl
2015-10-03T06:21:35Z dougk2 joined #sbcl
2015-10-03T06:25:43Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-03T06:31:38Z angavrilov quit (Remote host closed the connection)
2015-10-03T06:31:50Z Shinmera joined #sbcl
2015-10-03T06:38:16Z nikki93 joined #sbcl
2015-10-03T06:43:00Z smokeink quit (Ping timeout: 264 seconds)
2015-10-03T06:48:30Z psy_ quit (Ping timeout: 260 seconds)
2015-10-03T07:00:44Z smokeink joined #sbcl
2015-10-03T07:31:03Z nikki93 quit (Remote host closed the connection)
2015-10-03T07:31:26Z nikki93 joined #sbcl
2015-10-03T07:31:27Z nikki93 quit (Remote host closed the connection)
2015-10-03T07:32:10Z psy_ joined #sbcl
2015-10-03T08:02:51Z psy_ quit (Ping timeout: 250 seconds)
2015-10-03T08:27:25Z Ettore joined #sbcl
2015-10-03T09:30:15Z Ettore quit (Quit: Leaving.)
2015-10-03T09:38:25Z mattprelude joined #sbcl
2015-10-03T09:41:22Z mattprelude quit (Client Quit)
2015-10-03T09:48:27Z stassats joined #sbcl
2015-10-03T09:48:33Z stassats quit (Changing host)
2015-10-03T09:48:33Z stassats joined #sbcl
2015-10-03T09:55:31Z rj-code quit (Ping timeout: 256 seconds)
2015-10-03T09:56:44Z rj-code joined #sbcl
2015-10-03T10:11:15Z psy_ joined #sbcl
2015-10-03T10:40:11Z fridim_ joined #sbcl
2015-10-03T11:01:00Z logand joined #sbcl
2015-10-03T11:17:45Z psy_ quit (Remote host closed the connection)
2015-10-03T11:36:44Z stassats quit (Remote host closed the connection)
2015-10-03T11:59:49Z psy_ joined #sbcl
2015-10-03T12:43:09Z mattprelude joined #sbcl
2015-10-03T14:16:30Z angavrilov joined #sbcl
2015-10-03T14:40:51Z dougk2 joined #sbcl
2015-10-03T15:08:54Z karswell quit (Remote host closed the connection)
2015-10-03T15:09:06Z karswell joined #sbcl
2015-10-03T15:12:34Z logand quit (Remote host closed the connection)
2015-10-03T15:26:07Z karswell quit (Read error: Connection reset by peer)
2015-10-03T15:26:21Z karswell joined #sbcl
2015-10-03T15:30:00Z smokeink quit (Remote host closed the connection)
2015-10-03T15:51:36Z stassats joined #sbcl
2015-10-03T16:01:08Z fridim_ quit (Ping timeout: 250 seconds)
2015-10-03T16:08:32Z karswell quit (Read error: Connection reset by peer)
2015-10-03T16:08:45Z karswell joined #sbcl
2015-10-03T16:21:32Z scymtym_ joined #sbcl
2015-10-03T16:41:32Z prxq joined #sbcl
2015-10-03T16:49:23Z psy_ quit (Ping timeout: 264 seconds)
2015-10-03T17:15:06Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-03T17:34:46Z dougk2 joined #sbcl
2015-10-03T17:39:48Z |3b| quit (Read error: Connection reset by peer)
2015-10-03T17:42:21Z |3b| joined #sbcl
2015-10-03T17:48:51Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-03T18:40:31Z zacts joined #sbcl
2015-10-03T18:45:36Z dougk2 joined #sbcl
2015-10-03T18:46:21Z Bike quit (Ping timeout: 250 seconds)
2015-10-03T19:04:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-03T19:56:18Z prxq quit (Read error: Connection reset by peer)
2015-10-03T20:13:19Z Bike joined #sbcl
2015-10-03T20:19:10Z gingerale quit (Ping timeout: 272 seconds)
2015-10-03T20:19:20Z Bike quit (Ping timeout: 268 seconds)
2015-10-03T20:31:18Z Bike joined #sbcl
2015-10-03T21:01:02Z Ettore joined #sbcl
2015-10-03T21:01:06Z Ettore quit (Read error: Connection reset by peer)
2015-10-03T21:01:28Z Ettore joined #sbcl
2015-10-03T21:01:31Z Ettore quit (Read error: Connection reset by peer)
2015-10-03T21:01:59Z Ettore joined #sbcl
2015-10-03T21:02:17Z attila_lendvai joined #sbcl
2015-10-03T21:39:57Z stassats: got linkage table working
2015-10-03T21:47:36Z nyef: Congratulations.
2015-10-03T21:48:07Z stassats: need to resolve large PC offsets for compute-lra/compute-code
2015-10-03T21:48:37Z stassats: arm64 has an ADRP instruction, calculates a 4KB aligned offset from PC
2015-10-03T21:49:06Z stassats: so, it'll require two instruction ADRP + ADD immediate, or ADRP + LDR immediate offset
2015-10-03T21:49:35Z stassats: and then alien callbacks
2015-10-03T21:49:55Z stassats: i'll probably merge things tomorrow, even if i don't get everything working
2015-10-03T21:50:29Z stassats: don't know about threads, this or the next release
2015-10-03T21:59:31Z Ettore quit (Ping timeout: 246 seconds)
2015-10-03T21:59:46Z Ettore joined #sbcl
2015-10-03T22:25:30Z flavioc joined #sbcl
2015-10-03T22:39:44Z smokeink joined #sbcl
2015-10-03T22:40:46Z Ettore quit (Quit: Leaving.)
2015-10-03T22:43:46Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-03T22:47:22Z flavioc quit (Quit: Lost terminal)
2015-10-03T23:11:38Z karswell quit (Read error: Connection reset by peer)
2015-10-03T23:11:53Z karswell joined #sbcl
2015-10-03T23:20:25Z waaron` quit (Read error: Connection reset by peer)
2015-10-03T23:43:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-04T00:03:06Z mattprelude quit (Ping timeout: 240 seconds)
2015-10-04T00:07:48Z stassats quit (Ping timeout: 272 seconds)
2015-10-04T00:51:39Z karswell quit (Read error: Connection reset by peer)
2015-10-04T00:51:52Z karswell joined #sbcl
2015-10-04T01:06:35Z dougk2 joined #sbcl
2015-10-04T01:32:34Z Blkt quit (Quit: No Ping reply in 180 seconds.)
2015-10-04T01:32:50Z Blkt joined #sbcl
2015-10-04T01:55:11Z White_Flame: is there any opinion for/against including quicklisp in the sbcl distribution?
2015-10-04T02:02:30Z scymtym_ quit (Ping timeout: 255 seconds)
2015-10-04T02:09:24Z DeadTrickster quit (Ping timeout: 264 seconds)
2015-10-04T02:27:03Z hydan joined #sbcl
2015-10-04T02:29:50Z salva quit (Ping timeout: 240 seconds)
2015-10-04T02:30:34Z salva joined #sbcl
2015-10-04T02:31:08Z karswell quit (Read error: Connection reset by peer)
2015-10-04T02:31:21Z karswell joined #sbcl
2015-10-04T02:36:45Z hydan quit (Ping timeout: 268 seconds)
2015-10-04T03:08:45Z karswell quit (Remote host closed the connection)
2015-10-04T03:08:53Z karswell joined #sbcl
2015-10-04T03:40:06Z _8hzp quit (Ping timeout: 240 seconds)
2015-10-04T03:51:29Z psy_ joined #sbcl
2015-10-04T03:52:08Z psy_ quit (Remote host closed the connection)
2015-10-04T03:58:46Z dougk2 quit (Ping timeout: 268 seconds)
2015-10-04T03:59:19Z psy_ joined #sbcl
2015-10-04T04:14:26Z _iwc joined #sbcl
2015-10-04T04:26:45Z _iwc quit (K-Lined)
2015-10-04T05:28:36Z gingerale joined #sbcl
2015-10-04T05:47:44Z psy_ quit (Read error: Connection reset by peer)
2015-10-04T06:02:46Z smokeink quit (Ping timeout: 240 seconds)
2015-10-04T06:11:31Z DeadTrickster joined #sbcl
2015-10-04T06:42:39Z smokeink joined #sbcl
2015-10-04T06:51:27Z scymtym_ joined #sbcl
2015-10-04T06:53:02Z Shinmera joined #sbcl
2015-10-04T08:36:56Z |3b| quit (Read error: Connection reset by peer)
2015-10-04T08:39:44Z |3b| joined #sbcl
2015-10-04T08:43:53Z mattprelude joined #sbcl
2015-10-04T09:28:55Z stassats joined #sbcl
2015-10-04T10:48:53Z flavioc joined #sbcl
2015-10-04T11:03:40Z flavioc quit (Ping timeout: 246 seconds)
2015-10-04T11:06:05Z fridim_ joined #sbcl
2015-10-04T11:16:17Z flavioc joined #sbcl
2015-10-04T11:17:28Z flavioc quit (Remote host closed the connection)
2015-10-04T11:33:17Z stassats: trying to test load-from-label with large offsets, and it's hard to generate a lot of code
2015-10-04T11:34:04Z stassats: register allocation slows things down
2015-10-04T11:38:20Z stassats: a vop with (loop repeat X do (inst nop)) is better
2015-10-04T12:31:41Z fikusz quit (Quit: Leaving)
2015-10-04T12:34:22Z fikusz joined #sbcl
2015-10-04T12:56:19Z dougk_ quit (Ping timeout: 240 seconds)
2015-10-04T13:09:26Z dougk_ joined #sbcl
2015-10-04T13:22:37Z stassats: hm, the ADRP instruction gives a 4K aligned offset from PC, but i need to know the remainder of the destination to compute the byte offset
2015-10-04T13:23:23Z stassats: i thought it just applied a 4K aligned offset from PC, but no, the destination needs to be aligned
2015-10-04T13:23:39Z stassats: looks like i'll need something like x86 fixups
2015-10-04T13:24:36Z stassats: or i can compute PC and its alignment first, but that'll take 3 instructions
2015-10-04T13:35:59Z stassats: ADRP would be much more useful if it just calculated offsets from PC
2015-10-04T14:33:09Z dougk2 joined #sbcl
2015-10-04T15:00:06Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-04T15:12:51Z stassats: why even load things from labels, a 32-bit immediate needs just two instructions, and not an 8 byte value plus however many instructions to find it
2015-10-04T15:13:54Z stassats: i just need a fixup spanning multiple instructions
2015-10-04T15:21:16Z nyef: Adding new fixup types isn't particularly hard.
2015-10-04T15:21:33Z yvm quit (Remote host closed the connection)
2015-10-04T15:22:17Z nyef: There's basically three touchpoints. code/target-vm, compiler/target/something (insts?), and compiler/generic/genesis.
2015-10-04T15:22:20Z stassats: load-from-label is only used for runtime functions and assembly routines, those are 32-bit, but are they always?
2015-10-04T15:22:43Z stassats: so, i'll need an emit-chooser to work on fixups
2015-10-04T15:23:30Z stassats: but a fixup length can change after emit-chooser is done
2015-10-04T15:24:26Z stassats: otherwise i'll need four instructions, that's more than the current 8-bytes + one instruction
2015-10-04T15:25:02Z stassats: but when the label is far away, i'll need 8-bytes + three instructions
2015-10-04T15:26:24Z stassats: ADRP + ADD/LDR is nice, but it'll need x86-style code fixups for something that happens rarely
2015-10-04T15:27:55Z nyef: Hrm...
2015-10-04T15:32:01Z stassats: will do the load-from-label thing for now, i can get 35 bits of displacement in 3 instructions
2015-10-04T15:32:54Z stassats: 19-bits from ADR, 16-bits from MOVZ to TMP, then a shift of 3 from ADD or LDR of those 16-bits to clear the 19-bits
2015-10-04T15:49:26Z stassats: but with load-from-label i have to drag around LIP
2015-10-04T15:49:58Z stassats: like for the allocation macro, and it looks like load-from-label is used wrong on ARM32 in ALLOCATION
2015-10-04T15:50:34Z nyef: I didn't do it!
2015-10-04T15:50:39Z stassats: sure
2015-10-04T15:50:41Z nyef: And you can't prove that it was me!
2015-10-04T15:51:23Z stassats: now back to weighing in loading addresses via a chain of instructions and adding LIP to allocation
2015-10-04T15:51:58Z stassats: of course the threaded build will have no problem with accessing boxed_region
2015-10-04T15:53:06Z stassats: how bad would it be to thrash LR without allocating it?
2015-10-04T15:54:45Z yvm joined #sbcl
2015-10-04T15:55:25Z stassats: just using two instructions for loading 32-bit addresses would be a win, but then will cause problems when they are no longer 32-bit
2015-10-04T15:56:10Z yvm quit (Read error: Connection reset by peer)
2015-10-04T15:59:08Z nyef: LR? Just make sure that there's nothing important in it when you start.
2015-10-04T15:59:32Z nyef: It's almost never used to hold anything important at VOP boundaries.
2015-10-04T15:59:54Z nyef: Notable exceptions being function-call and assembly-routine-call scenarios.
2015-10-04T15:59:55Z stassats: there's not many uses of ALLOCATION or WITH-FIXED-ALLOCATION, so i'll just modify everything
2015-10-04T16:06:05Z smokeink quit (Quit: Angelic v4.4 - http://angelic.flexnet.org)
2015-10-04T16:18:41Z smokeink joined #sbcl
2015-10-04T16:40:07Z stassats: another roadblock, the 19 bits of ADDR are sign extended, so i can only use 18 bites, but the left shift in LDR can only be of 3 bits
2015-10-04T16:40:23Z stassats: so i can't shift the 16 bits from MOVZ
2015-10-04T16:41:56Z stassats: oh wait, it's LDR that has 19 bit offsets, ADDR has 21, phew
2015-10-04T16:46:04Z yvm joined #sbcl
2015-10-04T16:53:50Z smokeink quit (Ping timeout: 265 seconds)
2015-10-04T16:54:45Z smokeink joined #sbcl
2015-10-04T17:07:40Z ASau` joined #sbcl
2015-10-04T17:11:16Z ASau quit (Ping timeout: 272 seconds)
2015-10-04T17:24:01Z ASau` is now known as ASau
2015-10-04T18:33:07Z stassats: looks like i got some working callbacks
2015-10-04T18:42:47Z stassats: time to build with sbcl-arm32, then with itself and i'll merge
2015-10-04T18:43:52Z stassats: get ready for a barrage of emails from sbcl-commits
2015-10-04T18:59:19Z foom quit (Ping timeout: 246 seconds)
2015-10-04T19:07:39Z Krystof: yay
2015-10-04T19:12:36Z foom joined #sbcl
2015-10-04T19:21:25Z karswell quit (Remote host closed the connection)
2015-10-04T19:21:31Z karswell` joined #sbcl
2015-10-04T19:24:29Z specbot quit (Disconnected by services)
2015-10-04T19:24:29Z minion quit (Disconnected by services)
2015-10-04T19:24:33Z specbot joined #sbcl
2015-10-04T19:25:14Z minion joined #sbcl
2015-10-04T19:27:26Z jackc- joined #sbcl
2015-10-04T19:27:38Z White__Flame joined #sbcl
2015-10-04T19:28:48Z dsp_ joined #sbcl
2015-10-04T19:30:42Z dim quit (Ping timeout: 260 seconds)
2015-10-04T19:30:43Z yvm quit (Ping timeout: 260 seconds)
2015-10-04T19:30:43Z White_Flame quit (Ping timeout: 260 seconds)
2015-10-04T19:30:44Z alchemis7 quit (Ping timeout: 260 seconds)
2015-10-04T19:32:24Z dsp__ quit (Ping timeout: 240 seconds)
2015-10-04T19:32:25Z jackc-_ quit (Ping timeout: 240 seconds)
2015-10-04T19:32:33Z dim joined #sbcl
2015-10-04T19:52:12Z stassats: the last self-build and committing, some things are not yet done, but there's no point in a separate branch
2015-10-04T19:54:48Z nyef: So, more or less where the original AMD port was when merged?
2015-10-04T19:55:19Z nyef: Err... s/AMD/ARM/
2015-10-04T19:59:40Z stassats: disassembly annotation is notably missing
2015-10-04T19:59:44Z stassats: some tests fail because of that
2015-10-04T20:06:25Z karswell` quit (Read error: Connection reset by peer)
2015-10-04T20:06:41Z karswell` joined #sbcl
2015-10-04T20:14:45Z stassats: some problems with a self build
2015-10-04T20:33:11Z stassats: (type-of nil) => #<MEMBER-TYPE NULL> is not exactly right
2015-10-04T20:33:54Z stassats: but that's not my fault is it?
2015-10-04T20:38:46Z Krystof: probably not
2015-10-04T20:39:27Z stassats: that wasted me some time
2015-10-04T20:39:35Z stassats: self build attempt #2
2015-10-04T20:39:52Z stassats: (and i already have enough of my own bugs to waste time)
2015-10-04T20:47:17Z attila_lendvai joined #sbcl
2015-10-04T20:47:17Z attila_lendvai quit (Changing host)
2015-10-04T20:47:17Z attila_lendvai joined #sbcl
2015-10-04T20:50:01Z leuler joined #sbcl
2015-10-04T20:56:16Z stassats: ok, self built
2015-10-04T20:56:25Z stassats: better push it before something else breaks
2015-10-04T21:00:45Z stassats: except that i broke the x86-64 build
2015-10-04T21:03:26Z stassats: that wasn't a clean tree, phew
2015-10-04T21:03:47Z karswell` is now known as karswell
2015-10-04T21:10:35Z stassats: alright, finally
2015-10-04T21:10:43Z stassats: "Sending one email per commit" oh boy
2015-10-04T21:11:17Z stassats: just 87 commits
2015-10-04T21:16:40Z stassats: and i overtook whn by the number of commits
2015-10-04T21:17:31Z gingerale quit (Remote host closed the connection)
2015-10-04T21:17:53Z leuler: stassats: Congratulations!
2015-10-04T21:18:46Z nyef: Where are those stats (commits per commiter) available?
2015-10-04T21:19:03Z stassats: git shortlog -s -n
2015-10-04T21:19:55Z nyef: Ah, thank you.
2015-10-04T21:20:30Z nyef: Hunh. I'm at 648 to whn's 769.
2015-10-04T21:20:43Z stassats: and SBCL will probably pass the 10,000 commits mark next year
2015-10-04T21:21:02Z nyef: Heh. "The Dow just hit 10,000 again, are you going to throw another party?"
2015-10-04T21:21:07Z Ettore joined #sbcl
2015-10-04T21:22:16Z stassats: damn
2015-10-04T21:22:21Z stassats: i pushed the wrong branch
2015-10-04T21:22:40Z stassats: oh, it's github being silly
2015-10-04T21:22:59Z nyef: I'm sortof surprised at the low number-of-commits attributed to various people that I think of as high-powered hackers involved in maintaining SBCL at various times.
2015-10-04T21:23:00Z stassats: or helpful. confusing for sure
2015-10-04T21:23:19Z stassats: less commits per change in pre-git days?
2015-10-04T21:23:32Z nyef: That's a plausibility.
2015-10-04T21:23:53Z stassats: some were via patches, perhaps
2015-10-04T21:24:14Z nyef: But my local system puts me at number six for total commits, which is kindof shocking for me.
2015-10-04T21:24:40Z stassats: the x86-64 port was just one big commit
2015-10-04T21:24:49Z nyef: Because I don't think of myself as having been particularly active compared to, say, pkhuong or dan?b.
2015-10-04T21:25:13Z nyef: Okay, that would skew the numbers for sure.
2015-10-04T21:25:18Z stassats: (which was done by danb)
2015-10-04T21:25:40Z nyef: And Krystof got the credit for the original Win32 port, IIRC, but I think it also went in as pretty much one commit.
2015-10-04T21:27:18Z nyef: But I'm also ahead of jsnell and lichteblau in number of commits. This is just _weird_.
2015-10-04T21:27:23Z stassats: the linux kernel has 547,000 commits
2015-10-04T21:27:38Z stassats: nyef: i guess your numbers jumped after the arm32 port
2015-10-04T21:27:38Z nyef: I really don't think of myself as being particularly active in SBCL hacking much of the time.
2015-10-04T21:27:47Z stassats: it had a lot of commits
2015-10-04T21:29:05Z nyef: Oh, I've had spates of activity, certainly.
2015-10-04T21:29:18Z nyef: PPC threads was a good 40-60 commits, IIRC, for example.
2015-10-04T21:29:27Z nyef: And then there was the 63-bit fixnum thing.
2015-10-04T21:29:44Z stassats: but it's interesting to see that sbcl overall has continuous activity
2015-10-04T21:29:53Z stassats: compared to other implementations
2015-10-04T21:30:19Z nyef: Usually, yes.
2015-10-04T21:30:47Z nyef: Hrm.
2015-10-04T21:31:06Z Shinmera: Comparing by number of lines changed might give a more satisfactory estimate
2015-10-04T21:31:06Z stassats: https://github.com/sbcl/sbcl/graphs/contributors has a graph, although it doesn't include committers without a github account
2015-10-04T21:31:14Z nyef: I'm actually semi-preparing for another spate of activity. I wonder if I'll pass WHN for commits by SBCL's next anniversary?
2015-10-04T21:31:26Z stassats: what are you planning on doing?
2015-10-04T21:31:43Z nyef: MIPS64/threads.
2015-10-04T21:32:26Z stassats: ok, and i'll do the PPC64 then?
2015-10-04T21:32:43Z stassats: i'm still tied up on arm64 for this month at least
2015-10-04T21:32:46Z nyef: Fine by me, I can't seem to keep a PPC working these days.
2015-10-04T21:33:11Z nyef: Hrm. I almost certainly don't have time to also try to rehab the HPPA backend this year.
2015-10-04T21:33:34Z nyef: Or Alpha, for that matter, though I might have time to get a single-core Alpha machine up and going.
2015-10-04T21:33:43Z stassats: what about OSF/1?
2015-10-04T21:34:09Z nyef: I'm mostly focusing on Linux at this point. Easier to get install environments for.
2015-10-04T21:34:45Z Ettore quit (Quit: Leaving.)
2015-10-04T21:49:55Z leuler: I just bisected the (type-of nil) issue.
2015-10-04T21:49:58Z leuler: The first bad commit could be any of: 20ab508 Fix failure to inline LAYOUT-OF into TYPE-OF, d73dabb Respect NOTINLINE declaration when deciding to inline (SATISFIES PRED), or 1b1e95d Fix self-hosted build problem I just caused.
2015-10-04T21:50:03Z leuler: The first two versions don't build, so bisect can't be more precise.
2015-10-04T21:50:23Z stassats: well, sorry you had to do that, i knew which commit caused it
2015-10-04T21:51:11Z stassats: i should have probably made a test, but i was racing to get a final self build
2015-10-04T21:51:30Z leuler: No harm done, I sometimes like bisecting ;-)
2015-10-04T21:52:32Z leuler: Indeed, I only now see you already fixed it.
2015-10-04T21:53:01Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-04T21:59:39Z leuler quit (Quit: ERC Version 5.1.2 $Revision: 1.796.2.6 $ (IRC client for Emacs))
2015-10-04T22:03:54Z Ettore joined #sbcl
2015-10-04T22:12:10Z znpy joined #sbcl
2015-10-04T22:16:57Z Ettore quit (Quit: Leaving.)
2015-10-04T22:28:10Z fridim_ quit (Ping timeout: 268 seconds)
2015-10-04T22:50:26Z znpy quit (Ping timeout: 260 seconds)
2015-10-04T23:05:02Z mattprelude quit (Ping timeout: 265 seconds)
2015-10-04T23:22:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-04T23:28:36Z dougk2 joined #sbcl
2015-10-04T23:31:36Z alchemis7 joined #sbcl
2015-10-04T23:33:47Z scymtym_ quit (Ping timeout: 264 seconds)
2015-10-04T23:48:33Z stassats quit (Read error: No route to host)
2015-10-04T23:49:16Z stassats joined #sbcl
2015-10-04T23:54:24Z stassats quit (Ping timeout: 264 seconds)
2015-10-05T00:39:06Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-05T00:47:17Z dougk2 joined #sbcl
2015-10-05T01:19:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-05T01:21:42Z dougk2 joined #sbcl
2015-10-05T01:55:50Z sipo is now known as nicdev_
2015-10-05T02:29:23Z karswell quit (Read error: Connection reset by peer)
2015-10-05T02:29:36Z karswell joined #sbcl
2015-10-05T02:39:47Z kanru joined #sbcl
2015-10-05T03:05:15Z DeadTrickster quit (Read error: No route to host)
2015-10-05T03:09:41Z kanru quit (Remote host closed the connection)
2015-10-05T03:17:55Z DeadTrickster joined #sbcl
2015-10-05T03:36:17Z kanru joined #sbcl
2015-10-05T03:53:22Z smokeink quit (Remote host closed the connection)
2015-10-05T03:55:40Z White__Flame is now known as White_Flame
2015-10-05T04:23:24Z smokeink joined #sbcl
2015-10-05T04:44:32Z nyef quit (Ping timeout: 244 seconds)
2015-10-05T04:45:22Z psy_ joined #sbcl
2015-10-05T04:45:47Z psy_ quit (Max SendQ exceeded)
2015-10-05T04:49:24Z psy_ joined #sbcl
2015-10-05T04:50:06Z psy_ quit (Max SendQ exceeded)
2015-10-05T04:50:26Z psy_ joined #sbcl
2015-10-05T04:52:57Z nyef joined #sbcl
2015-10-05T05:13:41Z nyef quit (Ping timeout: 256 seconds)
2015-10-05T05:25:43Z ym joined #sbcl
2015-10-05T05:26:27Z nyef joined #sbcl
2015-10-05T05:44:43Z psy_ quit (Ping timeout: 246 seconds)
2015-10-05T06:39:08Z igajsin1 joined #sbcl
2015-10-05T06:50:36Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-05T06:53:00Z znpy joined #sbcl
2015-10-05T06:57:48Z Cymew joined #sbcl
2015-10-05T06:58:58Z Shinmera joined #sbcl
2015-10-05T07:00:42Z gingerale joined #sbcl
2015-10-05T07:03:05Z Cymew quit (Ping timeout: 252 seconds)
2015-10-05T07:07:12Z znpy quit (Ping timeout: 250 seconds)
2015-10-05T07:28:50Z _8hzp joined #sbcl
2015-10-05T07:30:19Z attila_lendvai joined #sbcl
2015-10-05T07:38:14Z Cymew joined #sbcl
2015-10-05T08:05:06Z Ettore joined #sbcl
2015-10-05T08:16:17Z znpy joined #sbcl
2015-10-05T08:17:26Z ASau quit (Ping timeout: 240 seconds)
2015-10-05T08:27:28Z loke quit (Ping timeout: 246 seconds)
2015-10-05T08:28:19Z angavrilov quit (Ping timeout: 240 seconds)
2015-10-05T08:30:07Z loke joined #sbcl
2015-10-05T08:31:52Z angavrilov joined #sbcl
2015-10-05T08:42:33Z mattprelude joined #sbcl
2015-10-05T08:49:35Z znpy quit (Ping timeout: 256 seconds)
2015-10-05T08:52:37Z fridim_ joined #sbcl
2015-10-05T09:00:33Z heddwch quit (Read error: Connection reset by peer)
2015-10-05T09:01:19Z heddwch joined #sbcl
2015-10-05T09:03:42Z smokeink quit (Remote host closed the connection)
2015-10-05T09:16:03Z smokeink joined #sbcl
2015-10-05T09:19:15Z znpy joined #sbcl
2015-10-05T09:29:46Z karswell quit (Ping timeout: 260 seconds)
2015-10-05T09:53:18Z znpy quit (Ping timeout: 268 seconds)
2015-10-05T10:21:47Z sjl joined #sbcl
2015-10-05T10:47:14Z stassats joined #sbcl
2015-10-05T10:48:42Z znpy joined #sbcl
2015-10-05T10:57:05Z znpy quit (Ping timeout: 256 seconds)
2015-10-05T11:48:37Z DGASAU quit (Remote host closed the connection)
2015-10-05T11:50:08Z DGASAU joined #sbcl
2015-10-05T11:56:18Z fridim_ quit (Ping timeout: 260 seconds)
2015-10-05T12:28:35Z salva quit (Remote host closed the connection)
2015-10-05T12:33:11Z salva joined #sbcl
2015-10-05T12:33:33Z karswell joined #sbcl
2015-10-05T12:40:36Z stassats: did something change again on sourcerforge? my change to platform-table.html is not visible
2015-10-05T13:17:47Z dougk2 joined #sbcl
2015-10-05T13:20:18Z stassats: still not showing the new file
2015-10-05T13:21:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-05T13:26:49Z eudoxia joined #sbcl
2015-10-05T13:32:50Z karswell quit (Ping timeout: 268 seconds)
2015-10-05T13:37:07Z psy_ joined #sbcl
2015-10-05T13:37:32Z psy_ quit (Max SendQ exceeded)
2015-10-05T13:38:13Z psy_ joined #sbcl
2015-10-05T13:39:42Z Krystof: ... just checking.  Did you upload it?
2015-10-05T13:39:49Z stassats: twice
2015-10-05T13:40:12Z stassats: and it's different locally
2015-10-05T13:40:36Z stassats: let me check the destination
2015-10-05T13:41:08Z Krystof: the version at the destination that I got has 1.2.15 for windows
2015-10-05T13:42:04Z stassats: looks like i have some garbage crept in into my cheat sheet
2015-10-05T13:42:20Z smokeink quit (Remote host closed the connection)
2015-10-05T13:42:32Z stassats: works now
2015-10-05T13:43:47Z smokeink joined #sbcl
2015-10-05T14:10:59Z stassats: i'm being asked what's a good bounty for the ppc64 port, and i have no idea
2015-10-05T14:18:11Z nyef: Price it by value, not by cost.
2015-10-05T14:18:42Z stassats: i'll defer the decision, since i don't want the money
2015-10-05T14:19:06Z nyef: That is, don't figure "it takes $n$ hours, and a decent hourly rate for an SBCL hacker is \$$m$, therefore..."
2015-10-05T14:21:12Z flip214: it's IBM paying, so it needn't be *that* accurate ;)
2015-10-05T14:24:06Z stassats: ok, i suggested asking at sbcl-devel@
2015-10-05T14:24:13Z stassats: having to make a decision averted
2015-10-05T14:34:36Z psy_ quit (Ping timeout: 264 seconds)
2015-10-05T15:10:11Z Krystof: stassats: aren't there some extra instructions on ppc64?
2015-10-05T15:10:40Z stassats: haven't checked
2015-10-05T15:14:53Z stassats:  there are additional instructions, but the shared instructions are the same
2015-10-05T15:40:54Z p_l: I think the extra instructions are mostly added in both 32 and 64 bits versions
2015-10-05T15:42:36Z scymtym_ joined #sbcl
2015-10-05T15:50:20Z smokeink quit (Remote host closed the connection)
2015-10-05T15:50:38Z hydan joined #sbcl
2015-10-05T15:54:21Z Fare joined #sbcl
2015-10-05T15:55:07Z Fare: Somehow, my sbcl doesn't get recompiled with sb-dynamic-core, even though I use --with-sb-dynamic-core and it appears positively in local-target-features
2015-10-05T15:56:30Z stassats: it needs some other features, although it would tell you that
2015-10-05T15:56:45Z psy_ joined #sbcl
2015-10-05T15:56:58Z nyef: s/would/should/ ?
2015-10-05T15:57:32Z attila_lendvai: stassats: my 0.02... just get the money and donate it to something you think will make the world a better place. or invest it into the business of a promising young entrepreneur...
2015-10-05T15:59:21Z stassats: i don't really want to bother much
2015-10-05T16:00:12Z Fare: stassats: what other features do I need? where would it tell me that?
2015-10-05T16:00:26Z stassats: it will shout at you
2015-10-05T16:00:29Z stassats: so, that's not really it
2015-10-05T16:00:41Z stassats: for some inexplicable reason it needs sb-thread
2015-10-05T16:03:27Z nyef: The explication is in src/cold/shared.lisp. "We got lazy about updating the allocation sequences."
2015-10-05T16:04:43Z eudoxia_ joined #sbcl
2015-10-05T16:05:02Z stassats: arm64 has a 16-bit immediate operand for the BRK instruction
2015-10-05T16:05:07Z stassats: i can encode all kinds of stuff there
2015-10-05T16:05:38Z stassats: currently trying the trap number and internal-error number
2015-10-05T16:07:59Z eudoxia quit (Ping timeout: 240 seconds)
2015-10-05T16:08:45Z stassats: with some clever packing i can encode trap number, internal-error number and the length of the error arguments
2015-10-05T16:10:21Z eudoxia_ quit (Quit: Leaving)
2015-10-05T16:10:51Z nyef: Be careful with that thing. I tried to use the ARM BKPT instruction for that, only to find out that it basically just dumped the program into an infinite loop.
2015-10-05T16:11:33Z stassats: works so far
2015-10-05T16:12:44Z stassats: will just pack trap and error numbers for now
2015-10-05T16:12:59Z stassats: there's still instruction alignment, so i can't win much
2015-10-05T16:13:12Z stassats: i need to revisit sc-offset encoding and see what can be improved there
2015-10-05T16:14:25Z stassats: packing the trap number shaved 300K from the core
2015-10-05T16:16:08Z nyef: Average savings of about 1.4 instruction-words per trap?
2015-10-05T16:16:46Z stassats: i think most things can get packed into 4 bytes, so removing the trap number saved an additional alignment instruction
2015-10-05T16:27:56Z heddwch is now known as heddvvch
2015-10-05T16:27:59Z heddvvch is now known as heddwch
2015-10-05T16:35:05Z sjl quit (Ping timeout: 252 seconds)
2015-10-05T16:38:48Z stassats: now doing the same as on x86-64, invalid-arg-count-error without encoding nargs-offset
2015-10-05T16:39:31Z hydan quit (Remote host closed the connection)
2015-10-05T16:40:42Z Fare: ok, it was a weird bootstrap problem involving old versions of my multi-call binary script-holder, old versions of cl-launch, old versions of asdf, and old versions of sbcl.
2015-10-05T16:41:10Z Fare: which together conspired to NOT call make.sh with --with-sb-dynamic-core
2015-10-05T16:44:59Z stassats: that saves some more space, but i guess the additional support code doesn't show the difference
2015-10-05T16:45:45Z stassats: (lambda (x) (declare (fixnum x)) x) was 184 bytes, now it's 140
2015-10-05T16:53:29Z stassats: i need to do something with large offsets for branches
2015-10-05T16:54:32Z stassats: conditional branches are ±1MB, uncodintional 128MB
2015-10-05T16:54:44Z stassats: bit-set branches are ±32K
2015-10-05T16:54:57Z stassats: can't use bit-set branches for now
2015-10-05T16:57:59Z stassats: my trusty (sb-c:define-vop (fill) (:generator 1 (loop repeat 5000000 do (sb-assem:inst nop))))
2015-10-05T17:00:14Z stassats: (sb-c:define-vop (fill) (:info count) (:generator 1 (loop repeat count do (sb-assem:inst nop))))
2015-10-05T17:00:15Z stassats: better
2015-10-05T17:31:39Z eudoxia joined #sbcl
2015-10-05T17:52:00Z irsol quit (Ping timeout: 272 seconds)
2015-10-05T17:52:12Z stassats: fixed a couple more issues, now the port looks more or less usable
2015-10-05T18:18:35Z heddwch is now known as can
2015-10-05T18:18:40Z can is now known as heddwch
2015-10-05T18:22:12Z irsol joined #sbcl
2015-10-05T18:59:24Z sjl joined #sbcl
2015-10-05T19:07:55Z Bicyclidine joined #sbcl
2015-10-05T19:07:57Z eudoxia quit (Quit: Leaving)
2015-10-05T19:33:23Z dougk2 joined #sbcl
2015-10-05T19:41:19Z Fare quit (Ping timeout: 250 seconds)
2015-10-05T19:44:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-05T19:46:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-05T19:46:57Z dougk2 joined #sbcl
2015-10-05T20:03:11Z Bicyclidine quit (Ping timeout: 264 seconds)
2015-10-05T20:06:58Z Bicyclidine joined #sbcl
2015-10-05T20:08:30Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-05T20:08:31Z dougk2_ joined #sbcl
2015-10-05T20:15:39Z dougk2_ quit (Ping timeout: 240 seconds)
2015-10-05T20:18:21Z jsnell_ joined #sbcl
2015-10-05T20:20:13Z jsnell quit (Ping timeout: 252 seconds)
2015-10-05T20:20:32Z jsnell_ is now known as jsnell
2015-10-05T20:23:18Z dougk2 joined #sbcl
2015-10-05T20:26:56Z prxq joined #sbcl
2015-10-05T20:29:56Z gingerale quit (Remote host closed the connection)
2015-10-05T20:31:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-05T20:39:22Z Fare joined #sbcl
2015-10-05T20:59:32Z ASau joined #sbcl
2015-10-05T21:04:39Z Fare quit (Ping timeout: 240 seconds)
2015-10-05T21:12:54Z Fare joined #sbcl
2015-10-05T21:19:29Z clique joined #sbcl
2015-10-05T21:26:43Z karswell joined #sbcl
2015-10-05T21:35:30Z prxq quit (Remote host closed the connection)
2015-10-05T21:39:21Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-05T21:58:15Z Bicyclidine quit (Ping timeout: 250 seconds)
2015-10-05T22:04:59Z Bicyclidine joined #sbcl
2015-10-05T22:56:19Z Ettore quit (Quit: Leaving.)
2015-10-05T23:04:15Z clique quit (Ping timeout: 246 seconds)
2015-10-05T23:13:00Z Bicyclidine quit (Quit: leaving)
2015-10-05T23:14:28Z dougk2 joined #sbcl
2015-10-05T23:40:36Z Fare quit (Ping timeout: 246 seconds)
2015-10-06T00:53:39Z scymtym_ quit (Ping timeout: 240 seconds)
2015-10-06T01:02:39Z Fare joined #sbcl
2015-10-06T01:19:22Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-06T01:46:45Z stassats quit (Ping timeout: 256 seconds)
2015-10-06T02:15:40Z npatrick04 joined #sbcl
2015-10-06T02:28:36Z npatrick04 quit (Ping timeout: 255 seconds)
2015-10-06T02:34:13Z White_Flame quit (Ping timeout: 252 seconds)
2015-10-06T03:44:37Z _8hzp quit (Ping timeout: 252 seconds)
2015-10-06T04:05:57Z smokeink joined #sbcl
2015-10-06T04:32:28Z gabriel_laddel joined #sbcl
2015-10-06T05:30:23Z znpy joined #sbcl
2015-10-06T05:31:07Z White_Flame joined #sbcl
2015-10-06T05:56:58Z znpy quit (Ping timeout: 250 seconds)
2015-10-06T05:58:54Z Shinmera joined #sbcl
2015-10-06T06:52:39Z Fare quit (Ping timeout: 240 seconds)
2015-10-06T06:53:04Z ASau quit (Ping timeout: 246 seconds)
2015-10-06T06:53:45Z ASau joined #sbcl
2015-10-06T06:59:10Z gingerale joined #sbcl
2015-10-06T07:32:15Z fridim_ joined #sbcl
2015-10-06T07:49:35Z daimrod joined #sbcl
2015-10-06T08:04:26Z mattprelude quit (Ping timeout: 244 seconds)
2015-10-06T08:14:50Z ASau quit (Ping timeout: 240 seconds)
2015-10-06T08:21:22Z _8hzp joined #sbcl
2015-10-06T08:30:30Z joshe quit (Ping timeout: 246 seconds)
2015-10-06T08:30:54Z joshe joined #sbcl
2015-10-06T08:32:57Z psy_ quit (Read error: Connection reset by peer)
2015-10-06T08:59:39Z psy_ joined #sbcl
2015-10-06T09:00:32Z heddwch quit (Read error: Connection reset by peer)
2015-10-06T09:01:31Z heddwch joined #sbcl
2015-10-06T09:13:16Z smokeink quit (Ping timeout: 250 seconds)
2015-10-06T09:14:10Z smokeink joined #sbcl
2015-10-06T09:16:18Z ralt joined #sbcl
2015-10-06T09:52:46Z Ettore joined #sbcl
2015-10-06T09:53:31Z Ettore quit (Read error: Connection reset by peer)
2015-10-06T09:53:43Z Ettore joined #sbcl
2015-10-06T10:40:53Z fridim_ quit (Ping timeout: 265 seconds)
2015-10-06T10:43:55Z fridim_ joined #sbcl
2015-10-06T10:56:44Z _8hzp quit (Quit: ERC Version 5.3 (IRC client for Emacs))
2015-10-06T11:06:37Z psy_ quit (Ping timeout: 268 seconds)
2015-10-06T11:26:19Z gabriel_laddel quit (Ping timeout: 240 seconds)
2015-10-06T11:29:37Z psy_ joined #sbcl
2015-10-06T12:09:02Z smokeink quit (Remote host closed the connection)
2015-10-06T12:10:23Z smokeink joined #sbcl
2015-10-06T12:53:36Z stassats joined #sbcl
2015-10-06T12:55:30Z eudoxia joined #sbcl
2015-10-06T13:12:38Z dougk2 joined #sbcl
2015-10-06T13:16:49Z Fare joined #sbcl
2015-10-06T13:47:39Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-06T13:59:39Z Fare quit (Ping timeout: 240 seconds)
2015-10-06T14:01:31Z dougk2 joined #sbcl
2015-10-06T14:11:08Z dougk2 quit (Ping timeout: 265 seconds)
2015-10-06T14:12:03Z dougk2 joined #sbcl
2015-10-06T14:25:00Z dougk2 quit (Ping timeout: 264 seconds)
2015-10-06T14:26:01Z dougk2 joined #sbcl
2015-10-06T14:32:27Z psy_ quit (Ping timeout: 244 seconds)
2015-10-06T14:34:00Z dougk2 quit (Ping timeout: 264 seconds)
2015-10-06T14:34:20Z Cymew quit (Ping timeout: 265 seconds)
2015-10-06T14:35:14Z rszeno joined #sbcl
2015-10-06T14:56:34Z sjl quit (Ping timeout: 265 seconds)
2015-10-06T15:14:09Z Fare joined #sbcl
2015-10-06T15:30:43Z dougk2 joined #sbcl
2015-10-06T15:46:00Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-06T15:53:19Z dougk2 joined #sbcl
2015-10-06T15:56:10Z smokeink quit (Quit: peace)
2015-10-06T15:58:45Z Fare quit (Quit: Leaving)
2015-10-06T15:59:16Z DeadTrickster quit (Read error: No route to host)
2015-10-06T15:59:59Z DeadTrickster joined #sbcl
2015-10-06T16:00:26Z psy_ joined #sbcl
2015-10-06T16:01:45Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-06T16:08:43Z eudoxia_ joined #sbcl
2015-10-06T16:09:23Z eudoxia quit (Read error: Connection reset by peer)
2015-10-06T16:29:32Z eudoxia_ quit (Quit: Leaving)
2015-10-06T16:31:41Z dougk2 joined #sbcl
2015-10-06T16:32:38Z jlarocco quit (Read error: Connection reset by peer)
2015-10-06T16:36:08Z dougk2 quit (Ping timeout: 250 seconds)
2015-10-06T17:12:59Z sjl joined #sbcl
2015-10-06T17:26:14Z rszeno quit (Quit: Leaving.)
2015-10-06T17:32:23Z jlarocco joined #sbcl
2015-10-06T17:39:43Z dougk2 joined #sbcl
2015-10-06T18:12:27Z clique joined #sbcl
2015-10-06T18:24:06Z psy_ quit (Ping timeout: 246 seconds)
2015-10-06T18:33:35Z heddwch is now known as ehdwdch
2015-10-06T18:38:32Z gingerale quit (Remote host closed the connection)
2015-10-06T18:39:16Z ehdwdch is now known as heddwch
2015-10-06T18:39:56Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-06T18:41:36Z sjl quit (Ping timeout: 272 seconds)
2015-10-06T18:52:54Z stassats: what about untagged CODE?
2015-10-06T18:53:36Z stassats: that'll allow access to 4096 constants as opposed to 256
2015-10-06T18:55:53Z nyef: Workable, but you'll have to pay a heap-space search to find the start of the object pointed to by $CODE.
2015-10-06T18:55:59Z nyef: For each thread, for each GC.
2015-10-06T18:56:36Z stassats: make it always holding a code-object, make the tag implicit
2015-10-06T18:57:02Z stassats: or even a looser restriction, other-pointer objects
2015-10-06T18:57:42Z stassats: though, 256 constants is a large number
2015-10-06T18:57:59Z attila_lendvai joined #sbcl
2015-10-06T18:58:16Z stassats: it's a bit less, though
2015-10-06T19:00:06Z stassats: ok, it's actually 30 constants
2015-10-06T19:05:06Z stassats: 9 bits of signed offsets
2015-10-06T19:05:19Z stassats: that gives 256 bytes
2015-10-06T19:05:39Z stassats: for untagged 12-bit 8-byte offsets could be used
2015-10-06T19:05:54Z stassats: so, 30 vs 4092
2015-10-06T19:06:42Z stassats: what is an average number of constants?
2015-10-06T19:13:38Z stassats: 7 on average
2015-10-06T19:15:55Z stassats: SB-IMPL::DESCRIBE-FUNCTION has 106 constants
2015-10-06T19:20:50Z stassats: so, for disassemble to properly annotate constants, i would need to reconstruct the value loaded into a temporary register
2015-10-06T19:20:59Z stassats: which can have several different instructions
2015-10-06T19:22:40Z scymtym_ joined #sbcl
2015-10-06T19:23:36Z stassats: but for 8189 constants only one instruction is used, i'll settle for handling only that
2015-10-06T19:30:02Z clique left #sbcl
2015-10-06T19:33:59Z clique joined #sbcl
2015-10-06T19:53:07Z nyef: Problem is, $CODE can occasionally hold function-tagged objects.
2015-10-06T19:53:31Z stassats: there's enough registers to go around
2015-10-06T19:53:52Z nyef: Depends on your architecture.
2015-10-06T19:54:01Z stassats: i'm talking about arm64 only
2015-10-06T19:54:17Z stassats: where there's two kinds of LDR offset encoding
2015-10-06T19:54:39Z stassats: signed and 9-bit byte offsets, and unsigned 12-bit word offsets
2015-10-06T20:02:57Z Bike quit (Read error: No route to host)
2015-10-06T20:03:50Z nyef: Hrm.
2015-10-06T20:05:17Z nyef: $CODE is occasionally permitted to point to something other than a CODE-OBJECT, but I don't think it's permitted to point to arbitrary data, and I rather doubt that it's allowed to be a FIXNUM. Maybe if it has a fixtag, it's declared to be a "native" (untagged) pointer to a CODE-OBJECT?
2015-10-06T20:12:00Z stassats: well, code-object has a other-pointer lowtag, so, it can be any other pointer
2015-10-06T20:12:12Z stassats: not that there's a need in any other other pointers to be there
2015-10-06T20:12:29Z stassats: and looks like i got CODE and NULL annotation working
2015-10-06T20:12:44Z stassats: the disassembly glowed like a christmas tree
2015-10-06T20:13:13Z stassats: that should reduce the number of false positives in tests
2015-10-06T20:13:35Z stassats: because that's how it decided whether something calls a function
2015-10-06T20:13:39Z stassats: not the greatest idea...
2015-10-06T20:13:56Z stassats: just going through constants would be easier
2015-10-06T20:14:56Z Bike joined #sbcl
2015-10-06T20:18:36Z stassats: and i have some broken backtraces
2015-10-06T20:18:47Z stassats: can't find the escaped frame
2015-10-06T20:18:48Z prxq joined #sbcl
2015-10-06T20:26:36Z nyef: ...  Because the $CODE value in the register context is "wrong"?
2015-10-06T20:29:06Z clique quit (Ping timeout: 246 seconds)
2015-10-06T20:38:44Z Krystof quit (Ping timeout: 240 seconds)
2015-10-06T20:39:06Z nicdev_ quit (Read error: Connection reset by peer)
2015-10-06T20:43:46Z foom quit (*.net *.split)
2015-10-06T20:43:46Z dougk_ quit (*.net *.split)
2015-10-06T20:43:49Z chris2 quit (*.net *.split)
2015-10-06T20:43:50Z PuercoPop quit (*.net *.split)
2015-10-06T20:43:53Z brucem quit (*.net *.split)
2015-10-06T20:45:24Z |3b| quit (Excess Flood)
2015-10-06T20:46:18Z nicdev_` joined #sbcl
2015-10-06T20:46:18Z foom joined #sbcl
2015-10-06T20:46:18Z dougk_ joined #sbcl
2015-10-06T20:46:18Z chris2 joined #sbcl
2015-10-06T20:46:18Z PuercoPop joined #sbcl
2015-10-06T20:46:18Z brucem joined #sbcl
2015-10-06T20:46:53Z chris2 quit (Max SendQ exceeded)
2015-10-06T20:48:48Z chris2 joined #sbcl
2015-10-06T20:49:01Z Xof joined #sbcl
2015-10-06T20:53:50Z |3b| joined #sbcl
2015-10-06T20:55:58Z stassats: haven't dissected it
2015-10-06T20:56:06Z stassats: but i need, because it makes debugging hard
2015-10-06T21:00:41Z ASau joined #sbcl
2015-10-06T21:32:03Z fridim_ quit (Ping timeout: 255 seconds)
2015-10-06T21:35:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-06T21:36:45Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-06T21:46:30Z stassats: ok, something's wrong with LRA
2015-10-06T21:46:37Z stassats: which gets into compute-calling-frame
2015-10-06T21:48:51Z stassats: there's no reg_LRA, ok
2015-10-06T21:50:39Z prxq quit (Remote host closed the connection)
2015-10-06T22:09:31Z dougk2 joined #sbcl
2015-10-06T22:22:30Z drmeister quit (Remote host closed the connection)
2015-10-06T22:28:27Z stassats: can't quite understand why here it saves three things https://github.com/sbcl/sbcl/blob/master/src/runtime/interrupt.c#L689 but two here https://github.com/sbcl/sbcl/blob/master/src/runtime/arm-assem.S#L79
2015-10-06T22:28:43Z stassats: and why CSP is increased by two words, and not three
2015-10-06T22:32:09Z stassats: looks like compute-calling-frame accounts for that difference, still looks bizarre to me
2015-10-06T22:33:43Z sjl joined #sbcl
2015-10-06T22:40:03Z stassats: but the increment still bothers me
2015-10-06T22:43:49Z mattprelude joined #sbcl
2015-10-06T22:55:27Z karswell quit (Remote host closed the connection)
2015-10-06T23:08:14Z stassats: ok, isn't it just using pointer arithmetic?
2015-10-06T23:08:19Z nyef: The interrupt.c version is specifically so that it can be found and interpreted by FIND-ESCAPED-FRAME or something like that.
2015-10-06T23:08:32Z stassats: so the 8 that was there before was clearly more than 3 and that's why it never broke?
2015-10-06T23:09:01Z nyef: Probably.
2015-10-06T23:09:10Z nyef: Why is it 2 now?
2015-10-06T23:09:46Z stassats: it was 8
2015-10-06T23:09:59Z stassats: i changed it N_WORD_BYTES * 2
2015-10-06T23:10:30Z stassats: that's the value of 8 everywhere where that wasn't ifdefed out
2015-10-06T23:10:51Z stassats: so, it should just be 3
2015-10-06T23:11:04Z nyef: Mind possible alignment issues.
2015-10-06T23:11:29Z nyef: But yeah, it's clearly a pointer-sized value, and thus is already scaled by N_WORD_BYTES.
2015-10-06T23:11:42Z stassats: the control stack is misaligned without a problem
2015-10-06T23:11:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-06T23:12:41Z stassats: another 8 https://github.com/sbcl/sbcl/blob/master/src/runtime/interrupt.c#L661
2015-10-06T23:13:19Z stassats: which is done after the values are shoved there
2015-10-06T23:13:35Z stassats: though interrupts are claimed to be disabled
2015-10-06T23:14:13Z nyef: About the only thing in its favor.
2015-10-06T23:14:41Z stassats: my problem is that i reach call_into_lisp, and then it's all sideways
2015-10-06T23:14:55Z stassats: it picks up some garbage from somwhere instead of NIL
2015-10-06T23:16:05Z stassats: call_into_lisp is of course suspect, but i'm inspecting it for the twentieth time
2015-10-06T23:19:25Z stassats: so fffc makes a new frame and call_into_lisp makes a new frame
2015-10-06T23:20:51Z nyef: Is this the first call_into_lisp?
2015-10-06T23:21:04Z stassats: the  last
2015-10-06T23:21:07Z dougk2 joined #sbcl
2015-10-06T23:21:08Z nyef: Or from an error context?
2015-10-06T23:21:11Z stassats: after an internal error
2015-10-06T23:21:14Z Ettore quit (Quit: Leaving.)
2015-10-06T23:21:18Z nyef: Hrm.
2015-10-06T23:22:02Z specbot quit (Remote host closed the connection)
2015-10-06T23:22:02Z minion quit (Remote host closed the connection)
2015-10-06T23:22:20Z minion joined #sbcl
2015-10-06T23:23:00Z specbot joined #sbcl
2015-10-06T23:24:14Z stassats: ppc doesn't seem to even allocate a frame in call_into_lisp
2015-10-06T23:24:17Z nyef: call_into_lisp is finding something other than NIL for $NULL ?
2015-10-06T23:25:14Z stassats: compute-calling-frame doesn't look where it should look for NIL
2015-10-06T23:25:24Z stassats: looks like it's 16-byte too high
2015-10-06T23:25:33Z stassats: so it gets a wrong CFP
2015-10-06T23:25:37Z stassats: but how?
2015-10-06T23:25:38Z nyef: Which way around are your stack-grows values?
2015-10-06T23:26:03Z nyef: Umm... Or there's sometimes a frame-pointer bias for some reason, IIRC.
2015-10-06T23:26:06Z stassats: the usual on non-x86, up
2015-10-06T23:26:43Z nyef: PPC probably doesn't allocate a frame because everything fits in registers, and the XEP will do the allocation as long as the CFP is at least set up.
2015-10-06T23:27:15Z nyef: Hrm. Maybe.
2015-10-06T23:27:30Z stassats: time to unsheathe gdb
2015-10-06T23:29:37Z stassats: damn, i forget that gdb doesn't work well with BRK
2015-10-06T23:30:06Z stassats: even after i change the number from 0 to something else, though that might help
2015-10-06T23:30:10Z stassats: stupid gdb
2015-10-06T23:30:24Z nyef: I think that that was one of the pleasant surprises with the ARM port: gdb actually being mostly-usable when I needed it.
2015-10-06T23:31:08Z stassats: it was usable before during cold-init brk got invovled
2015-10-06T23:31:13Z stassats: now it's not usable
2015-10-06T23:31:21Z stassats: maybe i can patch gdb to ignore non-0 brks
2015-10-06T23:33:15Z stassats: or just write my own sane debugger
2015-10-06T23:36:19Z stassats: why doesn't undo_fake_foreign_function_call deallocate the frame? or where does it do that/
2015-10-06T23:37:24Z stassats: i guess it's not strictly necessary
2015-10-06T23:37:25Z nyef: On most platforms, the stack and frame pointers are in the register set.
2015-10-06T23:37:53Z nyef: Note that on ARM, the control stack pointer ISN'T part of the interrupt context, and thus undo_fake DOES reset it.
2015-10-06T23:38:18Z heddwch is now known as indekkusu
2015-10-06T23:39:23Z stassats: i'm not yet returning from call_into_lisp, so i'll shelved that thread
2015-10-06T23:43:10Z drmeister joined #sbcl
2015-10-06T23:44:08Z nyef: I'd ask if there's any chance that the NIL value that call_into_lisp is trying to load gets corrupted somehow, but it's in the text segment, isn't it?
2015-10-06T23:45:52Z stassats: what call_into_lisp does with the NIL value is not really relevant
2015-10-06T23:46:34Z stassats: the NIL produced by build_fake_control_stack_frames is at its place, the place is just wrong
2015-10-06T23:46:59Z scymtym_ quit (Ping timeout: 264 seconds)
2015-10-06T23:47:35Z stassats: ok, how come the frame pointer in internal-error fetched from the context is cut off at 32-bits?
2015-10-06T23:48:01Z stassats: because it's declared unsigned-int?
2015-10-06T23:48:03Z stassats: now that's bad
2015-10-06T23:49:15Z stassats: and was the source of the problem
2015-10-06T23:49:49Z nyef: Once again, the runtime turns out to be a source of bugs. /-:
2015-10-06T23:51:01Z indekkusu is now known as heddwch
2015-10-06T23:53:44Z stassats: ok, testing and shipping
2015-10-06T23:54:05Z stassats: improving disassembly and backtracess: good haul for today
2015-10-06T23:54:30Z stassats: maybe that'll fix sb-sprof
2015-10-06T23:56:50Z stassats: that it did, that it did
2015-10-06T23:57:10Z stassats: no contribs failing now
2015-10-07T00:39:46Z mattprelude quit (Ping timeout: 240 seconds)
2015-10-07T00:58:41Z karswell joined #sbcl
2015-10-07T01:12:37Z Bike quit (Quit: resta)
2015-10-07T01:14:14Z smokeink joined #sbcl
2015-10-07T01:16:20Z psilord quit (Quit: Leaving.)
2015-10-07T01:34:44Z heddwch is now known as yeddwcy
2015-10-07T01:37:05Z psilord joined #sbcl
2015-10-07T01:39:56Z yeddwcy is now known as heddwch
2015-10-07T01:40:25Z jdz quit (Ping timeout: 256 seconds)
2015-10-07T01:46:30Z jdz joined #sbcl
2015-10-07T02:01:27Z Bike joined #sbcl
2015-10-07T02:10:23Z psy_ joined #sbcl
2015-10-07T02:50:52Z gabriel_laddel joined #sbcl
2015-10-07T02:55:46Z npatrick04 joined #sbcl
2015-10-07T03:06:23Z dougk2_ joined #sbcl
2015-10-07T03:07:49Z dougk2 quit (Read error: Connection reset by peer)
2015-10-07T03:32:35Z psy_ quit (Ping timeout: 264 seconds)
2015-10-07T03:46:30Z stassats quit (Ping timeout: 250 seconds)
2015-10-07T03:48:47Z npatrick04 quit (Ping timeout: 250 seconds)
2015-10-07T04:55:10Z _iwc joined #sbcl
2015-10-07T04:57:16Z heddwch is now known as yangzhia
2015-10-07T04:57:59Z Xof quit (Ping timeout: 240 seconds)
2015-10-07T05:02:18Z yangzhia is now known as heddwch
2015-10-07T05:32:39Z psy_ joined #sbcl
2015-10-07T05:59:07Z Shinmera joined #sbcl
2015-10-07T05:59:13Z Xof joined #sbcl
2015-10-07T06:10:04Z dougk2_ quit (Ping timeout: 265 seconds)
2015-10-07T06:55:51Z Ettore joined #sbcl
2015-10-07T07:13:37Z Ettore quit (Quit: Leaving.)
2015-10-07T07:18:23Z fridim_ joined #sbcl
2015-10-07T07:25:59Z psy_ quit (Quit: Leaving)
2015-10-07T07:26:06Z Cymew joined #sbcl
2015-10-07T08:19:23Z ASau quit (Ping timeout: 264 seconds)
2015-10-07T09:00:28Z heddwch quit (Read error: Connection reset by peer)
2015-10-07T09:01:18Z heddwch joined #sbcl
2015-10-07T09:35:04Z hlavaty joined #sbcl
2015-10-07T09:38:45Z Ettore joined #sbcl
2015-10-07T09:55:04Z mprelude joined #sbcl
2015-10-07T10:07:21Z ym quit (Read error: Connection reset by peer)
2015-10-07T10:12:44Z anthracite joined #sbcl
2015-10-07T10:43:39Z smokeink quit (Ping timeout: 240 seconds)
2015-10-07T10:44:36Z smokeink joined #sbcl
2015-10-07T10:52:36Z smokeink quit (Ping timeout: 264 seconds)
2015-10-07T11:02:45Z anthracite quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2015-10-07T11:23:25Z smokeink joined #sbcl
2015-10-07T11:30:49Z ym joined #sbcl
2015-10-07T11:37:11Z smokeink quit (Remote host closed the connection)
2015-10-07T11:40:19Z smokeink joined #sbcl
2015-10-07T11:45:57Z sjl quit (Ping timeout: 252 seconds)
2015-10-07T12:02:11Z anthracite joined #sbcl
2015-10-07T12:10:54Z DeadTrickster quit (Ping timeout: 250 seconds)
2015-10-07T12:19:49Z dougk2 joined #sbcl
2015-10-07T12:23:20Z DeadTrickster joined #sbcl
2015-10-07T12:23:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-07T12:52:01Z Cymew quit (Ping timeout: 268 seconds)
2015-10-07T12:59:31Z karswell quit (Read error: Connection reset by peer)
2015-10-07T12:59:48Z karswell joined #sbcl
2015-10-07T13:04:23Z DGASAU quit (Ping timeout: 256 seconds)
2015-10-07T13:08:42Z DGASAU joined #sbcl
2015-10-07T13:14:54Z stassats joined #sbcl
2015-10-07T13:32:12Z eudoxia joined #sbcl
2015-10-07T13:39:51Z Cymew joined #sbcl
2015-10-07T14:10:25Z Cymew quit (Ping timeout: 252 seconds)
2015-10-07T14:15:55Z scymtym_ joined #sbcl
2015-10-07T14:43:13Z Cymew joined #sbcl
2015-10-07T14:47:33Z anthraci_ joined #sbcl
2015-10-07T14:48:47Z zacts` joined #sbcl
2015-10-07T14:49:55Z anthracite quit (Read error: Connection reset by peer)
2015-10-07T14:49:56Z heddwch quit (Read error: Connection reset by peer)
2015-10-07T14:49:56Z ym quit (Remote host closed the connection)
2015-10-07T14:49:56Z zacts quit (Remote host closed the connection)
2015-10-07T14:49:57Z heddwch joined #sbcl
2015-10-07T14:49:57Z White_Flame quit (Quit: No Ping reply in 180 seconds.)
2015-10-07T14:49:58Z ym joined #sbcl
2015-10-07T14:50:46Z White_Flame joined #sbcl
2015-10-07T14:52:21Z sjl joined #sbcl
2015-10-07T15:16:01Z gabriel_laddel quit (Remote host closed the connection)
2015-10-07T15:17:05Z Cymew quit (Read error: Connection reset by peer)
2015-10-07T15:20:17Z anthraci_ quit (Quit: Textual IRC Client: www.textualapp.com)
2015-10-07T15:39:43Z smokeink quit (Remote host closed the connection)
2015-10-07T15:42:12Z psy_ joined #sbcl
2015-10-07T15:42:43Z psy_ quit (Max SendQ exceeded)
2015-10-07T15:43:08Z psy_ joined #sbcl
2015-10-07T15:45:16Z gingerale joined #sbcl
2015-10-07T16:11:59Z hzp joined #sbcl
2015-10-07T16:13:07Z eudoxia_ joined #sbcl
2015-10-07T16:14:51Z eudoxia quit (Read error: Connection reset by peer)
2015-10-07T16:29:24Z blaaa joined #sbcl
2015-10-07T16:35:30Z sjl quit (Ping timeout: 265 seconds)
2015-10-07T16:51:29Z attila_lendvai joined #sbcl
2015-10-07T17:17:33Z blaaa quit (Ping timeout: 246 seconds)
2015-10-07T17:22:26Z eudoxia_ quit (Quit: Leaving)
2015-10-07T17:38:58Z dougk2 joined #sbcl
2015-10-07T17:52:19Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-07T18:15:15Z stassats: trying to use Store Pair on arm64, where the order is important
2015-10-07T18:15:53Z stassats: it says it's treated as two atomic stores, but i wonder if it really can get interrupted in the middle
2015-10-07T18:16:05Z stassats: with an interrupt that can be handled by sbcl that is
2015-10-07T18:19:59Z mprelude quit (Ping timeout: 240 seconds)
2015-10-07T18:21:19Z stassats: ok, only system error asynchronous interrupt can split an instruction
2015-10-07T18:21:22Z stassats: now to figure what that is
2015-10-07T18:22:22Z stassats: probably something like catching on fire
2015-10-07T18:22:35Z stassats: so, i can safely use store pair and not care about the order
2015-10-07T18:22:37Z dougk2 joined #sbcl
2015-10-07T18:27:32Z stassats: until there's a concurrent gc
2015-10-07T18:29:17Z DGASAU quit (Ping timeout: 252 seconds)
2015-10-07T18:30:19Z stassats: even then it's probably doesn't matter because of store reordering
2015-10-07T18:30:59Z scymtym quit (Ping timeout: 240 seconds)
2015-10-07T18:32:45Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-07T18:33:01Z mprelude joined #sbcl
2015-10-07T18:39:56Z DGASAU joined #sbcl
2015-10-07T18:47:08Z clique joined #sbcl
2015-10-07T18:58:00Z clique quit (Ping timeout: 246 seconds)
2015-10-07T19:01:00Z sjl joined #sbcl
2015-10-07T19:06:25Z stassats: shaved of 512KB by using STP in bind/unbind and rearranging static symbols for better offsets for *bsp* and *paa*
2015-10-07T19:14:20Z flavioc joined #sbcl
2015-10-07T19:15:41Z dougk2 joined #sbcl
2015-10-07T19:21:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-07T19:25:22Z gingerale- joined #sbcl
2015-10-07T19:26:17Z Shinmera- joined #sbcl
2015-10-07T19:26:17Z Shinmera quit (Read error: Connection reset by peer)
2015-10-07T19:28:19Z gingerale quit (Ping timeout: 259 seconds)
2015-10-07T19:33:55Z stassats: transformed the two loades in return-single into LDP, that saved just 60K, but it's pretty ugly and i need something generic that can coalesce loads and stores
2015-10-07T19:38:34Z gingerale- quit (Remote host closed the connection)
2015-10-07T20:05:28Z Shinmera- is now known as Shinmera
2015-10-07T20:23:12Z fridim_ quit (Ping timeout: 264 seconds)
2015-10-07T20:40:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-07T20:42:30Z angavrilov quit (Remote host closed the connection)
2015-10-07T20:47:40Z angavrilov joined #sbcl
2015-10-07T20:52:03Z angavrilov quit (Remote host closed the connection)
2015-10-07T20:57:12Z ASau joined #sbcl
2015-10-07T21:03:58Z angavrilov joined #sbcl
2015-10-07T21:07:30Z angavrilov quit (Read error: Connection reset by peer)
2015-10-07T21:07:58Z dougk2 joined #sbcl
2015-10-07T21:12:28Z prxq joined #sbcl
2015-10-07T21:13:00Z angavrilov joined #sbcl
2015-10-07T21:17:33Z angavrilov quit (Remote host closed the connection)
2015-10-07T21:19:21Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-07T21:24:44Z prxq quit (Remote host closed the connection)
2015-10-07T21:25:21Z angavrilov joined #sbcl
2015-10-07T21:29:33Z angavrilov quit (Remote host closed the connection)
2015-10-07T21:35:54Z angavrilov joined #sbcl
2015-10-07T21:41:12Z angavrilov quit (Read error: Connection reset by peer)
2015-10-07T21:50:16Z angavrilov joined #sbcl
2015-10-07T21:54:24Z angavrilov quit (Read error: Connection reset by peer)
2015-10-07T22:02:51Z angavrilov joined #sbcl
2015-10-07T22:13:53Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-07T22:16:05Z stassats: still having trouble with the debugger, the same things as yesterday but now with undefined functions
2015-10-07T22:16:53Z angavrilov quit (Ping timeout: 268 seconds)
2015-10-07T22:18:25Z nyef: stassats: Well, you know where to start: undefined_tramp.
2015-10-07T22:19:02Z stassats: but it doesn't touch stack...
2015-10-07T22:19:53Z stassats: it sets up reg_CODE, does that direct stack traversal?
2015-10-07T22:20:45Z stassats: but reg_CODE looks strange indeed
2015-10-07T22:21:16Z nyef: Also cross-check with the mechanism that the debugger uses to determine undefined-function frames.
2015-10-07T22:21:32Z nyef: The oddly-bogus reg_CODE value is significant, IIRC.
2015-10-07T22:22:16Z stassats: header with a function lowtag does seem strange
2015-10-07T22:24:24Z nyef: Actually, $CODE pointing to a function header is seen elsewhere on some platforms.
2015-10-07T22:27:06Z angavrilov joined #sbcl
2015-10-07T22:30:21Z stassats: that looks like it
2015-10-07T22:32:39Z stassats: ok, next problem is somehow THROW and CATCH end up feeding LRA to APPEND2
2015-10-07T22:35:15Z stassats: ok, no, undefined_tramp apperars to be still not working
2015-10-07T22:36:41Z stassats: which means, where was i looking?
2015-10-07T22:36:46Z stassats: probably wrong slime connection
2015-10-07T22:40:11Z stassats: frame-down somehow gets the same kind of garbage when looking for LRA
2015-10-07T22:40:12Z flavioc quit (Ping timeout: 246 seconds)
2015-10-07T22:42:15Z dougk2 joined #sbcl
2015-10-07T22:45:57Z Ettore quit (Quit: Leaving.)
2015-10-07T22:47:28Z angavrilov quit (Ping timeout: 250 seconds)
2015-10-07T22:53:22Z angavrilov joined #sbcl
2015-10-07T23:13:18Z angavrilov quit (Remote host closed the connection)
2015-10-07T23:16:53Z angavrilov joined #sbcl
2015-10-07T23:17:28Z stassats: ok, undefined_tramp wasn't double word aligned, but that still doesn't solve the problem
2015-10-07T23:20:57Z scymtym_ quit (Ping timeout: 255 seconds)
2015-10-07T23:23:26Z mprelude quit (Ping timeout: 250 seconds)
2015-10-07T23:27:19Z angavrilov quit (Ping timeout: 240 seconds)
2015-10-07T23:28:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-07T23:36:05Z angavrilov joined #sbcl
2015-10-07T23:49:06Z angavrilov quit (Remote host closed the connection)
2015-10-07T23:53:36Z angavrilov joined #sbcl
2015-10-07T23:57:06Z stassats: arm64 was missing from https://github.com/sbcl/sbcl/blob/master/src/code/debug-int.lisp#L2092
2015-10-07T23:57:39Z stassats: it probably was at (or sparc arm) previously
2015-10-08T00:03:59Z angavrilov quit (Ping timeout: 240 seconds)
2015-10-08T00:08:53Z nyef: That would do it.
2015-10-08T00:09:57Z stassats: now passing debug.impure tests
2015-10-08T00:10:07Z stassats: save for trace encapsulate, which fail on arm32 too
2015-10-08T00:10:28Z stassats: now for that throw/catch mystery
2015-10-08T00:11:21Z stassats: https://github.com/sbcl/sbcl/blob/master/src/code/debug-int.lisp#L991 is not great
2015-10-08T00:11:57Z stassats: neither is https://github.com/sbcl/sbcl/blob/master/src/code/debug-int.lisp#L2466
2015-10-08T00:13:40Z nyef: Mmm. Well played. Probably survived this long because the Alpha backend isn't a true 64-bit port.
2015-10-08T00:14:27Z angavrilov joined #sbcl
2015-10-08T00:14:36Z stassats: this alpha silliness is annoying
2015-10-08T00:14:49Z stassats: if it's 32-bit, why not make it really look like 32-bit?
2015-10-08T00:14:59Z stassats: but no, constantly sap-ref-32 just for alpha
2015-10-08T00:17:35Z nyef: It's on my list to fix up, just not at the top of my list to fix up.
2015-10-08T00:19:14Z stassats: ok, shaved about a megabyte off the core, fixed undefined functions and undefined alien backtraces
2015-10-08T00:19:17Z stassats: a good place to stop
2015-10-08T00:20:44Z stassats: i think i'll continue focusing on polishing this release cycle, will do threads next month
2015-10-08T00:23:52Z stassats: and i'm too lazy to backport optimizations to arm32
2015-10-08T00:24:25Z dougk2 joined #sbcl
2015-10-08T00:30:04Z angavrilov quit (Write error: Connection reset by peer)
2015-10-08T00:36:38Z angavrilov joined #sbcl
2015-10-08T00:39:19Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-08T00:43:24Z rszeno joined #sbcl
2015-10-08T00:48:39Z angavrilov quit (Ping timeout: 246 seconds)
2015-10-08T00:53:46Z dougk2 joined #sbcl
2015-10-08T00:58:28Z angavrilov joined #sbcl
2015-10-08T01:04:19Z eschatologist joined #sbcl
2015-10-08T01:07:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-08T01:12:04Z angavrilov quit (Remote host closed the connection)
2015-10-08T01:15:27Z angavrilov joined #sbcl
2015-10-08T01:16:52Z igajsin1 quit (Ping timeout: 246 seconds)
2015-10-08T01:19:05Z igajsin1 joined #sbcl
2015-10-08T01:19:27Z dougk2 joined #sbcl
2015-10-08T01:28:50Z angavrilov quit (Ping timeout: 240 seconds)
2015-10-08T01:30:06Z igajsin1 quit (Ping timeout: 255 seconds)
2015-10-08T01:37:58Z angavrilov joined #sbcl
2015-10-08T01:39:10Z momo-reina joined #sbcl
2015-10-08T01:39:50Z igajsin1 joined #sbcl
2015-10-08T01:45:00Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-08T01:46:16Z dougk2 joined #sbcl
2015-10-08T01:54:03Z angavrilov quit (Ping timeout: 252 seconds)
2015-10-08T02:03:19Z momo-reina quit (Ping timeout: 240 seconds)
2015-10-08T02:04:19Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-08T02:04:32Z angavrilov joined #sbcl
2015-10-08T02:05:54Z smokeink joined #sbcl
2015-10-08T02:20:37Z angavrilov quit (Remote host closed the connection)
2015-10-08T02:24:27Z angavrilov joined #sbcl
2015-10-08T02:26:03Z dougk2 joined #sbcl
2015-10-08T02:33:18Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-08T02:34:33Z dougk2 joined #sbcl
2015-10-08T02:49:37Z angavrilov quit (Read error: Connection reset by peer)
2015-10-08T02:53:35Z eschatologist quit (Quit: Textual IRC Client: www.textualapp.com)
2015-10-08T02:53:58Z angavrilov joined #sbcl
2015-10-08T02:58:30Z dougk2_ joined #sbcl
2015-10-08T03:00:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-08T03:10:25Z angavrilov quit (Ping timeout: 256 seconds)
2015-10-08T03:13:00Z dougk2_ quit (Ping timeout: 264 seconds)
2015-10-08T03:20:30Z angavrilov joined #sbcl
2015-10-08T03:22:43Z dougk2 joined #sbcl
2015-10-08T03:26:06Z angavrilov quit (Remote host closed the connection)
2015-10-08T03:30:27Z angavrilov joined #sbcl
2015-10-08T03:35:15Z psy_ quit (Ping timeout: 252 seconds)
2015-10-08T03:53:45Z angavrilov quit (Ping timeout: 244 seconds)
2015-10-08T04:01:28Z angavrilov joined #sbcl
2015-10-08T04:09:01Z tifa quit (Ping timeout: 268 seconds)
2015-10-08T04:10:13Z tifa joined #sbcl
2015-10-08T04:20:49Z angavrilov quit (Ping timeout: 250 seconds)
2015-10-08T04:28:36Z angavrilov joined #sbcl
2015-10-08T04:32:17Z karswell quit (Remote host closed the connection)
2015-10-08T04:32:21Z jfe_ joined #sbcl
2015-10-08T04:32:34Z karswell joined #sbcl
2015-10-08T04:34:11Z angavrilov quit (Remote host closed the connection)
2015-10-08T04:40:03Z angavrilov joined #sbcl
2015-10-08T04:43:37Z angavrilov quit (Read error: Connection reset by peer)
2015-10-08T04:55:41Z angavrilov joined #sbcl
2015-10-08T04:59:33Z angavrilov quit (Remote host closed the connection)
2015-10-08T05:10:29Z angavrilov joined #sbcl
2015-10-08T05:26:50Z angavrilov quit (Ping timeout: 240 seconds)
2015-10-08T05:35:40Z angavrilov joined #sbcl
2015-10-08T05:52:30Z angavrilov quit (Remote host closed the connection)
2015-10-08T05:53:38Z dougk2 quit (Ping timeout: 260 seconds)
2015-10-08T05:57:11Z Shinmera joined #sbcl
2015-10-08T05:57:17Z angavrilov joined #sbcl
2015-10-08T06:16:07Z angavrilov quit (Ping timeout: 246 seconds)
2015-10-08T06:20:02Z dougk2 joined #sbcl
2015-10-08T06:20:50Z ASau quit (Ping timeout: 240 seconds)
2015-10-08T06:23:22Z angavrilov joined #sbcl
2015-10-08T06:27:10Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-08T06:37:57Z angavrilov quit (Ping timeout: 246 seconds)
2015-10-08T06:43:34Z stassats quit (Ping timeout: 260 seconds)
2015-10-08T06:48:25Z angavrilov joined #sbcl
2015-10-08T07:07:22Z angavrilov quit (Ping timeout: 260 seconds)
2015-10-08T07:09:09Z angavrilov joined #sbcl
2015-10-08T07:23:11Z scymtym joined #sbcl
2015-10-08T07:24:50Z fridim_ joined #sbcl
2015-10-08T07:24:51Z igajsin1 quit (Read error: Connection reset by peer)
2015-10-08T07:25:30Z rszeno quit (Quit: Leaving.)
2015-10-08T07:25:33Z igajsin1 joined #sbcl
2015-10-08T07:26:48Z Cymew joined #sbcl
2015-10-08T08:06:10Z Ettore joined #sbcl
2015-10-08T08:19:19Z tifa quit (Ping timeout: 240 seconds)
2015-10-08T08:24:22Z zacts` quit (Ping timeout: 260 seconds)
2015-10-08T08:27:22Z attila_lendvai joined #sbcl
2015-10-08T08:31:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-08T09:00:33Z heddwch quit (Read error: Connection reset by peer)
2015-10-08T09:01:18Z heddwch joined #sbcl
2015-10-08T09:47:25Z smokeink quit (Ping timeout: 265 seconds)
2015-10-08T09:57:41Z smokeink_ joined #sbcl
2015-10-08T10:09:50Z smokeink_ quit (Ping timeout: 260 seconds)
2015-10-08T10:10:07Z smokeink joined #sbcl
2015-10-08T10:24:56Z gabriel_laddel joined #sbcl
2015-10-08T10:26:49Z nikki93 joined #sbcl
2015-10-08T10:36:51Z nikki93 quit (Remote host closed the connection)
2015-10-08T10:52:46Z smokeink quit (Ping timeout: 244 seconds)
2015-10-08T11:19:32Z gabriel_laddel quit (Remote host closed the connection)
2015-10-08T12:15:34Z dougk2 joined #sbcl
2015-10-08T12:26:12Z dougk2 quit (Ping timeout: 264 seconds)
2015-10-08T12:30:26Z dougk2 joined #sbcl
2015-10-08T12:36:39Z flip214 quit (Ping timeout: 240 seconds)
2015-10-08T12:37:44Z flip214 joined #sbcl
2015-10-08T12:39:57Z eudoxia joined #sbcl
2015-10-08T12:44:14Z dougk2 quit (Ping timeout: 250 seconds)
2015-10-08T13:01:20Z dougk2 joined #sbcl
2015-10-08T13:05:28Z smokeink joined #sbcl
2015-10-08T13:13:30Z smokeink quit (Ping timeout: 240 seconds)
2015-10-08T13:15:07Z smokeink joined #sbcl
2015-10-08T13:36:41Z stassats joined #sbcl
2015-10-08T13:43:42Z dougk2 quit (Ping timeout: 272 seconds)
2015-10-08T13:54:00Z smokeink quit (Ping timeout: 250 seconds)
2015-10-08T13:55:25Z smokeink joined #sbcl
2015-10-08T14:06:54Z dougk2 joined #sbcl
2015-10-08T14:07:59Z smokeink quit (Ping timeout: 264 seconds)
2015-10-08T14:08:46Z smokeink joined #sbcl
2015-10-08T14:18:57Z psy_ joined #sbcl
2015-10-08T14:22:01Z smokeink quit (Ping timeout: 244 seconds)
2015-10-08T14:23:20Z smokeink joined #sbcl
2015-10-08T14:28:55Z smokeink quit (Ping timeout: 246 seconds)
2015-10-08T14:29:11Z smokeink joined #sbcl
2015-10-08T14:29:23Z Cymew quit (Ping timeout: 252 seconds)
2015-10-08T14:32:41Z psy_ quit (Ping timeout: 250 seconds)
2015-10-08T14:34:18Z smokeink quit (Ping timeout: 246 seconds)
2015-10-08T14:35:34Z smokeink joined #sbcl
2015-10-08T14:58:02Z attila_lendvai joined #sbcl
2015-10-08T15:00:18Z smokeink quit (Ping timeout: 250 seconds)
2015-10-08T15:00:38Z smokeink joined #sbcl
2015-10-08T15:05:09Z smokeink quit (Ping timeout: 268 seconds)
2015-10-08T15:21:51Z dougk2 quit (Remote host closed the connection)
2015-10-08T15:32:22Z gingerale joined #sbcl
2015-10-08T15:44:44Z smokeink joined #sbcl
2015-10-08T15:59:50Z fikusz quit (Quit: Leaving)
2015-10-08T16:02:28Z fikusz joined #sbcl
2015-10-08T16:03:06Z heddwch quit (Quit: ZNC - 1.6.0 - http://znc.in)
2015-10-08T16:04:22Z heddwch joined #sbcl
2015-10-08T16:15:16Z Ettore quit (Quit: Leaving.)
2015-10-08T16:16:59Z eudoxia_ joined #sbcl
2015-10-08T16:18:26Z Ettore joined #sbcl
2015-10-08T16:20:55Z eudoxia quit (Ping timeout: 256 seconds)
2015-10-08T16:24:42Z smokeink quit (Ping timeout: 255 seconds)
2015-10-08T16:25:49Z smokeink joined #sbcl
2015-10-08T16:30:47Z eudoxia_ quit (Quit: Leaving)
2015-10-08T16:32:27Z psy_ joined #sbcl
2015-10-08T17:30:21Z smokeink quit (Ping timeout: 246 seconds)
2015-10-08T17:30:39Z stassats: using STP instead of two STR for zeroing stack allocated vectors is twice as fast
2015-10-08T17:37:57Z nikki93 joined #sbcl
2015-10-08T18:28:26Z nikki93 quit (Remote host closed the connection)
2015-10-08T18:39:23Z nikki93 joined #sbcl
2015-10-08T18:51:50Z karswell quit (Read error: Connection reset by peer)
2015-10-08T18:52:04Z karswell joined #sbcl
2015-10-08T18:56:29Z stassats quit (Read error: Connection reset by peer)
2015-10-08T19:10:22Z scymtym_ joined #sbcl
2015-10-08T19:31:46Z eudoxia joined #sbcl
2015-10-08T19:31:58Z nikki93 quit (Remote host closed the connection)
2015-10-08T19:53:59Z stassats joined #sbcl
2015-10-08T19:54:35Z stassats: doing some unbind-to-here benchmarking, and it looks that it's actually faster to throw than to unbind all the special manually
2015-10-08T19:54:40Z stassats: at least with a 100 specials
2015-10-08T19:56:24Z stassats: so, that's a potential optimization, unbind a lot of specials in a loop
2015-10-08T19:57:55Z smokeink joined #sbcl
2015-10-08T19:58:41Z nyef: Which is odd, because THROW/CATCH uses UNBIND-TO-HERE to do the unbinding?
2015-10-08T19:59:02Z stassats: that's my point
2015-10-08T19:59:03Z nyef: Or is it that the UNBIND operation is slow?
2015-10-08T19:59:11Z stassats: unbind-to-here is faster than normal execution
2015-10-08T19:59:20Z stassats: after some number of specials
2015-10-08T19:59:37Z nyef: That... shouldn't come as a surprise.
2015-10-08T19:59:55Z stassats: for each binding it loads BSP
2015-10-08T19:59:56Z nyef: Normal operation acts as an unrolled loop.
2015-10-08T20:00:10Z nyef: Unbind-to-here is the rolled version.
2015-10-08T20:00:12Z stassats: so, if at least that part was done once it would be better
2015-10-08T20:04:51Z tifa joined #sbcl
2015-10-08T20:06:24Z stassats: usage of LDP/STP didn't speed up unbind-to-here
2015-10-08T20:06:28Z stassats: but it made it more compact
2015-10-08T20:09:03Z stassats: maybe very marginally faster
2015-10-08T20:12:17Z flavioc joined #sbcl
2015-10-08T20:13:27Z tifa quit (Ping timeout: 246 seconds)
2015-10-08T20:31:19Z stassats: quite interestingly, 32-bit sbcl writes '(signed-byte 64) streams big endian
2015-10-08T20:31:25Z stassats: while 64-bit writes little-endian
2015-10-08T20:35:23Z gingerale quit (Remote host closed the connection)
2015-10-08T20:45:59Z nyef: Eh? And it's not endian-dependent?
2015-10-08T20:46:31Z stassats: i think it's using something handcoded outside of native formats
2015-10-08T20:47:05Z nyef: Yet more evidence of the fundamental sanity of my usual approach, using (unsigned-byte 8) streams and encoding anything else from there.
2015-10-08T20:47:29Z stassats: well, sure, anything else is only portable between itself
2015-10-08T20:47:58Z stassats: i rolled out my own streams, which can do 8, 16, 32 and 64-bit access at the same time
2015-10-08T20:48:40Z stassats: so, being stack to a stream being always one size from which it's not easy to produce things is not really useful
2015-10-08T20:48:53Z stassats: (unsigned-byte 64) is only reserved for quick and dirty hacks
2015-10-08T20:49:56Z stassats: ok, another test failure crossed
2015-10-08T20:50:14Z stassats: catch/throw and a pc-return object eludes
2015-10-08T20:50:31Z stassats: somehow it gets where it shouldn't be, and as a value it passes an LRA
2015-10-08T20:50:42Z stassats: but catch/throw in general work
2015-10-08T20:52:22Z stassats: gdb would be really helpful, but alas it doesn't like BRK
2015-10-08T20:53:10Z smokeink quit (Ping timeout: 240 seconds)
2015-10-08T20:53:29Z smokeink joined #sbcl
2015-10-08T20:56:51Z ASau joined #sbcl
2015-10-08T20:58:14Z stassats: lldb doesn't seem to be ported to aarch64, at least on debian
2015-10-08T21:00:24Z smokeink quit (Ping timeout: 264 seconds)
2015-10-08T21:01:01Z smokeink joined #sbcl
2015-10-08T21:07:02Z smokeink quit (Ping timeout: 272 seconds)
2015-10-08T21:07:14Z Quadrescence joined #sbcl
2015-10-08T21:13:37Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-08T21:18:34Z nyef: And writing your own debugger would be a pain, of course.
2015-10-08T21:19:23Z smokeink joined #sbcl
2015-10-08T21:19:44Z stassats: after looking at the gdb sources, certainly easier than understanding that
2015-10-08T21:22:11Z flavioc quit (Ping timeout: 250 seconds)
2015-10-08T21:22:14Z nyef: There is that.
2015-10-08T21:22:43Z eudoxia quit (Quit: Leaving)
2015-10-08T21:22:45Z nyef: I have a proof-of-concept for using ptrace on x86-64 SBCL, but that doesn't exactly help.
2015-10-08T21:22:49Z stassats: it's probably just returning into the wrong address
2015-10-08T21:23:03Z nyef: Very plausibly.
2015-10-08T21:23:31Z stassats: but why does it work when i test it?
2015-10-08T21:24:46Z smokeink quit (Ping timeout: 240 seconds)
2015-10-08T21:28:39Z stassats: huh, VISC processors?
2015-10-08T21:29:27Z stassats: looking at http://softmachines.com/soft-machines-unveils-visc-processor-and-soc-roadmap/
2015-10-08T21:30:27Z nikki93 joined #sbcl
2015-10-08T21:37:41Z smokeink joined #sbcl
2015-10-08T21:40:45Z pkhuong: virtual hardware threading
2015-10-08T21:53:34Z nyef: stassats: Still stuck with the catch/throw thing?
2015-10-08T21:55:54Z stassats: yeah
2015-10-08T21:56:57Z nyef: Could you do with another set of eyes on the assembly-routine?
2015-10-08T21:57:16Z stassats: that never hurts
2015-10-08T21:57:29Z nyef: A github link, then?
2015-10-08T21:58:08Z stassats: https://github.com/sbcl/sbcl/blob/master/src/assembly/arm64/assem-rtns.lisp#L149
2015-10-08T21:58:33Z nyef: Thank you.
2015-10-08T21:58:55Z stassats: https://github.com/sbcl/sbcl/blob/master/src/compiler/arm64/nlx.lisp
2015-10-08T22:00:05Z mprelude joined #sbcl
2015-10-08T22:08:04Z nyef: Your NLX-ENTRY VOP seems iffy.
2015-10-08T22:08:28Z nyef: The general case, the loop for TN-REF-ACROSS for the values.
2015-10-08T22:09:02Z nyef: Where do the flags for your :LT tests get set in the loop?
2015-10-08T22:09:38Z nyef: (Compare the ARM version, for example, with its INST SUBS  before the SC-CASE.)
2015-10-08T22:14:07Z stassats: that might be it
2015-10-08T22:15:58Z nyef: Other than a non-optimiality in NLX-ENTRY-MULTIPLE for when N-FIXNUM-TAG-BITS = WORD-SHIFT, I'm not seeing much else at this point.
2015-10-08T22:20:30Z smokeink_ joined #sbcl
2015-10-08T22:22:10Z smokeink quit (Ping timeout: 240 seconds)
2015-10-08T22:22:17Z stassats: that solved it
2015-10-08T22:36:10Z fridim_ quit (Ping timeout: 240 seconds)
2015-10-08T22:39:01Z nyef: Good. Any more bugs you're trying to fix today?
2015-10-08T22:42:25Z stassats: that's it for today, i think
2015-10-08T22:43:01Z stassats: have a strange gc.impure.lisp failure
2015-10-08T22:43:19Z stassats: "The assertion GC-HAPPEND failed"
2015-10-08T22:44:55Z stassats: and i think i broke unbind-to-here without noticing
2015-10-08T22:44:55Z stassats: oops
2015-10-08T22:47:18Z stassats: and there's only a few test failures left
2015-10-08T22:57:51Z heddwch is now known as ortie
2015-10-08T22:58:59Z ortie is now known as nortie
2015-10-08T23:22:53Z Ettore quit (Quit: Leaving.)
2015-10-08T23:25:07Z mprelude quit (Ping timeout: 246 seconds)
2015-10-08T23:30:24Z nikki93 quit (Ping timeout: 264 seconds)
2015-10-08T23:35:18Z stassats quit (Ping timeout: 260 seconds)
2015-10-08T23:36:42Z nikki93 joined #sbcl
2015-10-08T23:56:25Z stassats joined #sbcl
2015-10-08T23:57:34Z nortie is now known as heddwch
2015-10-09T00:20:06Z stassats quit (Ping timeout: 240 seconds)
2015-10-09T00:37:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-09T00:48:26Z eschatologist joined #sbcl
2015-10-09T00:48:54Z eschatologist quit (Client Quit)
2015-10-09T00:57:34Z nikki93 quit (Remote host closed the connection)
2015-10-09T01:34:42Z tifa joined #sbcl
2015-10-09T01:58:11Z nikki93 joined #sbcl
2015-10-09T02:03:47Z nikki93 quit (Ping timeout: 264 seconds)
2015-10-09T02:14:09Z heddwch quit (Quit: ZNC - 1.6.0 - http://znc.in)
2015-10-09T02:15:35Z heddwch joined #sbcl
2015-10-09T02:16:06Z scymtym_ quit (Ping timeout: 272 seconds)
2015-10-09T02:18:19Z heddwch quit (Client Quit)
2015-10-09T02:39:07Z heddwch joined #sbcl
2015-10-09T02:59:35Z nikki93 joined #sbcl
2015-10-09T03:03:59Z nikki93 quit (Ping timeout: 252 seconds)
2015-10-09T03:07:46Z karswell quit (Read error: Connection reset by peer)
2015-10-09T03:08:00Z karswell joined #sbcl
2015-10-09T03:32:34Z psy_ quit (Ping timeout: 246 seconds)
2015-10-09T05:56:53Z Shinmera joined #sbcl
2015-10-09T06:04:31Z nikki93 joined #sbcl
2015-10-09T06:08:49Z nikki93 quit (Ping timeout: 256 seconds)
2015-10-09T06:18:28Z ASau quit (Ping timeout: 246 seconds)
2015-10-09T06:24:07Z smokeink_ quit (Ping timeout: 256 seconds)
2015-10-09T06:27:43Z smokeink joined #sbcl
2015-10-09T06:32:48Z smokeink quit (Ping timeout: 264 seconds)
2015-10-09T06:33:25Z smokeink joined #sbcl
2015-10-09T07:06:28Z scymtym_ joined #sbcl
2015-10-09T07:24:43Z fridim_ joined #sbcl
2015-10-09T07:32:21Z Cymew joined #sbcl
2015-10-09T08:46:37Z stassats joined #sbcl
2015-10-09T08:50:06Z stassats quit (Remote host closed the connection)
2015-10-09T08:54:01Z Ettore joined #sbcl
2015-10-09T08:56:10Z smokeink quit (Ping timeout: 240 seconds)
2015-10-09T08:57:44Z smokeink joined #sbcl
2015-10-09T09:00:30Z heddwch quit (Read error: Connection reset by peer)
2015-10-09T09:01:19Z heddwch joined #sbcl
2015-10-09T09:12:16Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-09T09:24:46Z sjl quit (Ping timeout: 240 seconds)
2015-10-09T09:25:22Z attila_lendvai joined #sbcl
2015-10-09T10:03:06Z sjl joined #sbcl
2015-10-09T10:05:46Z karswell quit (Read error: Connection reset by peer)
2015-10-09T10:06:01Z karswell joined #sbcl
2015-10-09T10:31:56Z karswell quit (Read error: Connection reset by peer)
2015-10-09T10:31:59Z karswell` joined #sbcl
2015-10-09T10:51:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-09T11:32:01Z _iwc quit (Killed (Sigyn (Spam is off topic on freenode.)))
2015-10-09T12:30:37Z smokeink quit (Remote host closed the connection)
2015-10-09T12:47:14Z DeadTrickster quit (Ping timeout: 260 seconds)
2015-10-09T12:52:11Z DeadTrickster joined #sbcl
2015-10-09T13:03:51Z smokeink joined #sbcl
2015-10-09T13:08:54Z psilord quit (Quit: Leaving.)
2015-10-09T13:26:01Z eudoxia joined #sbcl
2015-10-09T13:58:59Z gingerale joined #sbcl
2015-10-09T14:16:44Z karswell` quit (Remote host closed the connection)
2015-10-09T14:35:50Z psilord joined #sbcl
2015-10-09T14:45:21Z scymtym_: i updated David's core relocation patch and got to the point where the target system survives a forced relocation and appears to work (probably somewhat by accident). would someone with better knowledge of the runtime be willing to have a look?
2015-10-09T14:49:14Z scymtym_ just got "The value -2569666360 is not of type (SIGNED-BYTE 32)." in SB-VM:FIXUP-CODE-OBJECT when loading a fasl into the relocated image
2015-10-09T14:59:47Z Cymew quit (Ping timeout: 256 seconds)
2015-10-09T15:13:20Z nyef: scymtym_: It's been a while since I've tangled with that patch. I'm not sure how much time I can spend on it right now, but I'm willing to take a look.
2015-10-09T15:13:43Z scymtym_: thank you, i will make a paste
2015-10-09T15:14:48Z nyef: You're on x86-64, aren't you?
2015-10-09T15:15:40Z jfe_ is now known as jfe
2015-10-09T15:16:09Z scymtym_: currently x86
2015-10-09T15:16:25Z scymtym_: only tested on this machine
2015-10-09T15:17:21Z joshe: out of curiosity, why are you interested in relocatable cores?
2015-10-09T15:17:34Z scymtym_: "paste too large" :(
2015-10-09T15:17:38Z nyef: Hrm. So, how would you have managed to overflow a (signed-byte 32) in a fixup merely by moving heap spaces around?
2015-10-09T15:18:56Z nyef: Are you using a traditional 2G/2G user/kernel split, or a 3G/1G split?
2015-10-09T15:19:31Z nyef: I think you can get that result by having part of your heap space up over the 2G mark on 32-bit.
2015-10-09T15:19:36Z scymtym_: nyef: http://www.techfak.uni-bielefeld.de/~jmoringe/0001-TODO-ask-David-give-credit-remove-TEST_-START-END-Op.patch
2015-10-09T15:22:30Z nyef: This... is going to take a bit to wade through.
2015-10-09T15:22:40Z scymtym_: i used the TEST_{START,END} hack to force a relocation and ended up with "relocating core to: 0x97d89000; from: 0x9000000"
2015-10-09T15:24:26Z scymtym_: joshe: i had trouble with the current personality()-based workaround on Ubuntu
2015-10-09T15:24:55Z joshe: oh interesting, I didn't know there was a need for it on linux
2015-10-09T15:26:42Z hzp: hmm, I guess that file header was intended to be the Gerd Moellmann-style germanified version of the SBCL license.  But if the file that I copied from didn't have the header, by all means leave it out.
2015-10-09T15:27:43Z scymtym_: iiuc, it is /only/ used on linux and only with specific kernel versions to disable address space layout randomization
2015-10-09T15:29:32Z joshe: openbsd has randomized malloc() and shared library addresses for a while, and now even executable addresses
2015-10-09T15:30:00Z joshe: sbcl runs fine but it would still be nice to have sbcl fit in better
2015-10-09T15:30:07Z jfe quit (Disconnected by services)
2015-10-09T15:30:15Z nyef: joshe: That... has to suck for anything that is really heavy on pointer-based data structures.
2015-10-09T15:30:47Z joshe: why? memory locality?
2015-10-09T15:31:37Z joshe: I don't know how small allocations within a page are handled
2015-10-09T15:32:51Z nyef: No, pointer fixups in static data and code... when it's almost all pointers.
2015-10-09T15:33:26Z scymtym_: nyef: i put in just enough work to make the patch apply, make sbcl compile with it and get to a usable repl on my machine. should i try to prepare/improve anything else before you go through the trouble of looking at the patch?
2015-10-09T15:33:32Z nyef: A simple example: An ITC Forth interpreter packaged as an executable is almost entirely pointers.
2015-10-09T15:33:34Z joshe: oh, right
2015-10-09T15:34:15Z joshe: I don't know what kind of code gcc generates when you ask it to do PIE, but my impression was the overhead wasn't huge
2015-10-09T15:34:59Z nyef: scymtym_: Nothing leaped out at me on the first pass. I see that you even got the FDEFN thing right for ARM.
2015-10-09T15:35:31Z scymtym_: well, cargo culted from the gc and untested, of course
2015-10-09T15:35:46Z scymtym_: s/gc/gc code/
2015-10-09T15:38:33Z scymtym_: also note that this is 99 % not my work, if that wasn't clear from the previous discussion
2015-10-09T15:40:49Z pkhuong: nyef: I think each section's base address is randomised, but not individual objects
2015-10-09T15:40:58Z pkhuong: (unless you do section/object :x)
2015-10-09T15:41:33Z nyef: pkhuong: But that still means fixing up each and every direct pointer within the section.
2015-10-09T15:42:11Z nyef: And, as I said, when the section is MOSTLY pointers...
2015-10-09T15:43:35Z nyef: Essentially, it'd be like trying to persuade the OS loader to apply relocations for static-space on cheneygc!
2015-10-09T16:07:09Z pkhuong: gnu ld does have support for splitting sections with too many relocs...
2015-10-09T16:25:21Z eudoxia quit (Ping timeout: 256 seconds)
2015-10-09T16:39:29Z stassats joined #sbcl
2015-10-09T16:46:30Z stassats: so, is that a problem with the static space / runtime being too far away from dynamic space?
2015-10-09T16:47:09Z stassats: can't imagine there to be 2G of space within a single code object
2015-10-09T16:48:10Z stassats: so, it probably shouldn't use relative jumps/calls for fixups
2015-10-09T16:48:18Z stassats: but it presumably worked before?
2015-10-09T16:48:41Z nyef: stassats: That's the impression I get. A relative fixup to read-only-space for assembly-routines, for example.
2015-10-09T16:55:58Z stassats: maybe it maps all spaces separately, if it did them all at once that wouldn't be a problem
2015-10-09T16:56:02Z stassats: and contiguously
2015-10-09T16:56:07Z stassats: if at all possible on x86
2015-10-09T16:56:51Z nyef: The spaces are rather disjoint, IIRX.
2015-10-09T16:56:54Z nyef: Err...
2015-10-09T16:56:55Z nyef: IIRC.
2015-10-09T17:02:08Z stassats: because they are static
2015-10-09T17:02:18Z stassats: if they are relocatable, might as well splice them
2015-10-09T17:03:22Z stassats: maybe there is a key to MMAP to map into the lower space?
2015-10-09T17:03:56Z stassats: then again, you can increase dynamic-space-size and some code objects might venture too far away
2015-10-09T17:09:40Z nyef: Part of the point of read-only-space is that the GC doesn't even need to look at it.
2015-10-09T17:10:09Z nyef: On the other hand, it's almost entirely CODE-OBJECTs, and without constant vectors, at that.
2015-10-09T17:10:43Z stassats: gc can not look at it where it is
2015-10-09T17:10:46Z stassats: wherever
2015-10-09T17:12:16Z nyef: I think that part of my point is that GC should skip it fairly easily and cheaply wherever it is.
2015-10-09T17:12:42Z nyef: We'd have to start tracking references to it if it's movable, though, of course.
2015-10-09T17:13:52Z stassats: my main desire for relocatable cores is actually relocating the runtime, i.e, it could be loaded as a shared object
2015-10-09T17:14:05Z stassats: which itself is not a problem, linkage table can solve that already
2015-10-09T17:14:18Z stassats: but it may step onto some of the spaces, which would have to get relocated
2015-10-09T17:14:42Z stassats: i wonder if i can cause that sb32 error without that patch
2015-10-09T17:15:11Z nyef: Right, fully-relocatable cores + libsbcl.so could be a decent flexibility win.
2015-10-09T17:18:15Z karswell` joined #sbcl
2015-10-09T17:24:25Z hlavaty quit (Remote host closed the connection)
2015-10-09T17:28:50Z stassats: ha, gotcha
2015-10-09T17:28:54Z stassats: The value -2177666960 is not of type (SIGNED-BYTE 32).
2015-10-09T17:29:45Z stassats: (progn (defvar *z* (make-array 498435455 :initial-element -1)) (sb-sys:without-gcing (loop do (compile nil `(lambda () (throw 'x 10))))))
2015-10-09T17:31:21Z stassats: with 3300MB of dynamic-space-size
2015-10-09T17:33:27Z psy_ joined #sbcl
2015-10-09T17:34:05Z psy_ quit (Max SendQ exceeded)
2015-10-09T17:34:31Z psy_ joined #sbcl
2015-10-09T17:48:07Z stassats: filed as https://bugs.launchpad.net/sbcl/+bug/1504629
2015-10-09T17:49:54Z karswell` quit (Read error: Connection reset by peer)
2015-10-09T17:50:07Z karswell` joined #sbcl
2015-10-09T18:03:29Z karswell` quit (Read error: Connection reset by peer)
2015-10-09T18:03:42Z karswell` joined #sbcl
2015-10-09T18:06:26Z nyef: 3300MB of dynamic-space-size is basically three gigs, isn't it?
2015-10-09T18:08:27Z stassats: a tad more
2015-10-09T18:10:07Z nyef: Okay, so this was never expected to work, AIUI.
2015-10-09T18:10:29Z stassats: it wasn't thought about
2015-10-09T18:11:13Z nyef: I think that it was last thought about when the user/kernel split was 2G/2G.
2015-10-09T18:12:44Z nyef: Crazy question, but is the target address masked down to an unsigned-byte 32, or does the CPU do some sort of overflow/underflow checking/
2015-10-09T18:12:46Z nyef: ?
2015-10-09T18:14:15Z psy_ quit (Read error: Connection reset by peer)
2015-10-09T18:14:24Z stassats: well, it's a signed 32-bit offset
2015-10-09T18:14:55Z nyef: Applied to an unsigned 32-bit index into an unsigned 32-bit address space.
2015-10-09T18:15:02Z stassats: you mean what happens when PC+offset overflow?
2015-10-09T18:15:11Z nyef: Or underflow, yes.
2015-10-09T18:15:35Z stassats: just a GP fault?
2015-10-09T18:15:50Z nyef: Does it? Or does it just wrap?
2015-10-09T18:19:56Z stassats: tempEIP ← EIP + DEST; (* DEST is rel32 *) IF tempEIP is not within code segment limit THEN #GP(0); FI;
2015-10-09T18:19:56Z stassats:  
2015-10-09T18:19:57Z stassats:  
2015-10-09T18:20:03Z stassats: that's what happens
2015-10-09T18:20:21Z attila_lendvai joined #sbcl
2015-10-09T18:20:21Z attila_lendvai quit (Changing host)
2015-10-09T18:20:21Z attila_lendvai joined #sbcl
2015-10-09T18:20:33Z pkhuong: nyef: pretty sure it does a modular add.
2015-10-09T18:20:46Z pkhuong: not that useful on x86-64 though
2015-10-09T18:20:52Z sjl quit (Ping timeout: 246 seconds)
2015-10-09T18:21:06Z psy_ joined #sbcl
2015-10-09T18:21:20Z stassats: so, now i need to find the defintion of intel's pseudocode
2015-10-09T18:21:28Z psy_ quit (Max SendQ exceeded)
2015-10-09T18:24:09Z nyef: pkhuong: But this is a 32-bit x86 issue, not an x86-64 issue. Or are you saying that the compatibility mode isn't quite that compatible?
2015-10-09T18:24:34Z pkhuong: I'm just saying that we'll have the same problem, but worse, on x86-64
2015-10-09T18:24:55Z pkhuong: oh wait no. we use a different strategy
2015-10-09T18:24:57Z pkhuong: never mind
2015-10-09T18:24:58Z nyef: I have a vague impression that, at least at some point, we had a requirement that the high bit on any object pointer be zero.
2015-10-09T18:25:09Z nyef: Something to do with EQ hashing or similar.
2015-10-09T18:25:28Z pkhuong: that's unlikely to be broken on x86-64, at least until sb-unikernel
2015-10-09T18:33:25Z mattprelude joined #sbcl
2015-10-09T18:36:12Z stassats: ok, i just need to try it
2015-10-09T18:37:46Z stassats: there's no problem on x86-64 because it doesn't even implement relative CALLs
2015-10-09T18:37:48Z stassats: only relative jumps
2015-10-09T18:38:41Z stassats: or not so fast
2015-10-09T18:45:31Z stassats: ok, we never JMP relatively to assembly routines
2015-10-09T18:53:57Z gingerale- joined #sbcl
2015-10-09T18:54:14Z gingerale quit (Disconnected by services)
2015-10-09T18:54:43Z gingerale- is now known as gingerale
2015-10-09T19:09:50Z leuler joined #sbcl
2015-10-09T19:11:02Z sjl joined #sbcl
2015-10-09T19:14:04Z nyef: Right, x86-64 never emits a relative branch to outside of a code-object in order to make relocation easier. None of that code-fixup-envectorization that we have on x86.
2015-10-09T19:14:22Z stassats: ok, let's make a 2GB code object
2015-10-09T19:15:14Z nyef: (Note that setting up the fixup-vector in Genesis was one of the two major fixes to enable the relocation branch to work on cold-cores.)
2015-10-09T19:19:08Z stassats: managed The value 2147483706 is not of type (SIGNED-BYTE 32). on x86-64
2015-10-09T19:19:18Z stassats: but this time during SB-ASSEM::PROCESS-BACK-PATCHES
2015-10-09T19:20:09Z stassats: though, 2 gigabyte code objects are probably not a problem
2015-10-09T19:20:19Z stassats: you would never be able to compile such a thing
2015-10-09T19:33:19Z nikki93 joined #sbcl
2015-10-09T19:39:51Z nikki93 quit (Remote host closed the connection)
2015-10-09T19:41:11Z smokeink quit (Ping timeout: 250 seconds)
2015-10-09T20:54:40Z prxq joined #sbcl
2015-10-09T21:13:27Z leuler quit (Quit: ERC Version 5.1.2 $Revision: 1.796.2.6 $ (IRC client for Emacs))
2015-10-09T21:17:28Z ASau joined #sbcl
2015-10-09T21:18:58Z flavioc joined #sbcl
2015-10-09T21:30:20Z sjl quit (Read error: Connection reset by peer)
2015-10-09T21:31:03Z sjl joined #sbcl
2015-10-09T21:51:46Z flavioc quit (Ping timeout: 240 seconds)
2015-10-09T22:02:12Z stassats: looks like i've fixed relative fixups by using modular addition
2015-10-09T22:04:03Z stassats: my modular math is not strong, maybe there is a better formula than http://paste.lisp.org/display/156532
2015-10-09T22:05:25Z stassats: hold on, found a better one
2015-10-09T22:06:50Z stassats: http://paste.lisp.org/display/156532#1 i think that's what i was looking for
2015-10-09T22:07:40Z gingerale quit (Remote host closed the connection)
2015-10-09T22:08:45Z stassats: only tested with backward jumps, need to make sure forwards works too
2015-10-09T22:10:13Z stassats: seems right as well
2015-10-09T22:10:54Z stassats: ok, the math is right, but i get a 33-bit value
2015-10-09T22:11:49Z stassats: so, an additional (ldb (byte 32 0)) around it
2015-10-09T22:16:13Z karswell` quit (Remote host closed the connection)
2015-10-09T22:16:28Z karswell` joined #sbcl
2015-10-09T22:18:25Z psilord quit (Quit: Leaving.)
2015-10-09T22:19:31Z stassats: or just one around everything
2015-10-09T22:26:19Z stassats: and now i came back to (ldb (byte 32 0) rel-val)
2015-10-09T22:32:10Z Bike quit (Quit: leaving)
2015-10-09T22:41:42Z stassats: scymtym_: you should try your patch now
2015-10-09T22:45:04Z Bike joined #sbcl
2015-10-09T22:50:07Z prxq quit (Remote host closed the connection)
2015-10-09T23:12:41Z Ettore quit (Quit: Leaving.)
2015-10-09T23:15:49Z sjl quit (Ping timeout: 252 seconds)
2015-10-09T23:17:19Z fridim_ quit (Ping timeout: 240 seconds)
2015-10-09T23:22:49Z sjl joined #sbcl
2015-10-09T23:38:27Z psilord joined #sbcl
2015-10-10T00:01:11Z Bicyclidine joined #sbcl
2015-10-10T00:04:41Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-10T00:11:19Z mattprelude quit (Ping timeout: 240 seconds)
2015-10-10T00:13:01Z stassats quit (Ping timeout: 252 seconds)
2015-10-10T00:18:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-10T00:36:48Z Bicyclidine quit (Ping timeout: 244 seconds)
2015-10-10T00:39:07Z Bicyclidine joined #sbcl
2015-10-10T00:53:42Z rszeno joined #sbcl
2015-10-10T01:03:21Z Quadrescence joined #sbcl
2015-10-10T01:53:28Z nikki93 joined #sbcl
2015-10-10T02:10:54Z smokeink joined #sbcl
2015-10-10T02:29:14Z rszeno quit (Quit: Leaving.)
2015-10-10T02:38:46Z smokeink quit (Ping timeout: 240 seconds)
2015-10-10T02:40:07Z smokeink joined #sbcl
2015-10-10T02:44:10Z Bicyclidine quit (Ping timeout: 272 seconds)
2015-10-10T02:50:26Z Bicyclidine joined #sbcl
2015-10-10T03:08:46Z Bicyclidine quit (Ping timeout: 240 seconds)
2015-10-10T03:10:52Z Bicyclidine joined #sbcl
2015-10-10T03:19:02Z nikki93 quit (Remote host closed the connection)
2015-10-10T03:32:32Z Bicyclidine quit (Ping timeout: 250 seconds)
2015-10-10T03:49:26Z Bicyclidine joined #sbcl
2015-10-10T03:53:58Z Bicyclidine quit (Ping timeout: 260 seconds)
2015-10-10T03:55:41Z Bicyclidine joined #sbcl
2015-10-10T04:14:15Z Bicyclidine quit (Ping timeout: 250 seconds)
2015-10-10T04:16:09Z Bicyclidine joined #sbcl
2015-10-10T04:43:52Z Bicyclidine quit (Ping timeout: 246 seconds)
2015-10-10T05:50:16Z momo-reina joined #sbcl
2015-10-10T05:50:24Z momo-reina quit (Remote host closed the connection)
2015-10-10T05:58:59Z momo-reina joined #sbcl
2015-10-10T06:03:05Z psy_ joined #sbcl
2015-10-10T06:15:49Z momo-reina quit (Remote host closed the connection)
2015-10-10T06:16:12Z momo-reina joined #sbcl
2015-10-10T06:45:42Z fridim_ joined #sbcl
2015-10-10T06:47:44Z jlarocco_ joined #sbcl
2015-10-10T07:16:24Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-10T07:25:53Z Shinmera joined #sbcl
2015-10-10T07:34:56Z gingerale joined #sbcl
2015-10-10T07:35:41Z jlarocco_ quit (Quit: Leaving)
2015-10-10T07:49:30Z momo-reina quit (Ping timeout: 250 seconds)
2015-10-10T09:02:39Z stassats joined #sbcl
2015-10-10T09:08:21Z Bike quit (Remote host closed the connection)
2015-10-10T09:10:18Z Bike joined #sbcl
2015-10-10T09:11:26Z stassats quit (Remote host closed the connection)
2015-10-10T09:20:23Z stassats joined #sbcl
2015-10-10T09:32:47Z Ettore joined #sbcl
2015-10-10T09:50:06Z Xof: as ever when I try to do something... current status: baffled
2015-10-10T09:50:38Z Xof: (and also trying to debug things that are basically unconnected to the thing I am trying to do...)
2015-10-10T09:51:33Z stassats: the confused thing?
2015-10-10T09:51:46Z stassats: did it ever work or did i break it?
2015-10-10T09:51:54Z Xof: it's not your fault
2015-10-10T09:52:16Z Xof: it's a combination of several things, at most one I understand :)
2015-10-10T09:53:38Z Xof: the sb-unicode:confusable-p bug that I've just reported, I only found because I'm trying to build with readtable-case :invert
2015-10-10T09:53:49Z Xof: and that caused an explosion in cold-init
2015-10-10T09:54:02Z Xof: (because of NIL entries)
2015-10-10T09:54:14Z Xof: but that explosion should still be there even without readtable-case :invert, but it isn't
2015-10-10T09:54:44Z Xof: however, I now have a core, so maybe our extensive test suite will help me find the bits of broken behaviour
2015-10-10T09:55:33Z stassats: maybe it normally disables unicode tricks during xc/cold-init?
2015-10-10T09:56:07Z Xof: no, this is from a bad READ or a bad MAP or a bad COERCE or something
2015-10-10T09:56:14Z Xof: "or something" is the problem
2015-10-10T09:57:20Z stassats: i have an awful (setf (sb-impl::%readtable-normalization *readtable*) nil (sb-impl::%readtable-normalization sb-impl::*standard-readtable*) nil) in my sbcl config
2015-10-10T09:57:39Z stassats: because readtable normalization is a pretty silly thing
2015-10-10T09:59:21Z Xof: I found myself being pretty entertained by it
2015-10-10T09:59:49Z Xof: it did baffle me for a little while before I remembered that it existed
2015-10-10T10:00:56Z Xof: gah, target-unicode.lisp is full of random case
2015-10-10T10:01:31Z Xof: :east-asion
2015-10-10T10:06:38Z attila_lendvai joined #sbcl
2015-10-10T10:06:38Z attila_lendvai quit (Changing host)
2015-10-10T10:06:38Z attila_lendvai joined #sbcl
2015-10-10T10:14:03Z Xof: oh!  A bit less baffled
2015-10-10T10:27:09Z scymtym_: stassats: loading the fasl into the relocated core seems to work with your fix in place
2015-10-10T10:49:18Z White_Flame quit (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.)
2015-10-10T10:49:19Z igajsin_m joined #sbcl
2015-10-10T10:49:37Z White_Flame joined #sbcl
2015-10-10T11:15:26Z stassats: ugh, i think the problem with STP/LDP and binding stack exhaustion is that one of the operations is causing a stack guard protection
2015-10-10T11:15:34Z stassats: and there is a data abort and something happens
2015-10-10T11:18:44Z stassats: or something else happens
2015-10-10T11:19:16Z flavioc joined #sbcl
2015-10-10T11:26:38Z igajsin_m quit (Ping timeout: 260 seconds)
2015-10-10T11:32:12Z sjl quit (Ping timeout: 255 seconds)
2015-10-10T12:17:19Z flavioc quit (Ping timeout: 240 seconds)
2015-10-10T12:43:07Z DeadTrickster quit (Read error: Connection reset by peer)
2015-10-10T12:43:45Z DeadTrickster joined #sbcl
2015-10-10T12:54:23Z nyef: stassats: What? How? The page size is a multiple of the binding-stack-element size... and since when did we have a guard page on the bindstack in the first place?
2015-10-10T12:54:57Z stassats: no, the cause is swiftly reprotects the return page and then hits the guard again
2015-10-10T13:18:51Z stassats: i'm using post-index, need to research what that does on a fault, though, unbind-to-here shouldn't be the one to trigger it
2015-10-10T13:25:58Z nicdev_` is now known as nicdev
2015-10-10T13:38:42Z daimrod quit (Ping timeout: 268 seconds)
2015-10-10T13:40:32Z nyef: ... Are you blowing past the bottom of the bind-stack?
2015-10-10T13:41:05Z stassats: it dips below the guard page, thereby reprotecting it, and then goes up again
2015-10-10T13:44:08Z stassats: i can actually imagine a scenario where that can happen normally
2015-10-10T13:44:43Z DeadTrickster: sorry for offtopic but where can I find Christophe Rhodes, what his nick here?
2015-10-10T13:45:26Z stassats: DeadTrickster: Xof
2015-10-10T13:45:32Z DeadTrickster: thx
2015-10-10T13:46:13Z stassats: dougk_: that last change, does that mean no longer ! packages in cold-init?
2015-10-10T13:48:51Z psy_ quit (Ping timeout: 250 seconds)
2015-10-10T13:50:45Z psy_ joined #sbcl
2015-10-10T14:16:18Z stassats quit (Remote host closed the connection)
2015-10-10T14:20:19Z igajsin1 quit (Quit: Leaving.)
2015-10-10T14:31:23Z psy_ quit (Ping timeout: 264 seconds)
2015-10-10T14:35:30Z stassats joined #sbcl
2015-10-10T14:42:16Z karswell` is now known as karswell
2015-10-10T15:06:10Z flavioc joined #sbcl
2015-10-10T15:10:05Z stassats: looks like i need pre-index, not post-index, doh
2015-10-10T15:11:36Z nyef: pre-index one way, post-index the other?
2015-10-10T15:12:11Z stassats: there's only one way
2015-10-10T15:13:59Z stassats: i also need to optimize assem.S things to use post-index/pre-index only once, not for each load
2015-10-10T15:16:41Z stassats: but i have to benchmark it first
2015-10-10T15:18:28Z flavioc quit (Ping timeout: 272 seconds)
2015-10-10T15:18:45Z stassats: shaved off some 24KB by using a pre-index in unbind
2015-10-10T15:19:50Z flavioc joined #sbcl
2015-10-10T15:20:27Z stassats: but first to fix the remaining test failures
2015-10-10T15:40:00Z flavioc quit (Ping timeout: 264 seconds)
2015-10-10T15:54:50Z psy_ joined #sbcl
2015-10-10T16:00:01Z stassats: only two failures left
2015-10-10T16:02:42Z smokeink quit (Remote host closed the connection)
2015-10-10T16:11:37Z stassats: a test in float.impure.lisp so convoluted, i can't actually figure out which part of it fails
2015-10-10T17:28:58Z fikusz quit (Remote host closed the connection)
2015-10-10T17:30:58Z fikusz joined #sbcl
2015-10-10T17:38:13Z nikki93 joined #sbcl
2015-10-10T17:43:23Z nikki93 quit (Remote host closed the connection)
2015-10-10T17:48:03Z nikki93 joined #sbcl
2015-10-10T18:18:30Z ym: Why (let ((rpart 1/3)) #C(0 rpart)) causes error?
2015-10-10T18:18:46Z stassats: that's not how #C works
2015-10-10T18:18:54Z stassats: ask in #lisp
2015-10-10T18:19:01Z ym: Ok, thanks.
2015-10-10T18:38:58Z karswell quit (Read error: Connection reset by peer)
2015-10-10T18:39:13Z karswell joined #sbcl
2015-10-10T18:39:45Z nikki93 quit (Remote host closed the connection)
2015-10-10T18:56:04Z francogrex joined #sbcl
2015-10-10T18:56:28Z francogrex: hi how come when I use time sbcl --script test.lisp it is faster than sbcl --script test.fasl ? test.lisp contains a recursive fibonacci
2015-10-10T18:56:50Z fridim_ quit (Ping timeout: 240 seconds)
2015-10-10T18:57:04Z stassats: how come you don't know how to benchmark?
2015-10-10T18:57:54Z francogrex: ?
2015-10-10T18:58:07Z stassats: or to ask questions
2015-10-10T18:59:35Z francogrex: ?
2015-10-10T19:06:08Z fridim_ joined #sbcl
2015-10-10T19:17:40Z francogrex: stassats: http://paste.lisp.org/display/156566
2015-10-10T19:19:16Z stassats: the compiled version is faster
2015-10-10T19:19:39Z stassats: you are loading another version
2015-10-10T19:21:33Z francogrex: what other version?
2015-10-10T19:21:44Z stassats: you tell me
2015-10-10T19:22:15Z francogrex: it's the only one compiled from fib.lisp
2015-10-10T19:23:39Z stassats: a) that can't really happen b) it actually doesn't happen
2015-10-10T19:23:41Z francogrex: same if I do (time (load "fib.fasl")) & (time (load "fib.lisp")) few secs
2015-10-10T19:24:06Z francogrex: stassats: on your computer and with your version of sbcl maybe
2015-10-10T19:24:23Z stassats: with any version of sbcl on any computer
2015-10-10T19:24:37Z francogrex: on mine it's always a few secs longer the fasl
2015-10-10T19:24:50Z francogrex: get the fuck outta here
2015-10-10T19:25:08Z stassats: few seconds? the paste shows it's 200ms
2015-10-10T19:26:02Z francogrex: milliseconds
2015-10-10T19:26:49Z stassats: it's just not possible for the compiled version to be slower
2015-10-10T19:26:56Z stassats: so, you're measuring is bogus
2015-10-10T19:26:59Z stassats: your
2015-10-10T19:27:49Z francogrex: it maybe but is it not also possible that the loading of the fasl file takes longer than a lisp... at least initially
2015-10-10T19:29:00Z stassats: no
2015-10-10T19:29:44Z stassats: your machine may be so flaky it's not suitable for measurements
2015-10-10T19:30:17Z francogrex: but see this pls: http://paste.lisp.org/display/156566#1
2015-10-10T19:31:38Z stassats: obviously when you are measuring something you shouldn't have anything else running on the machine and cpu scaling disabled
2015-10-10T19:32:00Z stassats: otherwise you'll just get the price of potatoes
2015-10-10T19:32:15Z stassats: and not any value numbers
2015-10-10T19:32:17Z stassats: valid
2015-10-10T19:32:27Z cades joined #sbcl
2015-10-10T19:52:20Z clique joined #sbcl
2015-10-10T19:58:48Z francogrex left #sbcl
2015-10-10T20:03:33Z clique quit (Quit: Page closed)
2015-10-10T20:15:48Z stassats: one more failure fixed, and the gc.impure thing is left
2015-10-10T20:42:58Z nyef_ joined #sbcl
2015-10-10T20:43:38Z gucore joined #sbcl
2015-10-10T20:49:42Z stassats: for some reason after-gc hooks aren't run
2015-10-10T20:52:33Z nyef quit (Disconnected by services)
2015-10-10T20:52:38Z nyef_ is now known as nyef
2015-10-10T20:53:13Z stassats: moreover, gc isn't being run after receive-pending-interrupt when *gc-pending* is T
2015-10-10T20:53:18Z stassats: but *gc-pending* becomes NIL
2015-10-10T20:54:39Z fridim_ quit (*.net *.split)
2015-10-10T20:54:39Z psy_ quit (*.net *.split)
2015-10-10T20:54:43Z stassats: GC, the function
2015-10-10T20:55:28Z nikki93 joined #sbcl
2015-10-10T20:56:34Z stassats: DYNDEBUG crashes the gc, bummer
2015-10-10T21:00:13Z psy_ joined #sbcl
2015-10-10T21:01:51Z stassats: i'm not returning the lisp value from call-into-lisp
2015-10-10T21:04:52Z nikki93 quit (Remote host closed the connection)
2015-10-10T21:05:25Z stassats: no more test failures
2015-10-10T21:05:49Z stassats: unexpected ones
2015-10-10T21:06:13Z psy_ quit (Max SendQ exceeded)
2015-10-10T21:06:43Z psy_ joined #sbcl
2015-10-10T21:27:52Z nikki93 joined #sbcl
2015-10-10T21:31:00Z mood quit (Ping timeout: 264 seconds)
2015-10-10T21:31:55Z antoszka quit (Ping timeout: 260 seconds)
2015-10-10T21:32:53Z nikki93 quit (Remote host closed the connection)
2015-10-10T21:33:09Z mood_ joined #sbcl
2015-10-10T21:33:17Z antoszka joined #sbcl
2015-10-10T21:33:37Z mood_ is now known as mood
2015-10-10T21:34:00Z minion quit (Disconnected by services)
2015-10-10T21:34:04Z minion joined #sbcl
2015-10-10T21:34:47Z jdz_ joined #sbcl
2015-10-10T21:38:00Z Ober_ joined #sbcl
2015-10-10T21:40:16Z stassats: using one pre-index/post-index doesn't seem to make any difference, but gcc does that
2015-10-10T21:40:59Z stassats: looking at the optimization manual, the base register update takes 1 cycle, while LDR/STR take 4 cycles, and that's done in parallel
2015-10-10T21:41:15Z stassats: though i suppose that's more energy efficient, since it utilizes only one pipeline
2015-10-10T21:43:05Z jdz quit (*.net *.split)
2015-10-10T21:43:08Z rj-code quit (*.net *.split)
2015-10-10T21:43:10Z Ober quit (*.net *.split)
2015-10-10T21:53:13Z karswell quit (Read error: Connection reset by peer)
2015-10-10T21:53:27Z karswell joined #sbcl
2015-10-10T21:53:53Z rj-code joined #sbcl
2015-10-10T21:54:04Z rj-code quit (Max SendQ exceeded)
2015-10-10T21:54:15Z rj-code joined #sbcl
2015-10-10T22:26:17Z nikki93 joined #sbcl
2015-10-10T22:30:41Z nikki93 quit (Ping timeout: 244 seconds)
2015-10-10T22:37:01Z stassats: the new post-index/pre-index do not work well with a stack growing upwards
2015-10-10T22:45:39Z stassats: testing tomorrow some quicklisp stuff, adding missing disassemblies and it'll be a complete port
2015-10-10T22:51:52Z Quadrescence joined #sbcl
2015-10-10T22:52:51Z Ettore quit (Quit: Leaving.)
2015-10-10T23:00:11Z gucore quit (Ping timeout: 264 seconds)
2015-10-10T23:13:29Z gingerale quit (Remote host closed the connection)
2015-10-10T23:21:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-10T23:23:01Z stassats: it can run hunchentoot
2015-10-10T23:50:56Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-11T00:00:24Z stassats quit (Ping timeout: 246 seconds)
2015-10-11T01:00:02Z smokeink joined #sbcl
2015-10-11T01:27:42Z karswell quit (Ping timeout: 250 seconds)
2015-10-11T01:32:48Z scymtym_ quit (Ping timeout: 246 seconds)
2015-10-11T03:31:50Z Quadrescence: What might it take to allow dynamic heap growing?
2015-10-11T03:32:37Z nyef: Starting small, and having a good story about how the gencgc page tables work when they're not linearly allocated.
2015-10-11T04:15:05Z smokeink quit (Quit: Angelic v4.4 - http://angelic.flexnet.org)
2015-10-11T04:47:10Z Quadrescence: Is there a way within SBCL to determine OS X sub-version? I want to only enable a feature if it's OS 10.9 or higher.
2015-10-11T04:48:21Z nyef: run-program "uname -a"?
2015-10-11T04:50:37Z Quadrescence: That says the Darwin kernel version. I wonder if they coincide with OS X version.
2015-10-11T05:11:42Z Quadrescence: Can anyone comment on this WIP patch for SBCL? Some questions: Where/how should I export the function? How can I get SBCL to understand and infer the types better?
2015-10-11T05:11:44Z Quadrescence: http://paste.lisp.org/display/156580
2015-10-11T05:34:16Z Quadrescence: Wrote to the mailing list.
2015-10-11T06:34:18Z psy_ quit (Ping timeout: 255 seconds)
2015-10-11T07:02:09Z psy joined #sbcl
2015-10-11T07:15:11Z gucore joined #sbcl
2015-10-11T07:15:53Z psy quit (Remote host closed the connection)
2015-10-11T07:16:56Z psy joined #sbcl
2015-10-11T07:25:04Z Shinmera joined #sbcl
2015-10-11T07:52:05Z Xof: nyef: there was a lichteblau patch for that, wasn't there?
2015-10-11T07:53:42Z gingerale joined #sbcl
2015-10-11T07:56:33Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-11T08:02:26Z ASau` joined #sbcl
2015-10-11T08:06:06Z ASau quit (Ping timeout: 260 seconds)
2015-10-11T08:12:24Z gucore quit (Ping timeout: 255 seconds)
2015-10-11T08:35:21Z scymtym_ joined #sbcl
2015-10-11T10:31:16Z stassats joined #sbcl
2015-10-11T10:53:41Z attila_lendvai joined #sbcl
2015-10-11T11:46:00Z dougk_ quit (Ping timeout: 246 seconds)
2015-10-11T11:51:48Z ASau` is now known as ASau
2015-10-11T11:58:58Z dougk_ joined #sbcl
2015-10-11T12:11:12Z stassats quit (Ping timeout: 272 seconds)
2015-10-11T12:36:11Z Ettore joined #sbcl
2015-10-11T12:56:21Z psy quit (Ping timeout: 246 seconds)
2015-10-11T12:57:12Z psy joined #sbcl
2015-10-11T13:58:18Z heddwch quit (Ping timeout: 265 seconds)
2015-10-11T14:31:13Z stassats joined #sbcl
2015-10-11T15:41:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-11T16:37:02Z heddwch joined #sbcl
2015-10-11T16:38:53Z heddwch quit (Read error: Connection reset by peer)
2015-10-11T16:43:12Z heddwch joined #sbcl
2015-10-11T16:55:58Z evdev-ebev joined #sbcl
2015-10-11T17:14:19Z stassats: can run commonqt on arm64
2015-10-11T17:30:21Z flip214: all the qt packages are available on arm64, too; and amd64 runs commonqt, so why should the combination arm64 be prohibited?
2015-10-11T17:30:49Z stassats: huh?
2015-10-11T17:41:57Z gucore joined #sbcl
2015-10-11T17:46:57Z Xof: flip214: stassats is pleased that his arm64 port of sbcl is functional
2015-10-11T17:48:21Z Shinmera: stassats: does that mean I have to add support for arm64 to qt-libs
2015-10-11T17:48:23Z nyef: For what it's worth, I'm impressed at how quickly the arm64 port came together.
2015-10-11T17:48:57Z foom quit (Ping timeout: 246 seconds)
2015-10-11T17:48:58Z stassats: Shinmera: sure, does it support arm32 already?
2015-10-11T17:49:08Z Shinmera: Nope.
2015-10-11T17:49:16Z Shinmera: Only x86_64 (and x86 on windows)
2015-10-11T17:49:50Z Shinmera: Well, technically anything that can compile smoke on linux might work already, I just haven't tested it.
2015-10-11T17:50:17Z stassats: ok, now i need to figure out how to make disassemble to produce aliases when complicated relations are involved, not just Rn == XZR
2015-10-11T17:50:27Z stassats: but Rn == Rm, is there an easy way?
2015-10-11T17:52:56Z stassats: maybe something to do with prefilter?
2015-10-11T17:53:48Z stassats: i guess i can just use a dedicated printer, but that's not nice
2015-10-11T17:54:30Z stassats: don't really want to extend the disassemble either, it gives me enough headache as it is
2015-10-11T17:57:49Z stassats: looks like i can use :control
2015-10-11T18:00:27Z stassats: or not
2015-10-11T18:02:18Z foom joined #sbcl
2015-10-11T18:03:17Z stassats: extending it is, sigh
2015-10-11T18:09:09Z flip214: oh, sorry. misread.
2015-10-11T18:09:13Z flip214: stassats: congratulation!
2015-10-11T18:45:17Z Quadrescence joined #sbcl
2015-10-11T20:07:40Z stassats: looks like i disassemble everything, but complex aliases are still not aliased
2015-10-11T20:08:08Z stassats: i'll need some kind of :test key to the printer
2015-10-11T20:09:17Z stassats: the disassembler really does suck, and assembler too, i used a macro for easier emit-.. definition
2015-10-11T20:09:26Z stassats: the whole thing needs to be rewritten
2015-10-11T20:09:58Z stassats: too bad it's likely to get forgotten until the next port
2015-10-11T20:32:21Z nikki93 joined #sbcl
2015-10-11T20:33:35Z nyef: So... If we're lucky, it'll be remembered before the end of the year?
2015-10-11T20:34:36Z psy quit (Remote host closed the connection)
2015-10-11T20:34:36Z stassats: the instructions we already have are good for ppc, it only has some additional ones
2015-10-11T20:35:15Z stassats: it'll be hard to modify the [dis]assembler without being able to recompile, maybe make the new one switchable? that sounds pretty horrible
2015-10-11T20:36:27Z attila_lendvai joined #sbcl
2015-10-11T20:37:49Z nikki93_ joined #sbcl
2015-10-11T20:38:12Z nikki93 quit (Ping timeout: 250 seconds)
2015-10-11T20:38:24Z stassats: that's what i used https://github.com/sbcl/sbcl/blob/master/src/compiler/arm64/insts.lisp#L680
2015-10-11T20:38:29Z stassats: compare it to https://github.com/sbcl/sbcl/blob/master/src/compiler/arm/insts.lisp#L1024
2015-10-11T20:38:55Z stassats: it gives arg-list display, don't need to copy shared values, etc.
2015-10-11T20:39:26Z stassats: now, define-instruction-format has a lot of boilerplate, i think emitting and printer definitions could be combined into a single macro
2015-10-11T20:42:47Z nyef: That gets a touch iffy, I think.
2015-10-11T20:44:28Z stassats: say it uses the same def-emitter definition, and then something similar to define-instruction-format, but it can refer to fields by name, not through BYTE
2015-10-11T20:44:37Z stassats: and constant fields are extracted automatically
2015-10-11T20:46:39Z nikki93_ quit (Remote host closed the connection)
2015-10-11T20:47:34Z stassats: i should have done that before i started on disassembling, but i was in a hurry
2015-10-11T21:04:39Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-11T21:06:11Z nikki93 joined #sbcl
2015-10-11T21:06:39Z nikki93 quit (Remote host closed the connection)
2015-10-11T21:11:21Z nikki93 joined #sbcl
2015-10-11T21:28:05Z gingerale quit (Remote host closed the connection)
2015-10-11T21:29:47Z stassats: i guess it's time to take a break from ARM stuff, can't think of anything else to be done, besides threads
2015-10-11T21:30:01Z stassats: and threads are for the next month
2015-10-11T21:40:39Z gucore quit (Ping timeout: 252 seconds)
2015-10-11T21:54:33Z nikki93 quit (Ping timeout: 255 seconds)
2015-10-11T21:56:40Z nikki93 joined #sbcl
2015-10-11T23:09:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-11T23:45:36Z Ettore quit (Quit: Leaving.)
2015-10-12T00:08:22Z mood: There appears to be a typo in sb-sprof, it's been there since 2013, and I don't understand why this hasn't caused any trouble before.
2015-10-12T00:09:27Z mood: On line 735 of sb-sprof.lisp, in the macro definition of with-profiling, there's an extra comma that causes evaluation of *samples* at compile time
2015-10-12T00:10:07Z mood: Initially, however, *samples* is nil, and the function call fails because it's not an instance of SAMPLES
2015-10-12T00:11:30Z stassats: i don't see it
2015-10-12T00:12:16Z mood: There's (let ((,last-index, (samples-index...
2015-10-12T00:12:24Z mood: The comma after last-index shouldn't be there
2015-10-12T00:12:58Z stassats: ok, that's a good way to hide
2015-10-12T00:14:14Z stassats: it only happens with :loop t
2015-10-12T00:16:25Z mood: Yes, is that really so uncommon to use? It was the first thing I tried
2015-10-12T00:16:37Z stassats: i never even use with-profiling
2015-10-12T00:28:26Z stassats quit (Ping timeout: 260 seconds)
2015-10-12T00:31:29Z nyef: ... Is there a :RACIAL option to WITH-PROFILING ?
2015-10-12T00:38:13Z nikki93 quit (Remote host closed the connection)
2015-10-12T00:43:37Z scymtym_ quit (Ping timeout: 252 seconds)
2015-10-12T00:52:13Z Quadrescence: nyef, ha
2015-10-12T01:39:24Z nikki93 joined #sbcl
2015-10-12T01:43:38Z nikki93 quit (Remote host closed the connection)
2015-10-12T05:57:27Z nikki93 joined #sbcl
2015-10-12T06:17:39Z ASau quit (Ping timeout: 268 seconds)
2015-10-12T06:31:46Z Shinmera joined #sbcl
2015-10-12T06:44:21Z jlarocco_ joined #sbcl
2015-10-12T07:29:59Z Cymew joined #sbcl
2015-10-12T07:33:32Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-12T07:45:55Z nikki93 quit (Remote host closed the connection)
2015-10-12T07:57:18Z nikki93 joined #sbcl
2015-10-12T07:57:18Z nikki93 quit (Remote host closed the connection)
2015-10-12T08:03:14Z PuercoPop quit (K-Lined)
2015-10-12T08:05:32Z PuercoPop joined #sbcl
2015-10-12T09:00:33Z heddwch quit (Read error: Connection reset by peer)
2015-10-12T09:01:24Z heddwch joined #sbcl
2015-10-12T09:15:44Z loke: Why does the following return a second value of NIL? My reading of the spec says it should return true:
2015-10-12T09:15:45Z loke: (compile 'foo (lambda (x y) (1+ x)))
2015-10-12T09:15:57Z loke: "The secondary value, warnings-p, is false if no conditions of type error or warning were detected by the compiler, and true otherwise."
2015-10-12T09:16:10Z flip214: loke: y wasn't used?
2015-10-12T09:16:16Z flip214: so you'd get a warning?
2015-10-12T09:16:26Z loke: Correct. That raises a warning, so I'd expect warnings-p to be true.
2015-10-12T09:17:57Z Xof: in your expression, what is the type of the second argument that compile receives?
2015-10-12T09:18:02Z loke: The third return value (error-p) has a special exception for style-warning, so it might be that SBCL's implementation of COMPILE accidentally checks it for warnings-p as well.
2015-10-12T09:18:12Z Xof: loke: alternatively, you are doing something wrong
2015-10-12T09:18:40Z loke: Xof: I see what you mean.
2015-10-12T09:18:58Z loke: You're saying it's already compiled by the time it hits the COMPILE call
2015-10-12T09:19:08Z Xof: I am!
2015-10-12T09:19:20Z loke: Hah, you are right of course.
2015-10-12T09:19:20Z loke: (compile 'foo '(lambda (x y) (1+ x)))
2015-10-12T09:19:25Z loke: gives the expected result.
2015-10-12T09:19:50Z loke: Thank you for clearing that up.
2015-10-12T09:21:11Z loke: Speaking of the spec... I have a question about how to read it.
2015-10-12T09:22:48Z loke: Some time ago, stassats was explaining to me that the spec said that SORT is guaranteed to sort a vector in-place, when I wondered why it only warns about not assigning the result when being called with a list, not an array
2015-10-12T09:23:31Z loke: The spec explictily says that the return value can be different for vectors, but he explained that that section didn't apply.
2015-10-12T09:23:40Z loke: Which sections does apply?
2015-10-12T09:25:01Z loke: I mean, how can I tell what and what does not apply? The page for SORT is the most confusing to me, but there are others that has had me stumped.
2015-10-12T09:27:56Z Xof: stassats might be referring to <http://www.lispworks.com/documentation/HyperSpec/Body/01_dc.htm>
2015-10-12T09:29:50Z loke: I see
2015-10-12T09:31:47Z loke: So I have to cut out Notes and Examples from my reading (which contains the two main points to suggest that it can return different values for vectors)
2015-10-12T09:32:29Z Shinmera: Also note http://cliki.net/Proposed%20ANSI%20Revisions%20and%20Clarifications
2015-10-12T09:34:05Z loke: Shinmera: Thanks, I never saw that one.
2015-10-12T09:34:20Z loke: It would be nice if it had a section on sort :-)
2015-10-12T09:34:44Z Shinmera: Well it's a wiki. You know what to do.
2015-10-12T09:35:47Z Xof: loke: you could also study existing practice.  Do any implementations actually return a different value if you give them a non-simple array?
2015-10-12T09:36:03Z loke: Shinmera: well, for this specific point I'm not qualified to write anything. When I read the specific page (even after excluding Notes and Examples) I still don't read it as guaranteeing in-place sort.
2015-10-12T09:36:25Z loke: Xof: No, but I can see how an implementation may end up doing it.
2015-10-12T09:37:21Z loke: That said, this is such an irrelevant matter anyway, since assigning the result of sort is perfectly harmless.
2015-10-12T09:59:09Z Ettore joined #sbcl
2015-10-12T10:17:27Z attila_lendvai joined #sbcl
2015-10-12T10:47:53Z mattprelude joined #sbcl
2015-10-12T11:05:56Z stassats joined #sbcl
2015-10-12T11:48:47Z Posterdati quit (Ping timeout: 264 seconds)
2015-10-12T11:49:22Z _iwc joined #sbcl
2015-10-12T11:56:54Z _iwc quit (K-Lined)
2015-10-12T12:15:06Z momo-reina joined #sbcl
2015-10-12T12:20:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-12T12:27:01Z stassats: i hear there are more ARM servers coming up, serving hunchentoot from an arm board would be a possibility
2015-10-12T12:27:14Z stassats: needs threads first, but that's just a SMOP
2015-10-12T12:43:58Z chris2 quit (Quit: trotz alledem!)
2015-10-12T13:36:28Z gingerale joined #sbcl
2015-10-12T13:45:58Z sjl joined #sbcl
2015-10-12T15:01:09Z stassats quit (Remote host closed the connection)
2015-10-12T15:34:31Z stassats joined #sbcl
2015-10-12T15:40:39Z Cymew quit (Ping timeout: 240 seconds)
2015-10-12T15:46:04Z momo-reina quit (Remote host closed the connection)
2015-10-12T15:46:25Z momo-reina joined #sbcl
2015-10-12T16:46:16Z zacts joined #sbcl
2015-10-12T16:58:36Z sjl quit (Ping timeout: 264 seconds)
2015-10-12T17:00:08Z psy_ joined #sbcl
2015-10-12T17:00:33Z psy_ quit (Max SendQ exceeded)
2015-10-12T17:01:12Z psy_ joined #sbcl
2015-10-12T17:41:37Z nikki93 joined #sbcl
2015-10-12T17:45:42Z sjl joined #sbcl
2015-10-12T17:47:29Z nikki93 quit (Remote host closed the connection)
2015-10-12T17:49:00Z momo-reina quit (Ping timeout: 244 seconds)
2015-10-12T17:55:19Z ASau joined #sbcl
2015-10-12T17:57:44Z ASau quit (Remote host closed the connection)
2015-10-12T17:58:11Z ASau joined #sbcl
2015-10-12T18:19:38Z zacts quit (Quit: ERC Version 5.3 (IRC client for Emacs))
2015-10-12T18:32:47Z gingerale quit (Remote host closed the connection)
2015-10-12T19:13:11Z Posterdati joined #sbcl
2015-10-12T19:18:59Z Posterdati quit (Ping timeout: 268 seconds)
2015-10-12T19:19:44Z scymtym_ joined #sbcl
2015-10-12T19:31:44Z Posterdati joined #sbcl
2015-10-12T19:58:45Z attila_lendvai joined #sbcl
2015-10-12T20:29:48Z karswell joined #sbcl
2015-10-12T20:32:45Z Bicyclidine joined #sbcl
2015-10-12T20:38:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-12T20:39:47Z whiteline joined #sbcl
2015-10-12T21:03:51Z Bicyclidine quit (Ping timeout: 256 seconds)
2015-10-12T21:19:01Z chris2 joined #sbcl
2015-10-12T21:20:32Z Bicyclidine joined #sbcl
2015-10-12T21:47:19Z sjl quit (Quit: WeeChat 1.3)
2015-10-12T21:49:25Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-12T22:07:09Z reb quit (Ping timeout: 255 seconds)
2015-10-12T22:21:21Z rpg joined #sbcl
2015-10-12T22:24:29Z abeaumont joined #sbcl
2015-10-12T22:29:32Z Ettore quit (Quit: Leaving.)
2015-10-12T23:26:52Z beaumonta joined #sbcl
2015-10-12T23:29:21Z abeaumont quit (Ping timeout: 268 seconds)
2015-10-12T23:44:09Z rpg quit (Quit: rpg)
2015-10-12T23:49:20Z mattprelude quit (Ping timeout: 272 seconds)
2015-10-12T23:55:39Z scymtym_ quit (Ping timeout: 246 seconds)
2015-10-12T23:58:15Z Bicyclidine quit (Quit: leaving)
2015-10-13T00:00:37Z christop1 quit (Ping timeout: 252 seconds)
2015-10-13T00:16:46Z christop1 joined #sbcl
2015-10-13T00:21:15Z stassats quit (Ping timeout: 255 seconds)
2015-10-13T00:22:23Z ralt quit (Ping timeout: 264 seconds)
2015-10-13T01:35:26Z hydan joined #sbcl
2015-10-13T01:50:27Z hydan quit (Remote host closed the connection)
2015-10-13T01:59:51Z nikki93 joined #sbcl
2015-10-13T02:05:24Z nikki93 quit (Remote host closed the connection)
2015-10-13T02:56:52Z ASau quit (Changing host)
2015-10-13T02:56:52Z ASau joined #sbcl
2015-10-13T03:22:43Z rpg joined #sbcl
2015-10-13T03:27:57Z kanru quit (Remote host closed the connection)
2015-10-13T03:33:09Z psy_ quit (Ping timeout: 256 seconds)
2015-10-13T03:54:11Z kanru joined #sbcl
2015-10-13T03:58:51Z heddwch quit (Read error: Connection reset by peer)
2015-10-13T04:00:44Z heddwch joined #sbcl
2015-10-13T04:09:39Z rpg quit (Quit: rpg)
2015-10-13T04:30:10Z dustinm` quit (Read error: Connection reset by peer)
2015-10-13T04:38:03Z dustinm` joined #sbcl
2015-10-13T04:39:14Z nikki93 joined #sbcl
2015-10-13T04:40:09Z jackc-_ joined #sbcl
2015-10-13T04:42:01Z jackc- quit (Ping timeout: 268 seconds)
2015-10-13T04:43:28Z Quadrescence joined #sbcl
2015-10-13T04:49:57Z psy_ joined #sbcl
2015-10-13T05:06:25Z nikki93 quit (Remote host closed the connection)
2015-10-13T05:14:25Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-13T05:23:15Z karswell quit (Read error: Connection reset by peer)
2015-10-13T05:23:30Z karswell joined #sbcl
2015-10-13T05:27:26Z karswell quit (Read error: Connection reset by peer)
2015-10-13T05:47:53Z Shinmera joined #sbcl
2015-10-13T06:25:23Z gingerale joined #sbcl
2015-10-13T06:25:35Z irsol quit (Ping timeout: 250 seconds)
2015-10-13T06:27:09Z irsol joined #sbcl
2015-10-13T06:45:00Z Travis__ joined #sbcl
2015-10-13T06:48:56Z Travis__ left #sbcl
2015-10-13T06:54:26Z psy_ quit (Remote host closed the connection)
2015-10-13T07:21:46Z ASau quit (Ping timeout: 240 seconds)
2015-10-13T07:28:27Z Cymew joined #sbcl
2015-10-13T07:36:32Z jlarocco_ quit (Ping timeout: 268 seconds)
2015-10-13T07:44:52Z zacts joined #sbcl
2015-10-13T08:42:31Z attila_lendvai joined #sbcl
2015-10-13T08:45:52Z DeadTrickster quit (*.net *.split)
2015-10-13T08:46:08Z DeadTrickster joined #sbcl
2015-10-13T09:00:34Z heddwch quit (Read error: Connection reset by peer)
2015-10-13T09:01:02Z heddwch joined #sbcl
2015-10-13T10:14:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-13T11:03:00Z stassats joined #sbcl
2015-10-13T11:35:21Z Bike quit (Ping timeout: 255 seconds)
2015-10-13T12:30:41Z christop1 is now known as christoph
2015-10-13T12:31:10Z christoph is now known as Guest3619
2015-10-13T12:32:50Z Guest3619 is now known as christoph_d
2015-10-13T12:32:55Z christoph_d is now known as christoph_e
2015-10-13T13:50:07Z eudoxia joined #sbcl
2015-10-13T14:03:47Z reb joined #sbcl
2015-10-13T14:27:46Z attila_lendvai joined #sbcl
2015-10-13T14:45:33Z stassats quit (Remote host closed the connection)
2015-10-13T14:52:46Z Cymew quit (Ping timeout: 244 seconds)
2015-10-13T15:37:26Z sjl joined #sbcl
2015-10-13T15:38:41Z scymtym_ joined #sbcl
2015-10-13T15:42:32Z stassats joined #sbcl
2015-10-13T15:52:04Z stassats quit (Remote host closed the connection)
2015-10-13T16:01:59Z jlarocco quit (Ping timeout: 240 seconds)
2015-10-13T16:10:54Z stassats joined #sbcl
2015-10-13T16:18:09Z evdev-ebev quit (Ping timeout: 246 seconds)
2015-10-13T16:22:35Z carvite quit (Ping timeout: 250 seconds)
2015-10-13T16:32:22Z Bike joined #sbcl
2015-10-13T16:37:33Z eudoxia_ joined #sbcl
2015-10-13T16:38:24Z eudoxia quit (Read error: Connection reset by peer)
2015-10-13T16:43:04Z attila_lendvai: the use of expand for building makes it hard to build sbcl on yocto linux (playing with an intel edison board)
2015-10-13T16:43:25Z psy_ joined #sbcl
2015-10-13T16:46:26Z gabriel_laddel joined #sbcl
2015-10-13T16:58:15Z eudoxia_ quit (Quit: Leaving)
2015-10-13T17:07:49Z jlarocco joined #sbcl
2015-10-13T17:23:54Z heddwch quit (Ping timeout: 265 seconds)
2015-10-13T17:28:45Z heddwch joined #sbcl
2015-10-13T17:33:34Z heddwch quit (Ping timeout: 265 seconds)
2015-10-13T17:34:35Z heddwch joined #sbcl
2015-10-13T17:43:14Z heddwch quit (Ping timeout: 265 seconds)
2015-10-13T17:44:15Z heddwch joined #sbcl
2015-10-13T17:45:18Z heddwch quit (Excess Flood)
2015-10-13T17:45:25Z heddwch joined #sbcl
2015-10-13T18:21:11Z whiteline quit (Ping timeout: 264 seconds)
2015-10-13T18:23:22Z carvite joined #sbcl
2015-10-13T18:37:19Z scymtym_ quit (Ping timeout: 240 seconds)
2015-10-13T18:41:15Z mattprelude joined #sbcl
2015-10-13T18:41:53Z zacts: how would one port sbcl to a new platform? (namely http://minix3.org on the i386 platform)
2015-10-13T18:42:03Z zacts: s/that last 'platform'/architecture/
2015-10-13T18:43:17Z Xof: on a new OS, the main things to port are in the src/runtime directory
2015-10-13T18:43:35Z Xof: you could take a look to see how different the x86/BSD and x86/Linux builds are, and work out from that what to change for Minix
2015-10-13T18:43:49Z Xof: mostly what you will need to deal with are ABI, signal handling, and threads
2015-10-13T18:45:12Z zacts: ok
2015-10-13T18:45:26Z zacts: and would I need to utilize gnu clisp to crossbuild sbcl to minix?
2015-10-13T18:45:38Z zacts: or could I crossbuild from another platform, say Linux or NetBSD?
2015-10-13T18:46:59Z Xof: you can cross-build from another platform
2015-10-13T18:47:13Z zacts: ah ok neat
2015-10-13T18:47:16Z zacts: thanks
2015-10-13T18:47:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-13T18:48:07Z Xof: run make-config.sh on your target (minix) machine with --host-location
2015-10-13T18:48:51Z Xof: then run make-host-1.sh on your host, make-target-1.sh on your target, make-host-2.sh on your host, make-target-2.sh on your target
2015-10-13T18:51:28Z clique joined #sbcl
2015-10-13T19:00:44Z sjl quit (Ping timeout: 272 seconds)
2015-10-13T19:09:24Z Bicyclidine joined #sbcl
2015-10-13T19:12:12Z Ettore joined #sbcl
2015-10-13T19:25:31Z clique quit (Quit: Page closed)
2015-10-13T19:28:58Z gingerale quit (Remote host closed the connection)
2015-10-13T19:38:49Z whiteline joined #sbcl
2015-10-13T19:41:06Z pkhuong quit (Ping timeout: 240 seconds)
2015-10-13T19:46:10Z reb quit (Remote host closed the connection)
2015-10-13T19:48:49Z pkhuong joined #sbcl
2015-10-13T19:49:12Z pkhuong is now known as Guest62169
2015-10-13T19:50:50Z whiteline: are there any plans to expose the zlib functions used to compress image code as an extension?
2015-10-13T19:53:45Z prxq joined #sbcl
2015-10-13T19:54:46Z reb joined #sbcl
2015-10-13T19:55:56Z mattprelude quit (Ping timeout: 268 seconds)
2015-10-13T19:56:42Z Xof: I doubt it
2015-10-13T19:59:30Z stassats: you can do it trivially yourself
2015-10-13T19:59:36Z stassats: just as any other ffi
2015-10-13T20:05:06Z attila_lendvai joined #sbcl
2015-10-13T20:23:44Z stassats quit (Remote host closed the connection)
2015-10-13T20:48:29Z beaumonta is now known as abeaumont
2015-10-13T20:49:39Z gabriel_laddel quit (Remote host closed the connection)
2015-10-13T20:54:39Z scymtym_ joined #sbcl
2015-10-13T20:56:19Z ASau joined #sbcl
2015-10-13T20:59:08Z scymtym_: attila_lendvai: would something like https://github.com/scymtym/sbcl/commit/290b334784a06483d17365812173fecd9e88650b help?
2015-10-13T21:01:59Z attila_lendvai: scymtym, let me start a build and see if it works
2015-10-13T21:02:34Z scymtym_: i should eliminate the only use of expand i could find
2015-10-13T21:02:42Z scymtym_: s/i/it/
2015-10-13T21:03:17Z attila_lendvai: how nice that github supports zip download... there's no git in the repo on yokto linux
2015-10-13T21:05:25Z attila_lendvai: scymtym: I got an ascii decoding error on tests/compiler.pure.lisp
2015-10-13T21:06:10Z attila_lendvai: but when I've unwound from the sbcl debugger it continued building
2015-10-13T21:06:49Z attila_lendvai: another positive change is that I only had to chmod +x make.sh, not the othes in tools-for-build/
2015-10-13T21:07:55Z scymtym_: attila_lendvai: was it running tools-for-build/canonicalize-whitespace.lisp when the error occurred?
2015-10-13T21:10:15Z attila_lendvai: scymtym: CANONICALIZE-WHITESPACE is in the backtrace
2015-10-13T21:10:15Z scymtym_: and tests/compiler.pure.lisp apparently has non-ASCII characters
2015-10-13T21:10:26Z scymtym_: i should probably unconditionally use UTF-8 external format
2015-10-13T21:12:26Z karswell joined #sbcl
2015-10-13T21:13:05Z scymtym_: attila_lendvai: can you try https://github.com/scymtym/sbcl/commit/92ad5ab7e9daee7b2309de7dfd6cc881dc2f538a
2015-10-13T21:13:33Z Quadrescence joined #sbcl
2015-10-13T21:16:02Z zacts quit (Read error: Connection reset by peer)
2015-10-13T21:16:09Z attila_lendvai: scymtym: now it's utf-8 error on tests/data/compile-file-pos-utf16be.lisp :)
2015-10-13T21:16:38Z scymtym_: no fair!
2015-10-13T21:26:01Z ccl-logbot joined #sbcl
2015-10-13T21:26:01Z 
2015-10-13T21:26:01Z names: ccl-logbot Quadrescence karswell ASau scymtym_ attila_lendvai reb prxq Guest62169 whiteline Ettore Bicyclidine carvite heddwch jlarocco psy_ Bike DeadTrickster irsol jackc-_ dustinm` kanru christoph_e abeaumont chris2 Posterdati PuercoPop foom dougk_ rj-code Ober_ jdz_ minion antoszka mood nyef cades fikusz White_Flame psilord tifa flip214 scymtym angavrilov DGASAU hzp ym Xof drmeister specbot |3b| brucem nicdev joshe jsnell salva loke alchemis7 dim dsp_
2015-10-13T21:26:01Z names: Blkt p_l Intensity fe[nl]ix chu frankS2 failproofshark jozip pootler ferada jackdaniel luis- pchrist sigjuice les yauz
2015-10-13T21:28:53Z attila_lendvai: scymtym: I'm about to head bedwards... let me know if you need a little testing in the next few minutes, or otherwise I'll be around later tomorrow.
2015-10-13T21:29:20Z scymtym_: attila_lendvai: it should be fine, thanks
2015-10-13T21:29:31Z scymtym_: only two files are not UTF-8-decodable
2015-10-13T21:29:46Z scymtym_: reader.impure.lisp uses latin-1 encoding, probably because credit for a bug report is given to Levente Mészáros
2015-10-13T21:30:24Z scymtym_: i will push a fix and you can test at your discretion, if that's OK
2015-10-13T21:39:50Z karswell quit (Read error: Connection reset by peer)
2015-10-13T21:40:01Z karswell joined #sbcl
2015-10-13T21:47:56Z karswell quit (Remote host closed the connection)
2015-10-13T21:48:03Z karswell joined #sbcl
2015-10-13T21:54:53Z attila_lendvai: scymtym: got some "reading an ignored variable: STREAM", but otherwise build is running
2015-10-13T21:55:25Z scymtym_: hm, put that in for ccl, forgot to re-check with sbcl
2015-10-13T22:01:15Z prxq quit (Remote host closed the connection)
2015-10-13T22:14:55Z stassats joined #sbcl
2015-10-13T22:20:18Z attila_lendvai: that damn linux-nm was not executable, build has stopped somewhere in the middle...
2015-10-13T22:22:09Z stassats: nm is ran at the one make-target-1
2015-10-13T22:22:29Z stassats: if you don't want to repeat steps, you can just do make-target-1, make-host-2, make-target-2, make-target-contribs
2015-10-13T22:22:42Z stassats: that'll only save you make-host-1, i.e. not much
2015-10-13T22:25:54Z attila_lendvai: ok, then this atom thingy is not fast... :)
2015-10-13T22:27:38Z stassats: do the existing builds not work?
2015-10-13T22:27:44Z stassats: is that a library issue?
2015-10-13T22:28:12Z stassats: i'd just build a core with sb-dynamic-core, then recompile the runtime as needed
2015-10-13T22:29:42Z attila_lendvai: the x86 build is kinda old, but works. that's what I'm using for bootstrap... I just wanted to run a quick rebuild if it's a full linux after all. not crucial, though.
2015-10-13T22:30:03Z stassats: it's 64-bit, ain't it?
2015-10-13T22:31:20Z stassats: anyhow, you could make a x86 build on a faster machine
2015-10-13T22:34:34Z attila_lendvai: it's seems to be 32 bits, https://en.wikipedia.org/wiki/Intel_Edison
2015-10-13T22:34:44Z stassats: ok, so it's not an atom
2015-10-13T22:38:21Z sjl joined #sbcl
2015-10-13T23:10:17Z karswell quit (Read error: Connection reset by peer)
2015-10-13T23:10:33Z karswell joined #sbcl
2015-10-13T23:14:02Z attila_lendvai: the build run successfully. the linux is certainly 32 bits according to uname, and the wikipedia page says it's a dual atom. not fast, but it runs on 500mA USB and hardly got warm.
2015-10-13T23:14:29Z attila_lendvai heads to bed
2015-10-13T23:16:15Z stassats: well, if it's an atom, then it's 64-bit
2015-10-13T23:16:26Z stassats: the kernel must be just 32-bit
2015-10-13T23:17:10Z stassats: unless they are using some cut down atoms
2015-10-13T23:17:22Z stassats: which is pretty unlikely
2015-10-13T23:18:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-13T23:20:32Z Ettore quit (Quit: Leaving.)
2015-10-13T23:38:33Z scymtym_ quit (Ping timeout: 268 seconds)
2015-10-13T23:44:35Z Bicyclidine quit (Quit: moheee)
2015-10-14T00:19:08Z karswell quit (Read error: Connection reset by peer)
2015-10-14T00:35:20Z DeadTrickster quit (Remote host closed the connection)
2015-10-14T00:37:19Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-14T00:38:28Z DeadTrickster joined #sbcl
2015-10-14T00:39:36Z karswell joined #sbcl
2015-10-14T01:27:42Z whiteline quit (Ping timeout: 268 seconds)
2015-10-14T03:32:34Z Bike quit (Read error: No route to host)
2015-10-14T03:33:04Z Bike joined #sbcl
2015-10-14T03:33:59Z psy_ quit (Ping timeout: 260 seconds)
2015-10-14T03:44:17Z Quadrescence joined #sbcl
2015-10-14T04:03:37Z stassats quit (Ping timeout: 252 seconds)
2015-10-14T04:39:02Z psy_ joined #sbcl
2015-10-14T04:39:33Z psy_ quit (Max SendQ exceeded)
2015-10-14T04:40:50Z psy_ joined #sbcl
2015-10-14T05:34:37Z gingerale joined #sbcl
2015-10-14T06:14:29Z Shinmera joined #sbcl
2015-10-14T06:21:29Z ASau quit (Ping timeout: 252 seconds)
2015-10-14T06:32:58Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-14T07:06:41Z Cymew joined #sbcl
2015-10-14T07:11:29Z Cymew quit (Read error: Connection reset by peer)
2015-10-14T07:16:34Z Cymew joined #sbcl
2015-10-14T07:17:14Z psy_ quit (Read error: Connection reset by peer)
2015-10-14T08:03:24Z Ettore joined #sbcl
2015-10-14T09:00:33Z heddwch quit (Read error: Connection reset by peer)
2015-10-14T09:01:01Z heddwch joined #sbcl
2015-10-14T09:22:38Z bsmr joined #sbcl
2015-10-14T09:30:30Z mattprelude joined #sbcl
2015-10-14T09:57:38Z whiteline joined #sbcl
2015-10-14T09:58:01Z Ettore quit (Quit: Leaving.)
2015-10-14T10:01:03Z Ettore joined #sbcl
2015-10-14T11:31:46Z attila_lendvai joined #sbcl
2015-10-14T11:36:52Z rszeno joined #sbcl
2015-10-14T12:03:33Z sjl quit (Ping timeout: 246 seconds)
2015-10-14T12:04:20Z stassats joined #sbcl
2015-10-14T12:14:58Z rszeno quit (Quit: Leaving.)
2015-10-14T12:35:44Z bsmr quit (Quit: Leaving)
2015-10-14T13:03:25Z stassats: i wonder if having two alloc_tramps would be beneficial, the second one which doesn't save FPU registers
2015-10-14T13:04:24Z stassats: on arm that would save 512 bytes, 256 bytes on x86-64
2015-10-14T13:16:41Z eudoxia joined #sbcl
2015-10-14T14:37:54Z Cymew quit (Ping timeout: 246 seconds)
2015-10-14T14:45:07Z stassats quit (Ping timeout: 256 seconds)
2015-10-14T15:46:25Z hlavaty joined #sbcl
2015-10-14T15:47:23Z psy_ joined #sbcl
2015-10-14T15:47:53Z psy_ quit (Max SendQ exceeded)
2015-10-14T15:48:20Z psy_ joined #sbcl
2015-10-14T16:18:40Z nikki93 joined #sbcl
2015-10-14T16:35:27Z sjl joined #sbcl
2015-10-14T16:40:11Z yauz_2 joined #sbcl
2015-10-14T16:41:40Z eudoxia_ joined #sbcl
2015-10-14T16:42:42Z eudoxia quit (Read error: Connection reset by peer)
2015-10-14T16:44:26Z les` joined #sbcl
2015-10-14T16:50:01Z karswell quit (*.net *.split)
2015-10-14T16:50:01Z Guest62169 quit (*.net *.split)
2015-10-14T16:50:01Z Ober_ quit (*.net *.split)
2015-10-14T16:50:02Z failproofshark quit (*.net *.split)
2015-10-14T16:50:03Z les quit (*.net *.split)
2015-10-14T16:50:03Z yauz quit (*.net *.split)
2015-10-14T16:50:03Z nikki93 quit (*.net *.split)
2015-10-14T16:50:04Z heddwch quit (*.net *.split)
2015-10-14T16:50:04Z carvite quit (*.net *.split)
2015-10-14T16:50:04Z tifa quit (*.net *.split)
2015-10-14T16:50:04Z ym quit (*.net *.split)
2015-10-14T16:50:04Z chu quit (*.net *.split)
2015-10-14T16:50:05Z sigjuice quit (*.net *.split)
2015-10-14T16:52:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-14T16:53:59Z eudoxia_ quit (Quit: Leaving)
2015-10-14T16:59:55Z nikki93 joined #sbcl
2015-10-14T16:59:55Z heddwch joined #sbcl
2015-10-14T16:59:55Z carvite joined #sbcl
2015-10-14T16:59:55Z tifa joined #sbcl
2015-10-14T16:59:55Z ym joined #sbcl
2015-10-14T16:59:55Z chu joined #sbcl
2015-10-14T16:59:55Z sigjuice joined #sbcl
2015-10-14T17:07:14Z nikki93 quit (*.net *.split)
2015-10-14T17:07:14Z heddwch quit (*.net *.split)
2015-10-14T17:07:15Z carvite quit (*.net *.split)
2015-10-14T17:07:15Z tifa quit (*.net *.split)
2015-10-14T17:07:15Z ym quit (*.net *.split)
2015-10-14T17:07:15Z chu quit (*.net *.split)
2015-10-14T17:07:15Z sigjuice quit (*.net *.split)
2015-10-14T17:07:37Z Ober joined #sbcl
2015-10-14T17:07:38Z pkhuong joined #sbcl
2015-10-14T17:07:43Z karswell joined #sbcl
2015-10-14T17:07:43Z failproofshark joined #sbcl
2015-10-14T17:07:59Z pkhuong is now known as Guest21642
2015-10-14T17:08:41Z attila_lendvai joined #sbcl
2015-10-14T17:08:41Z attila_lendvai quit (Changing host)
2015-10-14T17:08:41Z attila_lendvai joined #sbcl
2015-10-14T17:10:26Z nikki93 joined #sbcl
2015-10-14T17:10:26Z heddwch joined #sbcl
2015-10-14T17:10:26Z carvite joined #sbcl
2015-10-14T17:10:26Z tifa joined #sbcl
2015-10-14T17:10:26Z ym joined #sbcl
2015-10-14T17:10:26Z chu joined #sbcl
2015-10-14T17:10:26Z sigjuice joined #sbcl
2015-10-14T17:19:19Z zacts joined #sbcl
2015-10-14T17:24:46Z nikki93 quit (Remote host closed the connection)
2015-10-14T17:52:23Z psy_ quit (Read error: Connection reset by peer)
2015-10-14T17:52:39Z ASau joined #sbcl
2015-10-14T17:54:37Z psy_ joined #sbcl
2015-10-14T18:03:44Z stassats joined #sbcl
2015-10-14T18:05:51Z zacts quit (Ping timeout: 246 seconds)
2015-10-14T18:08:48Z jlarocco quit (Quit: Leaving)
2015-10-14T18:35:16Z jlarocco joined #sbcl
2015-10-14T18:40:01Z eudoxia joined #sbcl
2015-10-14T18:56:07Z fikusz quit (Quit: Leaving)
2015-10-14T19:00:59Z fikusz joined #sbcl
2015-10-14T19:34:15Z clique joined #sbcl
2015-10-14T19:46:42Z scymtym_ joined #sbcl
2015-10-14T19:48:23Z clique left #sbcl
2015-10-14T20:01:01Z psilord1 joined #sbcl
2015-10-14T20:05:35Z gingerale quit (Remote host closed the connection)
2015-10-14T20:14:33Z clique joined #sbcl
2015-10-14T20:29:02Z clique quit (Quit: Page closed)
2015-10-14T20:43:17Z Posterdati quit (Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/)
2015-10-14T20:46:26Z Posterdati joined #sbcl
2015-10-14T21:22:17Z Bicyclidine joined #sbcl
2015-10-14T21:37:35Z psilord1 quit (Quit: Leaving.)
2015-10-14T21:47:15Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-14T21:47:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-14T22:01:30Z eudoxia quit (Quit: Leaving)
2015-10-14T22:05:01Z Ettore quit (Quit: Leaving.)
2015-10-14T22:45:43Z mattprelude quit (Ping timeout: 260 seconds)
2015-10-14T23:52:11Z Bicyclidine quit (Quit: later)
2015-10-15T00:11:54Z scymtym_ quit (Ping timeout: 246 seconds)
2015-10-15T00:13:19Z hydan joined #sbcl
2015-10-15T00:15:33Z hlavaty` joined #sbcl
2015-10-15T00:15:59Z hlavaty quit (Ping timeout: 240 seconds)
2015-10-15T02:44:15Z Guest21642 is now known as pkhuong
2015-10-15T03:09:25Z Quadrescence joined #sbcl
2015-10-15T03:15:30Z stassats quit (Ping timeout: 240 seconds)
2015-10-15T03:24:28Z Intensity quit (Ping timeout: 244 seconds)
2015-10-15T03:30:07Z Intensity joined #sbcl
2015-10-15T03:39:36Z psy_ quit (Ping timeout: 268 seconds)
2015-10-15T03:42:15Z hydan quit (Remote host closed the connection)
2015-10-15T03:44:10Z hydan joined #sbcl
2015-10-15T04:29:17Z nikki93 joined #sbcl
2015-10-15T04:38:09Z psy_ joined #sbcl
2015-10-15T04:55:32Z nikki93 quit (Remote host closed the connection)
2015-10-15T05:53:38Z psy_ quit (Read error: Connection reset by peer)
2015-10-15T07:18:57Z ASau quit (Ping timeout: 265 seconds)
2015-10-15T07:19:25Z gingerale joined #sbcl
2015-10-15T08:25:01Z scymtym quit (Remote host closed the connection)
2015-10-15T08:34:34Z nikki93 joined #sbcl
2015-10-15T08:36:45Z Cymew joined #sbcl
2015-10-15T08:38:56Z Cymew quit (Remote host closed the connection)
2015-10-15T08:40:47Z nikki93 quit (Remote host closed the connection)
2015-10-15T08:42:46Z nikki93 joined #sbcl
2015-10-15T08:52:25Z Cymew joined #sbcl
2015-10-15T08:53:36Z attila_lendvai joined #sbcl
2015-10-15T08:53:36Z attila_lendvai quit (Changing host)
2015-10-15T08:53:36Z attila_lendvai joined #sbcl
2015-10-15T09:00:47Z heddwch quit (Read error: Connection reset by peer)
2015-10-15T09:01:32Z heddwch joined #sbcl
2015-10-15T09:04:32Z Ettore joined #sbcl
2015-10-15T09:09:47Z scymtym joined #sbcl
2015-10-15T09:42:50Z mattprelude joined #sbcl
2015-10-15T10:42:56Z nikki93 quit (Remote host closed the connection)
2015-10-15T12:01:51Z stassats joined #sbcl
2015-10-15T12:08:37Z _iwc joined #sbcl
2015-10-15T12:21:28Z _iwc quit (Killed (Sigyn (Spam is off topic on freenode.)))
2015-10-15T12:22:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-15T12:29:44Z chu quit (Ping timeout: 265 seconds)
2015-10-15T12:34:24Z _iwc joined #sbcl
2015-10-15T12:46:51Z _iwc quit (K-Lined)
2015-10-15T13:01:40Z rszeno joined #sbcl
2015-10-15T13:11:23Z psilord quit (Quit: Leaving.)
2015-10-15T13:36:12Z Cymew quit (Ping timeout: 246 seconds)
2015-10-15T13:43:11Z Cymew joined #sbcl
2015-10-15T13:46:54Z rszeno quit (Quit: Leaving.)
2015-10-15T14:05:59Z psilord joined #sbcl
2015-10-15T14:15:19Z eudoxia joined #sbcl
2015-10-15T14:20:43Z Cymew quit (Read error: Connection reset by peer)
2015-10-15T15:02:32Z scymtym_ joined #sbcl
2015-10-15T15:42:19Z attila_lendvai joined #sbcl
2015-10-15T16:17:06Z dra joined #sbcl
2015-10-15T16:24:07Z dra: Hi all. Consider these two defuns: (defun foo (x y) (+ x y)) and (defun bar () (foo 7 11)). Now, disassembling 'bar (x86-64) shows these two lines: MOV EDX 14 and MOV EDI 22. Why are my arguments multiplied by two? Thanks in advance.
2015-10-15T16:25:07Z heddwch quit (Ping timeout: 265 seconds)
2015-10-15T16:28:17Z Xof: because fixnums have a tag bit
2015-10-15T16:28:33Z scymtym_: dra: when sbcl passes objects across function boundaries, the objects are tagged so that they carry their (approximate) type with them. for fixnums (small integers), the tag information can … what Xof said :)
2015-10-15T16:29:24Z dra: Xof, scymtym_: Thanks!
2015-10-15T16:41:10Z sjl quit (Ping timeout: 240 seconds)
2015-10-15T16:46:12Z eudoxia_ joined #sbcl
2015-10-15T16:46:12Z eudoxia_ quit (Remote host closed the connection)
2015-10-15T16:50:11Z eudoxia quit (Ping timeout: 268 seconds)
2015-10-15T16:51:14Z sjl joined #sbcl
2015-10-15T16:56:11Z stassats: dra: moreover, all pointers are double-word aligned (16 byte), which means the low 4 bits are always zero
2015-10-15T16:56:31Z stassats: those can be used for tags, and masked off when pointers are accessed
2015-10-15T16:57:25Z stassats: fixnums claim all the tags with the first bit being 0, that leaves all the tags with the first bit 1 to other things, i.e. 3 bits for the tag
2015-10-15T16:57:55Z stassats: there are other immediate things, like characters and single floats, the rest are pointers
2015-10-15T17:01:00Z psy joined #sbcl
2015-10-15T17:01:32Z dra: stassats: Thanks a bunch, now all those weird offsets make a lot more sense.
2015-10-15T17:02:39Z stassats: x86's complicated effective addresses are really good at that
2015-10-15T17:04:21Z Ettore quit (Quit: Leaving.)
2015-10-15T17:04:36Z Ettore joined #sbcl
2015-10-15T17:07:08Z dra: stassats: Next line is MOV RAX, [RIP-93]. That address isn't in the disassembled code. Does SBCL let me inspect more of the compiled function?
2015-10-15T17:07:30Z stassats: that's loading a constant, if you inspect it in slime you can see them
2015-10-15T17:07:55Z dra: Ah nice, thanks.
2015-10-15T17:08:02Z stassats: it should also annotate what the constant is
2015-10-15T17:09:12Z dra: Yes, it's the definition of #'foo, see above.
2015-10-15T17:09:52Z dra: (Which I could just disassemble, too...)
2015-10-15T17:10:04Z stassats: slime only shows boxed constants, there are also unboxed constants
2015-10-15T17:35:26Z clique joined #sbcl
2015-10-15T17:38:14Z sjl quit (Ping timeout: 250 seconds)
2015-10-15T17:40:03Z dra quit (Quit: Leaving)
2015-10-15T18:07:57Z ASau joined #sbcl
2015-10-15T18:16:36Z scymtym_ quit (Remote host closed the connection)
2015-10-15T18:18:41Z scymtym_ joined #sbcl
2015-10-15T18:33:27Z heddwch joined #sbcl
2015-10-15T18:33:45Z clique quit (Quit: Page closed)
2015-10-15T18:49:57Z sjl joined #sbcl
2015-10-15T18:54:28Z sjl quit (Ping timeout: 244 seconds)
2015-10-15T19:05:21Z luis- quit (Ping timeout: 255 seconds)
2015-10-15T19:12:14Z luis joined #sbcl
2015-10-15T19:16:57Z luis quit (Ping timeout: 268 seconds)
2015-10-15T19:17:05Z luis joined #sbcl
2015-10-15T19:24:43Z gingerale quit (Remote host closed the connection)
2015-10-15T19:28:44Z clique joined #sbcl
2015-10-15T19:48:05Z clique quit (Quit: Page closed)
2015-10-15T20:06:52Z luis quit (Ping timeout: 250 seconds)
2015-10-15T20:07:07Z luis joined #sbcl
2015-10-15T20:26:00Z clique joined #sbcl
2015-10-15T20:36:06Z sjl joined #sbcl
2015-10-15T20:40:21Z sjl quit (Ping timeout: 250 seconds)
2015-10-15T20:55:27Z psilord quit (Quit: Leaving.)
2015-10-15T20:55:49Z psilord joined #sbcl
2015-10-15T21:14:02Z angavrilov quit (Remote host closed the connection)
2015-10-15T21:17:14Z jdz_ is now known as jdz
2015-10-15T21:19:23Z abeaumont quit (Remote host closed the connection)
2015-10-15T21:21:30Z abeaumont joined #sbcl
2015-10-15T21:26:43Z clique quit (Quit: Page closed)
2015-10-15T22:21:50Z sjl joined #sbcl
2015-10-15T22:26:10Z sjl quit (Ping timeout: 240 seconds)
2015-10-15T22:36:37Z psilord quit (Quit: Leaving.)
2015-10-15T23:23:41Z Ettore quit (Quit: Leaving.)
2015-10-15T23:26:45Z ASau` joined #sbcl
2015-10-15T23:30:17Z ASau quit (Write error: Connection reset by peer)
2015-10-15T23:32:28Z ASau` is now known as ASau
2015-10-15T23:34:19Z Quadrescence quit (Ping timeout: 240 seconds)
2015-10-15T23:34:21Z sigjuice quit (Ping timeout: 265 seconds)
2015-10-15T23:34:48Z heddwch quit (Ping timeout: 265 seconds)
2015-10-15T23:34:49Z carvite quit (Ping timeout: 265 seconds)
2015-10-15T23:34:49Z tifa quit (Ping timeout: 265 seconds)
2015-10-15T23:34:49Z ym quit (Ping timeout: 265 seconds)
2015-10-15T23:36:39Z carvite joined #sbcl
2015-10-15T23:37:50Z psilord joined #sbcl
2015-10-15T23:39:52Z sigjuice joined #sbcl
2015-10-15T23:59:18Z carvite quit (*.net *.split)
2015-10-15T23:59:19Z fikusz quit (*.net *.split)
2015-10-15T23:59:20Z rj-code quit (*.net *.split)
2015-10-15T23:59:20Z sigjuice quit (*.net *.split)
2015-10-15T23:59:21Z scymtym quit (*.net *.split)
2015-10-15T23:59:21Z hlavaty` quit (*.net *.split)
2015-10-15T23:59:21Z Bike quit (*.net *.split)
2015-10-15T23:59:22Z PuercoPop quit (*.net *.split)
2015-10-15T23:59:22Z minion quit (*.net *.split)
2015-10-15T23:59:22Z mood quit (*.net *.split)
2015-10-15T23:59:22Z nicdev quit (*.net *.split)
2015-10-15T23:59:23Z brucem quit (*.net *.split)
2015-10-15T23:59:24Z ASau quit (*.net *.split)
2015-10-15T23:59:24Z psy quit (*.net *.split)
2015-10-15T23:59:24Z failproofshark quit (*.net *.split)
2015-10-15T23:59:24Z yauz_2 quit (*.net *.split)
2015-10-15T23:59:26Z p_l quit (*.net *.split)
2015-10-15T23:59:26Z frankS2 quit (*.net *.split)
2015-10-15T23:59:26Z jozip quit (*.net *.split)
2015-10-15T23:59:26Z pootler quit (*.net *.split)
2015-10-15T23:59:26Z Posterdati quit (*.net *.split)
2015-10-15T23:59:27Z Ober quit (*.net *.split)
2015-10-15T23:59:27Z hzp quit (*.net *.split)
2015-10-15T23:59:29Z foom quit (*.net *.split)
2015-10-15T23:59:29Z dougk_ quit (*.net *.split)
2015-10-15T23:59:29Z antoszka quit (*.net *.split)
2015-10-15T23:59:29Z nyef quit (*.net *.split)
2015-10-15T23:59:29Z White_Flame quit (*.net *.split)
2015-10-15T23:59:29Z Xof quit (*.net *.split)
2015-10-15T23:59:30Z mattprelude quit (*.net *.split)
2015-10-15T23:59:30Z les` quit (*.net *.split)
2015-10-15T23:59:31Z dim quit (*.net *.split)
2015-10-15T23:59:32Z hydan quit (*.net *.split)
2015-10-15T23:59:32Z christoph_e quit (*.net *.split)
2015-10-15T23:59:33Z Blkt quit (*.net *.split)
2015-10-15T23:59:34Z fe[nl]ix quit (*.net *.split)
2015-10-15T23:59:34Z attila_lendvai quit (*.net *.split)
2015-10-15T23:59:34Z DeadTrickster quit (*.net *.split)
2015-10-15T23:59:35Z DGASAU quit (*.net *.split)
2015-10-15T23:59:35Z joshe quit (*.net *.split)
2015-10-15T23:59:36Z ferada quit (*.net *.split)
2015-10-15T23:59:37Z kanru quit (*.net *.split)
2015-10-15T23:59:37Z specbot quit (*.net *.split)
2015-10-15T23:59:37Z jsnell quit (*.net *.split)
2015-10-15T23:59:37Z salva quit (*.net *.split)
2015-10-15T23:59:38Z luis quit (*.net *.split)
2015-10-15T23:59:38Z whiteline quit (*.net *.split)
2015-10-15T23:59:39Z flip214 quit (*.net *.split)
2015-10-15T23:59:39Z psilord quit (*.net *.split)
2015-10-15T23:59:40Z pkhuong quit (*.net *.split)
2015-10-15T23:59:40Z jackc-_ quit (*.net *.split)
2015-10-15T23:59:40Z chris2 quit (*.net *.split)
2015-10-15T23:59:41Z jdz quit (*.net *.split)
2015-10-15T23:59:41Z cades quit (*.net *.split)
2015-10-15T23:59:41Z drmeister quit (*.net *.split)
2015-10-15T23:59:41Z |3b| quit (*.net *.split)
2015-10-15T23:59:41Z jackdaniel quit (*.net *.split)
2015-10-15T23:59:42Z pchrist quit (*.net *.split)
2015-10-15T23:59:42Z dustinm` quit (Max SendQ exceeded)
2015-10-16T00:06:31Z fikusz joined #sbcl
2015-10-16T00:06:31Z tifa joined #sbcl
2015-10-16T00:06:31Z rj-code joined #sbcl
2015-10-16T00:06:31Z dustinm` joined #sbcl
2015-10-16T00:07:28Z kanru joined #sbcl
2015-10-16T00:07:28Z specbot joined #sbcl
2015-10-16T00:07:28Z jsnell joined #sbcl
2015-10-16T00:07:28Z salva joined #sbcl
2015-10-16T00:08:56Z carvite joined #sbcl
2015-10-16T00:09:52Z luis joined #sbcl
2015-10-16T00:09:52Z flip214 joined #sbcl
2015-10-16T00:13:12Z hydan joined #sbcl
2015-10-16T00:13:12Z christoph_e joined #sbcl
2015-10-16T00:13:12Z Blkt joined #sbcl
2015-10-16T00:13:12Z fe[nl]ix joined #sbcl
2015-10-16T00:13:16Z frankS2 joined #sbcl
2015-10-16T00:13:16Z foom joined #sbcl
2015-10-16T00:13:16Z dougk_ joined #sbcl
2015-10-16T00:13:16Z antoszka joined #sbcl
2015-10-16T00:13:16Z nyef joined #sbcl
2015-10-16T00:13:16Z White_Flame joined #sbcl
2015-10-16T00:13:16Z Xof joined #sbcl
2015-10-16T00:13:28Z attila_lendvai joined #sbcl
2015-10-16T00:13:28Z DeadTrickster joined #sbcl
2015-10-16T00:13:28Z DGASAU joined #sbcl
2015-10-16T00:13:28Z joshe joined #sbcl
2015-10-16T00:13:28Z ferada joined #sbcl
2015-10-16T00:14:06Z Posterdati joined #sbcl
2015-10-16T00:14:06Z Ober joined #sbcl
2015-10-16T00:14:06Z hzp joined #sbcl
2015-10-16T00:14:10Z les` joined #sbcl
2015-10-16T00:14:10Z dim joined #sbcl
2015-10-16T00:14:27Z tifa quit (Quit: Saliendo)
2015-10-16T00:15:49Z ASau joined #sbcl
2015-10-16T00:15:49Z psy joined #sbcl
2015-10-16T00:15:49Z failproofshark joined #sbcl
2015-10-16T00:15:49Z yauz_2 joined #sbcl
2015-10-16T00:15:49Z p_l joined #sbcl
2015-10-16T00:15:49Z jozip joined #sbcl
2015-10-16T00:15:49Z pootler joined #sbcl
2015-10-16T00:16:22Z psilord joined #sbcl
2015-10-16T00:16:22Z pkhuong joined #sbcl
2015-10-16T00:16:22Z jackc-_ joined #sbcl
2015-10-16T00:16:22Z chris2 joined #sbcl
2015-10-16T00:16:22Z jdz joined #sbcl
2015-10-16T00:16:22Z cades joined #sbcl
2015-10-16T00:16:22Z drmeister joined #sbcl
2015-10-16T00:16:22Z |3b| joined #sbcl
2015-10-16T00:16:22Z jackdaniel joined #sbcl
2015-10-16T00:16:22Z pchrist joined #sbcl
2015-10-16T00:18:25Z frankS2 quit (Changing host)
2015-10-16T00:18:25Z frankS2 joined #sbcl
2015-10-16T00:19:56Z sigjuice joined #sbcl
2015-10-16T00:19:56Z scymtym joined #sbcl
2015-10-16T00:19:56Z hlavaty` joined #sbcl
2015-10-16T00:19:57Z Bike joined #sbcl
2015-10-16T00:19:57Z PuercoPop joined #sbcl
2015-10-16T00:19:57Z minion joined #sbcl
2015-10-16T00:19:57Z mood joined #sbcl
2015-10-16T00:19:57Z nicdev joined #sbcl
2015-10-16T00:19:57Z brucem joined #sbcl
2015-10-16T00:22:37Z sjl joined #sbcl
2015-10-16T00:27:10Z sjl quit (Ping timeout: 240 seconds)
2015-10-16T00:30:40Z heddwch joined #sbcl
2015-10-16T00:35:50Z Quadresce` joined #sbcl
2015-10-16T00:36:41Z Quadresce` is now known as Quadrescence
2015-10-16T00:36:53Z Quadrescence quit (Changing host)
2015-10-16T00:36:53Z Quadrescence joined #sbcl
2015-10-16T00:58:29Z tifa joined #sbcl
2015-10-16T01:14:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-16T01:56:35Z psy quit (Ping timeout: 264 seconds)
2015-10-16T01:58:19Z psy joined #sbcl
2015-10-16T02:01:59Z jlarocco_ joined #sbcl
2015-10-16T02:23:39Z sjl joined #sbcl
2015-10-16T02:28:26Z sjl quit (Ping timeout: 260 seconds)
2015-10-16T02:33:33Z psy quit (Ping timeout: 246 seconds)
2015-10-16T03:01:19Z rszeno joined #sbcl
2015-10-16T03:47:12Z rszeno quit (Quit: Leaving.)
2015-10-16T03:48:12Z jlarocco_ quit (Read error: Connection reset by peer)
2015-10-16T03:50:46Z nicdev` joined #sbcl
2015-10-16T03:51:03Z nicdev quit (Read error: Connection reset by peer)
2015-10-16T03:51:17Z stassats quit (Ping timeout: 244 seconds)
2015-10-16T04:24:25Z sjl joined #sbcl
2015-10-16T04:28:40Z sjl quit (Ping timeout: 250 seconds)
2015-10-16T05:09:35Z sjl joined #sbcl
2015-10-16T05:11:19Z psy_ joined #sbcl
2015-10-16T05:41:38Z sjl quit (Ping timeout: 260 seconds)
2015-10-16T05:50:05Z whiteline joined #sbcl
2015-10-16T05:57:51Z scymtym_ quit (Ping timeout: 255 seconds)
2015-10-16T06:07:39Z gingerale joined #sbcl
2015-10-16T06:14:50Z zacts joined #sbcl
2015-10-16T06:18:15Z psy_ quit (Ping timeout: 246 seconds)
2015-10-16T06:56:50Z ASau quit (Remote host closed the connection)
2015-10-16T06:59:38Z ASau joined #sbcl
2015-10-16T07:03:05Z ASau quit (Remote host closed the connection)
2015-10-16T07:06:26Z ASau joined #sbcl
2015-10-16T07:10:35Z ChanServ has set mode +o Xof
2015-10-16T07:20:16Z ASau quit (Ping timeout: 250 seconds)
2015-10-16T07:35:31Z Ettore joined #sbcl
2015-10-16T07:38:05Z sjl joined #sbcl
2015-10-16T07:42:14Z sjl quit (Ping timeout: 244 seconds)
2015-10-16T07:56:57Z Cymew joined #sbcl
2015-10-16T08:38:46Z sjl joined #sbcl
2015-10-16T08:43:54Z sjl quit (Ping timeout: 255 seconds)
2015-10-16T08:51:42Z psy_ joined #sbcl
2015-10-16T09:00:27Z heddwch quit (Read error: Connection reset by peer)
2015-10-16T09:01:14Z heddwch joined #sbcl
2015-10-16T09:06:20Z angavrilov joined #sbcl
2015-10-16T09:47:45Z rszeno joined #sbcl
2015-10-16T09:50:58Z Shinmera joined #sbcl
2015-10-16T10:06:28Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-16T10:09:16Z sjl joined #sbcl
2015-10-16T10:12:27Z scymtym: the commit "build: leave a *.o file that has all of the startup code." seems to have broken the MacOS build: https://ci.cor-lab.org/job/sbcl-master/1241/featureset=1,label=MAC_OS_mavericks_64bit/consoleFull
2015-10-16T10:23:48Z ym joined #sbcl
2015-10-16T11:12:53Z rszeno quit (Quit: Leaving.)
2015-10-16T11:29:22Z attila_lendvai joined #sbcl
2015-10-16T11:29:22Z attila_lendvai quit (Changing host)
2015-10-16T11:29:22Z attila_lendvai joined #sbcl
2015-10-16T11:40:59Z psy_ quit (Ping timeout: 260 seconds)
2015-10-16T13:26:18Z Cymew quit (Ping timeout: 272 seconds)
2015-10-16T13:39:37Z Cymew joined #sbcl
2015-10-16T14:02:06Z sjl quit (Ping timeout: 240 seconds)
2015-10-16T14:24:17Z Cymew quit (Read error: Connection reset by peer)
2015-10-16T14:36:53Z christoph_e quit (Quit: WeeChat 1.0.1)
2015-10-16T14:40:18Z eudoxia joined #sbcl
2015-10-16T14:43:16Z christoph_debian joined #sbcl
2015-10-16T14:53:50Z karswell quit (Read error: Connection reset by peer)
2015-10-16T14:54:26Z karswell joined #sbcl
2015-10-16T14:54:43Z stassats joined #sbcl
2015-10-16T15:31:23Z foom: wow, that's quite the opaque error.
2015-10-16T15:40:21Z stassats: the 32-bit linux build is broken too
2015-10-16T15:40:31Z stassats: ld: Relocatable linking with relocations from format elf32-i386 (alloc.o) to format elf64-x86-64 (sbcl.o) is not supported
2015-10-16T15:40:58Z foom: Ah, not enough flags passed to the ld call
2015-10-16T15:42:17Z foom: Needs to pass the arch, at least. Maybe some of those other things on macosx too
2015-10-16T15:43:08Z foom: Figuring out the right flag for arch to pass to ld without using the driver is not going to be easy.
2015-10-16T15:43:21Z foom: e.g. on linux it's elf_i386
2015-10-16T15:43:30Z foom: -m elf_i386 i mean
2015-10-16T15:43:35Z stassats: that was my concern when i saw that it calls ld directly
2015-10-16T15:43:41Z stassats: "something's not going to be right"
2015-10-16T15:44:54Z foom: Little GCC sat in bed, cried and cried, her eyes were red.
2015-10-16T15:46:16Z stassats: so, presumably gcc can't just combine .o files together?
2015-10-16T15:47:06Z foom: probably it can
2015-10-16T15:47:22Z foom: I think if you liked with -nostdlib -Wl,-r
2015-10-16T15:47:33Z foom: I expect that would call ld the right way
2015-10-16T15:47:49Z foom: linked, rather.
2015-10-16T15:47:58Z stassats: ok, i don't really want to concern myself with this
2015-10-16T15:48:47Z stassats: described the problem on the ML
2015-10-16T15:52:32Z jackdaniel:  /win 27
2015-10-16T16:00:25Z hlavaty` quit (Remote host closed the connection)
2015-10-16T16:22:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-16T16:50:05Z Ettore quit (Quit: Leaving.)
2015-10-16T16:50:10Z eudoxia_ joined #sbcl
2015-10-16T16:50:57Z eudoxia quit (Read error: Connection reset by peer)
2015-10-16T16:51:18Z karswell quit (Read error: Connection reset by peer)
2015-10-16T16:51:31Z karswell joined #sbcl
2015-10-16T17:05:46Z zacts quit (Ping timeout: 240 seconds)
2015-10-16T17:28:15Z eudoxia_ quit (Quit: Leaving)
2015-10-16T17:45:58Z ASau joined #sbcl
2015-10-16T17:58:31Z hydan` joined #sbcl
2015-10-16T18:09:44Z clique joined #sbcl
2015-10-16T18:24:51Z gingerale quit (Remote host closed the connection)
2015-10-16T18:46:41Z clique quit (Quit: Page closed)
2015-10-16T19:07:51Z abeaumont quit (Ping timeout: 244 seconds)
2015-10-16T19:11:33Z man213 joined #sbcl
2015-10-16T19:18:33Z Ettore joined #sbcl
2015-10-16T19:18:39Z mattprelude joined #sbcl
2015-10-16T19:20:26Z Shinmera joined #sbcl
2015-10-16T19:30:34Z stassats: trying to rethink call and call-named VOPs
2015-10-16T19:30:52Z stassats: we put results into standard locations and they are later moved into the requested locations
2015-10-16T19:31:13Z stassats: trying to load things directly into requested locations
2015-10-16T19:32:29Z stassats: maybe just using CMOV for NILling would be fast enough
2015-10-16T19:52:36Z alchemis7 quit (Ping timeout: 255 seconds)
2015-10-16T20:01:27Z nikki93 joined #sbcl
2015-10-16T20:01:27Z hydan`` joined #sbcl
2015-10-16T20:01:51Z hydan`` is now known as udzinari
2015-10-16T20:02:33Z udzinari is now known as hydan_
2015-10-16T20:03:01Z hydan_ is now known as hydan``
2015-10-16T20:04:35Z sjl joined #sbcl
2015-10-16T20:05:06Z igajsin joined #sbcl
2015-10-16T20:06:23Z scymtym_ joined #sbcl
2015-10-16T20:13:20Z sjl__ joined #sbcl
2015-10-16T20:15:33Z sjl quit (Ping timeout: 255 seconds)
2015-10-16T20:23:35Z nikki93 quit (Remote host closed the connection)
2015-10-16T20:30:25Z Bicyclidine joined #sbcl
2015-10-16T21:07:39Z nyef: On ARM32, which do we expect to be cheaper, constructing a value in two instructions, or doing a cached read from memory to get the value?
2015-10-16T21:08:23Z stassats: i have instructions tables for cortex a57
2015-10-16T21:09:10Z stassats: LDR latency is 4
2015-10-16T21:09:47Z stassats: ALU latency is 1
2015-10-16T21:10:07Z stassats: 2 for shifted ALU
2015-10-16T21:12:16Z stassats: i'd go for constructing
2015-10-16T21:12:23Z stassats: it should also be smaller
2015-10-16T21:20:37Z nyef: Smaller? How so?
2015-10-16T21:21:18Z stassats: could be smaller
2015-10-16T21:21:34Z stassats: if the memory locations is not encodable into a single instruction
2015-10-16T21:30:24Z man213 quit (Quit: Leaving.)
2015-10-16T21:46:41Z jlarocco_ joined #sbcl
2015-10-16T21:52:07Z attila_lendvai joined #sbcl
2015-10-16T22:09:11Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-16T22:18:34Z sjl__ quit (Ping timeout: 260 seconds)
2015-10-16T22:20:44Z ASau quit (Ping timeout: 250 seconds)
2015-10-16T22:31:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-16T22:31:50Z luis quit (Ping timeout: 240 seconds)
2015-10-16T22:32:20Z luis joined #sbcl
2015-10-16T22:42:26Z hydan` quit (Ping timeout: 240 seconds)
2015-10-16T22:54:31Z scymtym_ quit (Ping timeout: 256 seconds)
2015-10-16T22:58:37Z zacts joined #sbcl
2015-10-16T23:09:06Z zacts quit (Ping timeout: 246 seconds)
2015-10-16T23:17:51Z nikki93 joined #sbcl
2015-10-16T23:18:39Z nikki93_ joined #sbcl
2015-10-16T23:21:46Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-16T23:22:11Z nikki93_ quit (Remote host closed the connection)
2015-10-16T23:25:46Z hydan`` quit (Ping timeout: 260 seconds)
2015-10-16T23:29:32Z hydan` joined #sbcl
2015-10-16T23:33:49Z Ettore quit (Quit: Leaving.)
2015-10-16T23:34:36Z phoe_krk_ joined #sbcl
2015-10-16T23:35:07Z phoe_krk_: how can I debug a crashing SBCL?
2015-10-16T23:36:19Z stassats: by employing debugging techniques and thinking hard
2015-10-16T23:36:50Z mattprelude quit (Ping timeout: 272 seconds)
2015-10-16T23:37:11Z alchemis7 joined #sbcl
2015-10-16T23:43:50Z phoe_krk_: http://paste.lisp.org/display/156954 this is my ~/.swank.lisp file. right now, this is working. BUT if you move the bottom two lines to the top, calling (swank::eval-in-lisp '()) causes SBCL to explode
2015-10-16T23:44:34Z phoe_krk_: SBCL 1.2.16 Win64, slime 2.14
2015-10-16T23:45:13Z stassats: well, duh, how can you name two functions with the same name?
2015-10-16T23:45:38Z phoe_krk_: two functions with the same name?
2015-10-16T23:45:49Z stassats: that's what i said indeed
2015-10-16T23:45:52Z stassats: what are you even trying to do?
2015-10-16T23:46:42Z phoe_krk_: I'm silly.
2015-10-16T23:46:45Z phoe_krk_: I just realized.
2015-10-16T23:47:06Z phoe_krk_: That I'm overwriting SWANK::EVAL-IN-EMACS with a custom function that recurses infinitely.
2015-10-16T23:47:09Z phoe_krk_: As a result.
2015-10-16T23:47:11Z phoe_krk_: Therefore, bang.
2015-10-16T23:47:15Z stassats: right
2015-10-16T23:47:20Z stassats: so, what are you trying to achieve here?
2015-10-16T23:47:38Z phoe_krk_: the ability to run elisp code in emacs from inside SBCL REPL
2015-10-16T23:47:41Z phoe_krk_: and I achieved that.
2015-10-16T23:47:42Z phoe_krk_: just.
2015-10-16T23:47:51Z phoe_krk_: by not overwriting a vital function.
2015-10-16T23:48:20Z phoe_krk_: (eval-in-emacs `(progn (switch-to-buffer (get-buffer "*scratch*")))
2015-10-16T23:48:26Z phoe_krk_: This line works, sends me to scratch in emacs.
2015-10-16T23:50:14Z stassats: that should works out the box
2015-10-16T23:50:37Z stassats: why the readtable hack?
2015-10-16T23:50:43Z phoe_krk_: hm?
2015-10-16T23:50:46Z phoe_krk_: let me check
2015-10-16T23:50:56Z stassats: protip: (switch-to-buffer (get-buffer "*scratch*")) == (switch-to-buffer "*scratch*")
2015-10-16T23:51:19Z phoe_krk_: thanks
2015-10-16T23:51:58Z phoe_krk_:  (swank::eval-in-emacs `(progn (switch-to-buffer "*scratch*"))) errors for me
2015-10-16T23:52:05Z phoe_krk_: error: "Non-serializable return value: #<buffer *scratch*>"
2015-10-16T23:52:22Z phoe_krk_: it does what it asks, just fails to return properly, it seems
2015-10-16T23:52:25Z stassats: why not (swank:eval-in-emacs `(null (switch-to-buffer "*scratch*"))) then?
2015-10-16T23:53:13Z stassats: or if you really need the return value: (swank:eval-in-emacs `(prin1-to-string (switch-to-buffer "*scratch*")))
2015-10-16T23:53:33Z stassats: i think there was a function for that in slime
2015-10-16T23:53:39Z phoe_krk_: I don't need a return value
2015-10-16T23:54:00Z phoe_krk_: we were just thinking over at #lisp how to achieve RPC between emacs and SBCL
2015-10-16T23:54:05Z stassats: so, just don't return then
2015-10-16T23:54:21Z stassats: (swank:eval-in-emacs `(and (switch-to-buffer "*scratch*") t)) will not return any garbage
2015-10-16T23:55:16Z phoe_krk_: like
2015-10-16T23:55:18Z phoe_krk_: https://github.com/informatimago/emacs/blob/master/slime-rpc.el
2015-10-16T23:55:20Z phoe_krk_: which is 2010 code
2015-10-16T23:55:39Z phoe_krk_: but seemingly, this is just a pair of commands now
2015-10-16T23:55:54Z stassats: that would be a bad idea in 2010 too
2015-10-16T23:56:54Z phoe_krk_: okay
2015-10-16T23:57:04Z phoe_krk_: swank:eval-in-emacs allows me to evaluate elisp from within SBCL
2015-10-16T23:57:07Z phoe_krk_: what is the other way around?
2015-10-16T23:57:09Z phoe_krk_: slime-?
2015-10-16T23:57:25Z phoe_krk_: I want to evaluate SBCL from within elisp
2015-10-16T23:57:27Z stassats: slime-eval
2015-10-16T23:57:53Z stassats: and slime-eval-async
2015-10-16T23:58:04Z stassats: be aware, it reads everything in the swank-backend package
2015-10-16T23:58:40Z stassats: correction, it's now SWANK-IO-PACKAGE
2015-10-16T23:59:07Z phoe_krk_: CL> swank:eval-in-emacs / ELISP> slime-eval
2015-10-16T23:59:30Z jdz quit (Ping timeout: 252 seconds)
2015-10-16T23:59:48Z stassats: hm, i think i did something for non-readable results, but that doesn't seem to work
2015-10-17T00:01:28Z stassats: i certainly did, but it was broken by someone else, sigh
2015-10-17T00:02:24Z phoe_krk_: I'll work with that; thanks a lot!
2015-10-17T00:02:31Z stassats: maybe it needs another round of fixing
2015-10-17T00:03:46Z phoe_krk_: okay, going home~ thanks, guys
2015-10-17T00:03:56Z phoe_krk_ quit (Remote host closed the connection)
2015-10-17T00:03:57Z stassats: but, i guess error: "Non-serializable return value: #<buffer *scratch*>" is better than the horrible crash it had previously
2015-10-17T00:04:29Z stassats: but i can do a quick fix
2015-10-17T00:04:32Z stassats: a better one
2015-10-17T00:06:27Z jdz joined #sbcl
2015-10-17T00:07:40Z Bicyclidine quit (Quit: dead)
2015-10-17T00:17:59Z stassats: (swank:eval-in-emacs `(get-buffer "*scratch*")) => #<UNREDABLE-RESULT #<buffer *scratch*>>
2015-10-17T00:20:14Z nikki93 joined #sbcl
2015-10-17T00:30:15Z stassats: ok, that's commited
2015-10-17T00:30:29Z stassats: update slime when committing, did it just break my autocompletion?
2015-10-17T00:30:45Z stassats: bloody hell, can't update a thing without breaking
2015-10-17T00:39:09Z nikki93 quit (Remote host closed the connection)
2015-10-17T00:50:44Z nikki93 joined #sbcl
2015-10-17T01:23:25Z phoe_krk joined #sbcl
2015-10-17T02:01:01Z nikki93 quit (Remote host closed the connection)
2015-10-17T02:07:30Z phoe_krk quit (Remote host closed the connection)
2015-10-17T03:35:48Z sjl joined #sbcl
2015-10-17T03:40:00Z DeadTrickster quit (Read error: No route to host)
2015-10-17T03:40:13Z sjl quit (Ping timeout: 250 seconds)
2015-10-17T03:47:15Z zacts joined #sbcl
2015-10-17T03:51:46Z tifa quit (Quit: Saliendo)
2015-10-17T03:53:15Z DeadTrickster joined #sbcl
2015-10-17T04:15:49Z sjl joined #sbcl
2015-10-17T04:20:18Z sjl quit (Ping timeout: 265 seconds)
2015-10-17T04:21:06Z zacts quit (Quit: WeeChat 1.0.1)
2015-10-17T04:55:51Z sjl joined #sbcl
2015-10-17T05:00:47Z sjl quit (Ping timeout: 264 seconds)
2015-10-17T05:31:28Z stassats quit (Ping timeout: 244 seconds)
2015-10-17T05:35:58Z sjl joined #sbcl
2015-10-17T05:37:01Z zacts joined #sbcl
2015-10-17T05:40:20Z sjl quit (Ping timeout: 246 seconds)
2015-10-17T05:56:34Z hydan`` joined #sbcl
2015-10-17T06:12:29Z zacts quit (Quit: WeeChat 1.0.1)
2015-10-17T06:15:58Z sjl joined #sbcl
2015-10-17T06:17:50Z hydan`` quit (Ping timeout: 250 seconds)
2015-10-17T06:20:33Z sjl quit (Ping timeout: 244 seconds)
2015-10-17T06:56:03Z sjl joined #sbcl
2015-10-17T06:56:37Z zacts joined #sbcl
2015-10-17T07:00:18Z sjl quit (Ping timeout: 250 seconds)
2015-10-17T07:11:32Z gingerale joined #sbcl
2015-10-17T08:32:57Z beaumonta joined #sbcl
2015-10-17T08:40:31Z beaumonta is now known as abeaumont
2015-10-17T08:50:47Z nikki93 joined #sbcl
2015-10-17T08:51:03Z scymtym_ joined #sbcl
2015-10-17T08:56:12Z sjl joined #sbcl
2015-10-17T09:01:07Z sjl quit (Ping timeout: 265 seconds)
2015-10-17T09:33:21Z Ettore joined #sbcl
2015-10-17T09:51:21Z nikki93 quit (Remote host closed the connection)
2015-10-17T09:51:25Z mattprelude joined #sbcl
2015-10-17T09:51:58Z nikki93 joined #sbcl
2015-10-17T09:56:10Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-17T10:16:24Z sjl joined #sbcl
2015-10-17T10:21:06Z sjl quit (Ping timeout: 255 seconds)
2015-10-17T10:50:02Z sjl__ joined #sbcl
2015-10-17T11:06:50Z sjl__ quit (Ping timeout: 240 seconds)
2015-10-17T11:09:26Z c6248 joined #sbcl
2015-10-17T11:16:24Z c6248 quit (Ping timeout: 246 seconds)
2015-10-17T11:36:44Z edgar-rft joined #sbcl
2015-10-17T11:48:21Z momo-reina joined #sbcl
2015-10-17T11:49:32Z gargaml joined #sbcl
2015-10-17T12:03:42Z stassats joined #sbcl
2015-10-17T12:17:28Z momo-reina quit (Quit: ERC (IRC client for Emacs 24.5.1))
2015-10-17T12:22:59Z edgar-rft quit (Quit: edgar-rft)
2015-10-17T12:27:58Z rszeno joined #sbcl
2015-10-17T12:29:17Z Shinmera joined #sbcl
2015-10-17T12:34:55Z stassats` joined #sbcl
2015-10-17T12:36:46Z stassats quit (Ping timeout: 260 seconds)
2015-10-17T12:39:17Z karswell quit (Read error: Connection reset by peer)
2015-10-17T12:39:32Z karswell joined #sbcl
2015-10-17T12:56:41Z phoe_krk joined #sbcl
2015-10-17T13:07:30Z edgar-rft joined #sbcl
2015-10-17T13:31:24Z phoe_krk quit (Read error: Connection reset by peer)
2015-10-17T14:03:07Z phoe_krk joined #sbcl
2015-10-17T14:06:35Z heddwch quit (Ping timeout: 265 seconds)
2015-10-17T14:08:40Z fikusz quit (Remote host closed the connection)
2015-10-17T14:09:16Z heddwch joined #sbcl
2015-10-17T14:38:27Z Ettore quit (Quit: Leaving.)
2015-10-17T15:11:10Z specbot quit (Remote host closed the connection)
2015-10-17T15:11:10Z minion quit (Remote host closed the connection)
2015-10-17T15:13:40Z minion joined #sbcl
2015-10-17T15:14:20Z specbot joined #sbcl
2015-10-17T15:18:04Z fikusz joined #sbcl
2015-10-17T15:29:22Z hydan` quit (Ping timeout: 272 seconds)
2015-10-17T16:50:45Z hydan` joined #sbcl
2015-10-17T17:24:47Z Ettore joined #sbcl
2015-10-17T17:32:35Z attila_lendvai joined #sbcl
2015-10-17T17:40:32Z rszeno quit (Quit: Leaving.)
2015-10-17T17:42:05Z mattprelude quit (Ping timeout: 252 seconds)
2015-10-17T18:00:29Z zacts quit (Quit: WeeChat 1.0.1)
2015-10-17T18:36:50Z nikki93 joined #sbcl
2015-10-17T18:40:59Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-17T18:53:06Z zacts joined #sbcl
2015-10-17T18:55:54Z jlarocco_ quit (Quit: Leaving)
2015-10-17T18:57:27Z nikki93 joined #sbcl
2015-10-17T19:15:02Z jlarocco_ joined #sbcl
2015-10-17T19:24:42Z nikki93 quit (Ping timeout: 255 seconds)
2015-10-17T19:26:18Z nyef: Hunh. Finding a platform that SB-POSIX TERMIOS support simply Will Not Work For was easy, found one on my first attempt: NetBSD.
2015-10-17T19:27:28Z stassats`: that reminds me, i need to reenable dlsym on netbsd
2015-10-17T19:27:57Z nikki93 joined #sbcl
2015-10-17T19:30:00Z nyef: ... second attempt, FreeBSD, also obviously won't work.
2015-10-17T19:30:32Z nyef: Clearly, it's not worth digging any further.
2015-10-17T19:32:23Z nikki93_ joined #sbcl
2015-10-17T19:35:11Z nikki93 quit (Ping timeout: 244 seconds)
2015-10-17T19:46:24Z cades quit (Quit: Connection closed for inactivity)
2015-10-17T20:04:50Z hydan` quit (Ping timeout: 246 seconds)
2015-10-17T20:06:42Z Intensity quit (Ping timeout: 244 seconds)
2015-10-17T20:08:04Z |3b| gets a segfault in gc_init_tables during make-target-2 on x8664 windows 7
2015-10-17T20:13:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-17T20:22:16Z nikki93_ quit (Read error: Connection reset by peer)
2015-10-17T20:29:03Z attila_lendvai joined #sbcl
2015-10-17T20:29:03Z attila_lendvai quit (Changing host)
2015-10-17T20:29:03Z attila_lendvai joined #sbcl
2015-10-17T20:30:00Z |3b|: seems to be caused by f61d047c2d83ce6c7437a08223e69970ac561d2b "build: leave a *.o file that has all of the startup code."
2015-10-17T20:33:13Z stassats`: ok, that's one more failure
2015-10-17T20:33:28Z sjl joined #sbcl
2015-10-17T20:33:36Z stassats`: i'll just revert it until things are figured out
2015-10-17T20:33:48Z hydan` joined #sbcl
2015-10-17T20:50:48Z nikki93 joined #sbcl
2015-10-17T20:57:46Z hydan` quit (Ping timeout: 240 seconds)
2015-10-17T21:09:58Z edgar-rft quit (Quit: edgar-rft)
2015-10-17T21:11:17Z sjl quit (Ping timeout: 244 seconds)
2015-10-17T21:35:01Z zacts quit (Quit: WeeChat 1.0.1)
2015-10-17T21:37:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-17T21:42:28Z gargaml quit (Quit: WeeChat 1.1.1)
2015-10-17T21:45:43Z zacts joined #sbcl
2015-10-17T21:46:55Z sjl joined #sbcl
2015-10-17T21:51:03Z nikki93 quit (Read error: No route to host)
2015-10-17T21:52:30Z nikki93 joined #sbcl
2015-10-17T21:54:27Z heddwch quit (Ping timeout: 265 seconds)
2015-10-17T21:56:06Z heddwch joined #sbcl
2015-10-17T22:02:28Z gingerale quit (Remote host closed the connection)
2015-10-17T22:09:11Z heddwch quit (Quit: ZNC - 1.6.0 - http://znc.in)
2015-10-17T22:15:02Z sjl quit (Ping timeout: 246 seconds)
2015-10-17T22:43:43Z sjl joined #sbcl
2015-10-17T23:29:20Z Ettore quit (Quit: Leaving.)
2015-10-17T23:58:10Z sjl quit (Ping timeout: 240 seconds)
2015-10-18T00:02:05Z nikki93 quit (Remote host closed the connection)
2015-10-18T00:24:38Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-18T00:26:44Z nikki93 joined #sbcl
2015-10-18T01:09:28Z stassats` quit (Ping timeout: 244 seconds)
2015-10-18T01:30:31Z rszeno joined #sbcl
2015-10-18T01:46:32Z stassats` joined #sbcl
2015-10-18T01:55:27Z scymtym_ quit (Ping timeout: 244 seconds)
2015-10-18T02:08:50Z rszeno quit (Quit: Leaving.)
2015-10-18T02:21:40Z hydan` joined #sbcl
2015-10-18T02:46:33Z nikki93 quit (Remote host closed the connection)
2015-10-18T02:54:25Z stassats` quit (Remote host closed the connection)
2015-10-18T02:55:40Z nikki93 joined #sbcl
2015-10-18T02:56:10Z nikki93 quit (Remote host closed the connection)
2015-10-18T03:02:16Z nikki93 joined #sbcl
2015-10-18T03:34:17Z nikki93 quit (Remote host closed the connection)
2015-10-18T03:41:40Z nikki93 joined #sbcl
2015-10-18T03:43:24Z nikki93 quit (Remote host closed the connection)
2015-10-18T04:05:03Z stassats joined #sbcl
2015-10-18T04:54:14Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-18T04:55:49Z angavrilov_ joined #sbcl
2015-10-18T04:57:10Z scymtym_ joined #sbcl
2015-10-18T04:57:36Z Bike_ joined #sbcl
2015-10-18T04:58:17Z Bike quit (Disconnected by services)
2015-10-18T04:58:20Z Bike_ is now known as Bike
2015-10-18T05:02:51Z angavrilov quit (*.net *.split)
2015-10-18T05:02:51Z sigjuice quit (*.net *.split)
2015-10-18T05:02:51Z scymtym quit (*.net *.split)
2015-10-18T05:02:52Z PuercoPop quit (*.net *.split)
2015-10-18T05:02:52Z mood quit (*.net *.split)
2015-10-18T05:02:53Z brucem quit (*.net *.split)
2015-10-18T05:02:55Z mood_ joined #sbcl
2015-10-18T05:03:53Z sigjuice joined #sbcl
2015-10-18T05:04:57Z PuercoPop joined #sbcl
2015-10-18T05:07:23Z zacts quit (Ping timeout: 260 seconds)
2015-10-18T05:08:03Z brucem joined #sbcl
2015-10-18T05:10:13Z zacts joined #sbcl
2015-10-18T06:04:04Z gingerale joined #sbcl
2015-10-18T06:54:46Z edgar-rft joined #sbcl
2015-10-18T07:08:42Z stassats quit (Ping timeout: 260 seconds)
2015-10-18T08:30:50Z nikki93 joined #sbcl
2015-10-18T08:37:45Z angavrilov_ quit (Remote host closed the connection)
2015-10-18T08:44:41Z Shinmera joined #sbcl
2015-10-18T08:54:23Z angavrilov joined #sbcl
2015-10-18T09:03:36Z nikki93 quit (Remote host closed the connection)
2015-10-18T09:17:48Z scymtym joined #sbcl
2015-10-18T09:51:31Z sjl joined #sbcl
2015-10-18T09:57:30Z Ettore joined #sbcl
2015-10-18T10:00:46Z sjl quit (Ping timeout: 240 seconds)
2015-10-18T10:35:39Z dougk_ quit (Ping timeout: 240 seconds)
2015-10-18T10:37:40Z gucore joined #sbcl
2015-10-18T10:49:03Z dougk_ joined #sbcl
2015-10-18T11:10:59Z igajsin quit (Remote host closed the connection)
2015-10-18T11:42:06Z ym quit (Ping timeout: 255 seconds)
2015-10-18T11:42:09Z scymtym quit (Read error: Connection reset by peer)
2015-10-18T11:54:09Z scymtym joined #sbcl
2015-10-18T12:05:54Z Cymew joined #sbcl
2015-10-18T12:10:48Z Cymew quit (Ping timeout: 244 seconds)
2015-10-18T12:31:37Z rszeno joined #sbcl
2015-10-18T12:51:05Z mood_ is now known as mood
2015-10-18T12:58:19Z _iwc joined #sbcl
2015-10-18T13:08:39Z stassats joined #sbcl
2015-10-18T13:10:42Z _iwc quit (Killed (Sigyn (Spam is off topic on freenode.)))
2015-10-18T13:12:58Z rszeno quit (Quit: Leaving.)
2015-10-18T13:57:18Z stassats: gcc compile farm just got an armv8 machine
2015-10-18T13:57:24Z stassats: or four of them
2015-10-18T13:59:06Z stassats: APM X-Gene Mustang board, 8 cores 32GB RAM
2015-10-18T14:00:45Z p_l: nice
2015-10-18T14:17:03Z Cymew joined #sbcl
2015-10-18T14:24:18Z ym joined #sbcl
2015-10-18T14:45:30Z Cymew quit (Read error: Connection reset by peer)
2015-10-18T15:27:12Z nyef: ... We're not in code-freeze yet, are we?
2015-10-18T15:27:28Z brucem quit (Changing host)
2015-10-18T15:27:28Z brucem joined #sbcl
2015-10-18T15:28:46Z stassats: don't think so
2015-10-18T15:29:15Z nyef: Okay, in that case I should commit my two-line initial fix for sb-posix termios.
2015-10-18T15:29:46Z stassats: and i'll see what to do with that sbcl.o thing
2015-10-18T15:29:54Z stassats: either i can make it work or i'll revert it
2015-10-18T15:30:11Z stassats: but virtualbox crashes my system, fun, fun
2015-10-18T15:39:34Z stassats: an update fixed that, can now test on windows
2015-10-18T15:41:25Z stassats: ok, the APM X-Gene Mustang thingy is slower than my phone
2015-10-18T15:41:29Z stassats: at building sbcl
2015-10-18T15:49:19Z heddwch joined #sbcl
2015-10-18T15:51:05Z stassats: reverting it is
2015-10-18T16:21:46Z stassats: make-target-contrib is broken on netbsd, great
2015-10-18T16:29:46Z stassats: huh what, unset EXTRA_CFLAGS exits the script
2015-10-18T16:32:54Z stassats: ok, i need to update netbsd
2015-10-18T16:38:39Z foom quit (Ping timeout: 240 seconds)
2015-10-18T16:41:15Z stassats: they fixed that in NetBSD 7.0
2015-10-18T16:47:09Z nyef sighs.
2015-10-18T16:47:21Z nyef: I'm going to have to run a NetBSD VM on all of my Linux build boxes, aren't I?
2015-10-18T16:48:04Z nyef: Also a FreeBSD VM on those that are supported by FreeBSD...
2015-10-18T16:48:36Z joshe: I started to set up a buildbot cluster
2015-10-18T16:48:50Z joshe: but I never had any luck with netbsd
2015-10-18T16:52:01Z foom joined #sbcl
2015-10-18T16:54:12Z stassats: reenabled dladdr on netbsd, that's enough testing on marginal platforms for today
2015-10-18T16:54:37Z nyef: Aww, you don't want to make termios work on NetBSD?
2015-10-18T16:54:55Z stassats: i don't need termios anywhere
2015-10-18T16:55:58Z nyef: Fair enough.
2015-10-18T16:56:21Z nyef: And it turns out that I need more than just termios, so I'm also using some Linux IOCTLs as well.
2015-10-18T16:56:58Z stassats: 1.2.16.181, that's a large number, but can't beat the arm32 port commit numbers
2015-10-18T16:58:30Z nyef: What did we get to with the arm32 port? I forget...
2015-10-18T16:58:44Z stassats: something ridiculous
2015-10-18T16:58:47Z stassats: like 500
2015-10-18T16:58:55Z nyef: Heh. Neat.
2015-10-18T17:00:31Z stassats: 577
2015-10-18T17:01:22Z nyef: Yeah, that's going to be a tough number to beat.
2015-10-18T17:01:31Z ralt joined #sbcl
2015-10-18T17:01:32Z ralt: hi
2015-10-18T17:01:50Z ralt: is there some function to have (usleep)? sbcl-specific is fine
2015-10-18T17:01:57Z ralt: maybe in sb-ext or something like that
2015-10-18T17:02:15Z nyef: CL:SLEEP can take a floating-point argument, can't it?
2015-10-18T17:02:35Z ralt: oh, didn't know that
2015-10-18T17:02:41Z ralt: I just assumed it was integers only
2015-10-18T17:02:55Z nyef: clhs sleep
2015-10-18T17:02:55Z specbot: http://www.lispworks.com/reference/HyperSpec/Body/f_sleep.htm
2015-10-18T17:03:04Z ralt: real
2015-10-18T17:03:15Z ralt: that includes floating-point numbers, I guess
2015-10-18T17:03:16Z stassats: (sleep 1/1000)
2015-10-18T17:03:18Z ralt: thanks!
2015-10-18T17:04:08Z stassats: nyef: a couple from scratch port at the same time
2015-10-18T17:04:28Z stassats: itanium and m68k
2015-10-18T17:05:03Z nyef: Heh. Shall I get a Quardra and install Linux on it, then?
2015-10-18T17:05:24Z nyef: Err... Quadra.
2015-10-18T17:05:43Z nyef: Ooh. Even better idea, A/UX!
2015-10-18T17:06:28Z stassats: something with 68060 would be more useful
2015-10-18T17:06:42Z joshe: there's always a vax port
2015-10-18T17:07:12Z stassats: what about going totally crazy and porting to i286?
2015-10-18T17:08:28Z edgar-rft quit (Quit: edgar-rft)
2015-10-18T17:09:24Z nyef: What has an '060?
2015-10-18T17:09:38Z nyef: i286 _would_ be crazy. 16-bit, remember?
2015-10-18T17:09:53Z stassats: that's the point
2015-10-18T17:10:13Z stassats: some Amiga 4000T
2015-10-18T17:11:37Z stassats: DOS-i286
2015-10-18T17:13:45Z stassats: there's a Freescale ColdFire thingy
2015-10-18T17:14:33Z nyef: Hrm... Do we have any other reasonable options?
2015-10-18T17:15:42Z stassats: AVR32?
2015-10-18T17:15:45Z stassats: RISC-V?
2015-10-18T17:16:46Z clique joined #sbcl
2015-10-18T17:17:29Z joshe: super h?
2015-10-18T17:17:55Z Quadrescence joined #sbcl
2015-10-18T17:23:20Z stassats: openbsd is 20 years old, incidentally
2015-10-18T17:23:59Z nyef: Wow. In another year, it can drink booze in the USA.
2015-10-18T17:24:09Z joshe: yes
2015-10-18T17:24:28Z joshe: well, it's 20 years since the first commit in the repo
2015-10-18T17:28:45Z stassats: nyef: for now the reasonable options are bringing up 64-bit support for all other arches
2015-10-18T17:29:09Z Xof: stassats: is your netbsd version test the wrong way round?
2015-10-18T17:29:22Z stassats: Xof: what do you mean?
2015-10-18T17:29:35Z ralt quit (Ping timeout: 264 seconds)
2015-10-18T17:29:43Z phoe_krk quit (Remote host closed the connection)
2015-10-18T17:29:53Z Xof: I mean you seem to return 1 on high versions of netbsd
2015-10-18T17:30:10Z Xof: (or I am misreading the patchmail)
2015-10-18T17:30:58Z stassats: no point in reading, let's just see if it actually pushes the feature
2015-10-18T17:31:52Z stassats: :os-provides-dladdr is there
2015-10-18T17:32:34Z Xof: ...
2015-10-18T17:32:57Z Xof: then I don't understand
2015-10-18T17:33:01Z Xof: #if defined(__NetBSD_Version__) && __NetBSD_Version__ >= 700000001
2015-10-18T17:33:05Z nyef: stassats: Dibs on MIPS, Alpha, and HPPA!
2015-10-18T17:33:08Z Xof:   return 1;
2015-10-18T17:33:31Z stassats: ok, i don't understand now either
2015-10-18T17:36:01Z stassats: it does return 104
2015-10-18T17:36:39Z Xof: what's __NetBSD_VERSION__?
2015-10-18T17:36:46Z Xof: um
2015-10-18T17:36:48Z Xof: Version
2015-10-18T17:36:52Z Xof: however you capitalize it
2015-10-18T17:36:58Z stassats: maybe i added an extra 0
2015-10-18T17:37:14Z Xof: 399000800
2015-10-18T17:37:17Z Xof: 700000001
2015-10-18T17:37:24Z Xof: hm, no, length checks out :)
2015-10-18T17:37:42Z scymtym: do we have a preferred encoding for source files? or put differently, would changing the encoding of a source file from latin-1 to utf-8 break the non-sb-unicode build?
2015-10-18T17:38:12Z stassats: it wouldn't be a great idea
2015-10-18T17:39:12Z scymtym: i had the impression that we had utf-8 encoded files already. i'm going to have a closer look
2015-10-18T17:39:55Z stassats: Xof: i probably need to include /sys/param.h for __NetBSD_Version__ to appear
2015-10-18T17:40:01Z Xof: aha
2015-10-18T17:40:17Z scymtym is working on eliminating the expand dependency
2015-10-18T17:40:39Z stassats: now, how standard is sys/param.h?
2015-10-18T17:40:59Z scymtym quit (Quit: scymtym)
2015-10-18T17:41:16Z scymtym joined #sbcl
2015-10-18T17:43:27Z attila_lendvai joined #sbcl
2015-10-18T17:43:27Z attila_lendvai quit (Changing host)
2015-10-18T17:43:27Z attila_lendvai joined #sbcl
2015-10-18T17:48:14Z joshe: not very, afaik
2015-10-18T17:49:17Z joshe: the existence of the header itself is widespread but I don't think the contents are very consistent
2015-10-18T17:49:27Z stassats: Xof: fixed, good catch
2015-10-18T17:49:43Z stassats: that'll boost the commit number
2015-10-18T18:01:42Z clique left #sbcl
2015-10-18T18:12:07Z phoe_krk joined #sbcl
2015-10-18T18:18:31Z scymtym: we have 7 utf-8 files and one latin-1 file in tests/. the only non-ascii characters in that file are in a comment stating the name of a person who reported a bug. are you sure changing that file to utf-8 would cause problems?
2015-10-18T18:19:20Z nyef: scymtym: Why don't you do the conversion, and then a non-unicode build-and-test, and let us know how it goes?
2015-10-18T18:21:14Z scymtym: nyef: will do
2015-10-18T18:35:12Z jlarocco_ quit (Ping timeout: 244 seconds)
2015-10-18T18:42:10Z scymtym: no apparent problems during build or test on my system
2015-10-18T18:47:24Z jlarocco_ joined #sbcl
2015-10-18T19:13:45Z Fare joined #sbcl
2015-10-18T19:14:00Z Fare: how do I compile sbcl on windows? mingw ?
2015-10-18T19:14:07Z stassats: yes
2015-10-18T19:14:19Z stassats: Fare: did you see the ld issues?
2015-10-18T19:14:19Z Fare: what about mac? just regular apple dev kit?
2015-10-18T19:14:25Z stassats: i have no idea how to solve that one
2015-10-18T19:14:27Z Fare: I saw that you saw them
2015-10-18T19:14:50Z stassats: the windows crash is just weird, it compiles alright, but then goes bonkers running
2015-10-18T19:14:55Z Fare: I have to somehow unearth an antique XP install on a laptop or otherwise find myself a Windows VM that I could try things on
2015-10-18T19:15:20Z Fare: the refactoring ought to have been trivial :-(
2015-10-18T19:15:26Z Fare: especially with GNU tools.
2015-10-18T19:15:43Z Fare: what is the crash on a Mac? It worked on FreeBSD.
2015-10-18T19:16:06Z stassats: no crash, can't link the .o into sbcl
2015-10-18T19:17:04Z Fare: ?
2015-10-18T19:17:06Z Fare: :-(
2015-10-18T19:19:06Z Fare: so the ld -r step succeeds but the .o it creates is unusable?
2015-10-18T19:19:30Z stassats: yes
2015-10-18T19:19:54Z stassats: see my message in sbcl-commits
2015-10-18T19:21:18Z Fare: even weirder as ld -r seems to work for cmucl on macosx
2015-10-18T19:22:18Z Fare: (although... I'm not the one who tested that, so I don't know what exactly was tested.)
2015-10-18T19:23:19Z Fare: all I see is the commit a61d893a7fbe80905e9e056cc37c47d476074c36
2015-10-18T19:23:25Z Fare: is there more info?
2015-10-18T19:24:41Z stassats: as i said elsewhere, it doesn't work on os x, windows, some configurations of linux
2015-10-18T19:32:18Z Fare: some configurations of linux???
2015-10-18T19:40:14Z jlarocco_ quit (Ping timeout: 246 seconds)
2015-10-18T20:21:39Z Fare: with the gnu toolchain?
2015-10-18T20:22:04Z stassats: i wrote everything on the mailing list, on the launchpad ticket
2015-10-18T20:23:21Z nikki93 joined #sbcl
2015-10-18T20:25:50Z hydan` quit (Ping timeout: 250 seconds)
2015-10-18T20:27:59Z jlarocco_ joined #sbcl
2015-10-18T20:32:26Z gucore quit (Ping timeout: 240 seconds)
2015-10-18T20:40:01Z hydan` joined #sbcl
2015-10-18T20:40:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-18T20:44:42Z heddwch quit (Ping timeout: 265 seconds)
2015-10-18T21:11:46Z nikki93 quit (Remote host closed the connection)
2015-10-18T21:16:20Z flavioc_ joined #sbcl
2015-10-18T21:16:37Z flavioc_ quit (Client Quit)
2015-10-18T21:17:17Z Fare: thanks
2015-10-18T21:29:35Z nikki93 joined #sbcl
2015-10-18T21:33:22Z prxq joined #sbcl
2015-10-18T21:38:53Z hydan` quit (Ping timeout: 246 seconds)
2015-10-18T21:53:34Z heddwch joined #sbcl
2015-10-18T21:57:08Z sjl joined #sbcl
2015-10-18T22:04:01Z nikki93 quit (Remote host closed the connection)
2015-10-18T22:15:30Z gingerale quit (Remote host closed the connection)
2015-10-18T22:17:11Z prxq quit (Remote host closed the connection)
2015-10-18T22:40:45Z nikki93 joined #sbcl
2015-10-18T22:41:57Z Fare: ouch, there is no obvious equivalent to -m32 for GNU ld
2015-10-18T22:43:49Z nikki93_ joined #sbcl
2015-10-18T22:46:46Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-18T22:57:52Z sjl__ joined #sbcl
2015-10-18T22:59:59Z sjl quit (Ping timeout: 256 seconds)
2015-10-18T23:05:08Z nikki93 joined #sbcl
2015-10-18T23:07:11Z Ettore quit (Quit: Leaving.)
2015-10-18T23:08:00Z Fare: linux i386 seems to be about adding LDFLAGS = -m elf32-i386
2015-10-18T23:08:05Z Fare: or something
2015-10-18T23:08:35Z nikki93_ quit (Ping timeout: 264 seconds)
2015-10-18T23:08:45Z Fare: for windows... I'm not sure
2015-10-18T23:14:41Z nikki93 quit
2015-10-18T23:14:54Z Fare: I assume -m pe-i386
2015-10-18T23:34:22Z nyef: lp 1044465
2015-10-18T23:34:22Z specbot: https://bugs.launchpad.net/bugs/1044465
2015-10-18T23:34:54Z sjl__ quit (Ping timeout: 260 seconds)
2015-10-18T23:36:13Z nyef: So, did we ever figure out the whole "conditional DX" thing?
2015-10-18T23:39:27Z nyef: Or did we just leave it at "well, we fixed the bug, and we're too tired right now to figure out how to make the optimization bit work"?
2015-10-18T23:41:01Z stassats: tired? more like trying to preserve sanity by not wandering through the dungeons
2015-10-18T23:41:31Z nyef: Heh. Yeah, that too.
2015-10-18T23:42:05Z nyef: Too much chance of stumbling across something from the Lovecraft Mythos down there.
2015-10-18T23:42:25Z stassats: it's easy to make it worse without a complete understanding
2015-10-18T23:43:05Z nyef: I had a hard enough time getting the STACK analysis right, after all.
2015-10-18T23:43:36Z stassats: though, i'm sick of dealing with the backend, maybe something frontendish would be interesting
2015-10-18T23:44:08Z nyef: If you want frontendish, how about lp#310185 ?
2015-10-18T23:44:33Z stassats: meh
2015-10-18T23:44:59Z stassats: it'll make me want to rewrite external formats even more
2015-10-18T23:45:35Z stassats: i had an angle on making external formats for i/o and for ffi go faster
2015-10-18T23:45:43Z stassats: but there's a lot of stuff involved
2015-10-18T23:46:22Z nyef: If you do, please remember that I actually use the :macintosh external-format... And we may possibly want to port to a system that uses EBCDIC as a default external format at some point.
2015-10-18T23:51:11Z stassats: there's no reason for it to be restricted to ascii
2015-10-18T23:51:47Z stassats: and maybe doing clrfs first will enhance my understanding of the subject
2015-10-18T23:52:02Z stassats: but, that'll have to wait until i'm out of other things to work on
2015-10-18T23:53:03Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-19T00:07:58Z nyef sighs.
2015-10-19T00:08:13Z nyef: SPARC GENCGC isn't handling P-A correctly for runtime allocations.
2015-10-19T00:09:30Z nyef: And my notes say that PPC P-A is wrong for non-threaded GENCGC, but I have no idea if that is still the case.
2015-10-19T00:16:45Z phoe_krk quit (Remote host closed the connection)
2015-10-19T00:35:37Z stassats: nyef: i don't there are any runtime allocations
2015-10-19T00:35:51Z stassats: they are utterly broken everywhere, true
2015-10-19T00:47:19Z scymtym quit (Ping timeout: 240 seconds)
2015-10-19T00:51:01Z nyef: Yes, utterly broken everywhere, but at least the P-A handling should be made correct.
2015-10-19T00:51:37Z stassats: things in alloc.c just need to be taken out
2015-10-19T00:52:04Z nyef: IIRC, it's not just allocations that require P-A from the runtime.
2015-10-19T00:53:05Z stassats: i think now only alloc_number is used, and that should only produce fixnums
2015-10-19T00:53:22Z stassats: albeit    if (-0x20000000 < n && n < 0x20000000) return make_fixnum(n); is looking mightily suspect
2015-10-19T00:53:52Z stassats: nyef: is pa broken anywhere else?
2015-10-19T00:54:23Z nyef: Oh, probably.
2015-10-19T00:55:10Z stassats: in what fashion?
2015-10-19T00:55:27Z nyef: On the other hand, PA really only gets used by the runtime on GENCGC, so we're not too fussed about the non-gencgc platforms at this point.
2015-10-19T00:56:07Z nyef: Compare SPARC *_pseudo_atomic_* (sparc-arch.c) to the PPC versions.
2015-10-19T00:56:55Z nyef: Umm... I don't think that ARM would be affected, given that it partakes of the x86oid nature there.
2015-10-19T00:57:16Z nyef: And I don't believe that any other arch has gencgc?
2015-10-19T01:01:54Z stassats: so, where else is pa used in runtime together with ffca?
2015-10-19T01:02:02Z stassats: call_into_[c|lisp]?
2015-10-19T01:04:32Z nyef: Hrm.
2015-10-19T01:04:52Z nyef: I thought that there were more cases, but either I'm misremembering or they've been excised in the meantime.
2015-10-19T01:05:22Z stassats: i see if i can just axe pa_alloc
2015-10-19T01:10:31Z stassats: it's used for allocating some numbers for some signal handlers on sparc, hppa, and mips
2015-10-19T01:11:18Z stassats: sparc seems to be doing tagged adds, which might steer into non-fixnums
2015-10-19T01:11:38Z stassats: alloc_number is utterly broken here, ok, but sparc is pretty broken overall
2015-10-19T01:12:15Z stassats: or rather, when combined with solaris
2015-10-19T01:13:34Z stassats: if i had access to sparc i would have probably just removed tagged instructions
2015-10-19T01:13:44Z stassats: i bet they are microcoded and slow anyhow
2015-10-19T01:17:05Z nyef: Okay, so I should probably just schedule SPARC for a good going-over at some point?
2015-10-19T01:17:38Z stassats: and what's being done in sigfpe on other arches, no idea
2015-10-19T01:18:30Z stassats: does mips have overflow traps too?
2015-10-19T01:21:40Z stassats: ok, didn't know that
2015-10-19T01:22:00Z stassats: how can fixnums overflow though?
2015-10-19T01:22:20Z stassats: under addition
2015-10-19T01:26:32Z stassats: oh, if it doesn't untage them of course
2015-10-19T01:27:27Z stassats: ok, can't remove alloc_number just yet, but it's completely gc unsafe
2015-10-19T01:28:51Z stassats: even ignoring alloc_number, just look at this sequence https://github.com/sbcl/sbcl/blob/master/src/runtime/mips-arch.c#L463
2015-10-19T01:29:37Z nyef: MIPS, IIRC, is cheneygc-only.
2015-10-19T01:30:08Z stassats: an interrupt hits and allocates something?
2015-10-19T01:32:06Z nyef: I sortof see what it's doing, but...
2015-10-19T01:32:14Z nyef grabs his MIPS reference.
2015-10-19T01:32:31Z stassats: mips has ADD and ADDU, one is trapping, another isn't
2015-10-19T01:33:39Z nyef: Yes, that's exactly what's going on.
2015-10-19T01:34:05Z nyef: A FIXNUM-FIXNUM operation overflowing, and the trap handler doing the bignum allocation.
2015-10-19T01:34:14Z stassats: it's #+niled
2015-10-19T01:35:20Z stassats: except for %%nip-values
2015-10-19T01:35:34Z stassats: and signed-byte-32-len
2015-10-19T01:36:10Z stassats: and some amount of sub usage too
2015-10-19T01:36:27Z stassats: i can imagine it can go pretty bad when it overflows, which is pretty unlikely, but still
2015-10-19T01:36:47Z nyef: How about this? Leave it for now. I'm planning to spend some time going over the MIPS code soon, possibly this week.
2015-10-19T01:37:16Z stassats: good thing, this overflow thing is not used
2015-10-19T01:37:33Z stassats: so, it just needs to be comprehensively removed
2015-10-19T01:38:13Z stassats: allocating things from the runtime is a pretty bad proposition
2015-10-19T01:39:06Z Fare quit (Ping timeout: 255 seconds)
2015-10-19T01:42:18Z stassats: even if it were done right, trapping for bignum allocation? come on
2015-10-19T01:42:47Z stassats: this kills any semblance of performance for bignums
2015-10-19T01:43:44Z stassats: tadd is not used on sparc either, good
2015-10-19T01:50:54Z stassats: pa-risc has an interesting conditional execution of the subsequent instruction
2015-10-19T01:51:52Z Fare joined #sbcl
2015-10-19T02:02:29Z stassats: hppa doesn't use overflows either, so it seems
2015-10-19T02:04:47Z stassats: i haven't checked src/assembly
2015-10-19T02:05:46Z nyef: ... While I'm not planning on tackling HPPA this month, I might at least check to see if my HPPA hardware still works.
2015-10-19T02:05:54Z stassats: looks like it did use in the past, but not now
2015-10-19T02:06:16Z stassats: ok, so, all those sig[fpe|emt|bus]_handlers can go
2015-10-19T02:06:29Z stassats: that'll no longer hold alloc_number, and pa_alloc
2015-10-19T02:07:11Z stassats: and will no longer confuse people
2015-10-19T02:13:29Z stassats: i could do some serious clean up if i had access to all arches
2015-10-19T02:14:59Z sjl joined #sbcl
2015-10-19T02:15:06Z nyef: I currently do not have access to: PPC, ARM64. I _may_ not have working hardware for: Alpha.
2015-10-19T02:15:31Z stassats: the gcc compiler farm now has both ppc and arm64 (as of today)
2015-10-19T02:15:48Z nyef: Everything else, I can cover out of local resources.
2015-10-19T02:15:52Z stassats: my phone is still faster than both
2015-10-19T02:16:45Z stassats: i also have qemulated mips
2015-10-19T02:18:02Z nyef: I have real hardware.
2015-10-19T02:18:55Z nyef: So, the main thing that I need in order to be able to give you some level of access is a VPN.
2015-10-19T02:19:17Z stassats: i like real and _local_ hardware too
2015-10-19T02:19:21Z sjl quit (Ping timeout: 252 seconds)
2015-10-19T02:20:02Z nyef: Yes, well, so do I, but I can't help you with that.
2015-10-19T02:20:38Z stassats: i can help me with that: not care about things i'm not going to use
2015-10-19T02:20:58Z stassats: but i don't like when the VOPs split apart and eventually rot
2015-10-19T02:24:35Z stassats: the compiler harm once had everything available
2015-10-19T02:24:46Z stassats: but now it's just x86, arm and ppc
2015-10-19T02:25:08Z stassats: probably tells what actually survives in the real world
2015-10-19T02:45:06Z stassats quit (Ping timeout: 268 seconds)
2015-10-19T04:15:43Z sjl joined #sbcl
2015-10-19T04:20:22Z sjl quit (Ping timeout: 252 seconds)
2015-10-19T04:56:19Z luis quit (Ping timeout: 260 seconds)
2015-10-19T04:56:49Z luis joined #sbcl
2015-10-19T05:03:46Z luis quit (Ping timeout: 260 seconds)
2015-10-19T05:06:50Z luis joined #sbcl
2015-10-19T05:08:14Z nikki93 joined #sbcl
2015-10-19T05:16:31Z sjl joined #sbcl
2015-10-19T05:21:46Z sjl quit (Ping timeout: 272 seconds)
2015-10-19T05:40:08Z Fare quit (Ping timeout: 246 seconds)
2015-10-19T05:42:12Z nyef quit (Quit: G'night all.)
2015-10-19T06:20:39Z nikki93 quit (Remote host closed the connection)
2015-10-19T06:21:40Z nikki93 joined #sbcl
2015-10-19T06:37:25Z kanru quit (Read error: Connection reset by peer)
2015-10-19T06:38:19Z kanru joined #sbcl
2015-10-19T06:44:02Z nikki93 quit (Ping timeout: 265 seconds)
2015-10-19T06:45:32Z salva quit (Read error: Connection reset by peer)
2015-10-19T06:47:08Z gucore joined #sbcl
2015-10-19T06:47:14Z sjl joined #sbcl
2015-10-19T06:48:25Z nikki93 joined #sbcl
2015-10-19T06:51:27Z sjl quit (Ping timeout: 244 seconds)
2015-10-19T06:56:36Z heddwch quit (Ping timeout: 265 seconds)
2015-10-19T06:56:45Z gingerale joined #sbcl
2015-10-19T07:03:51Z Ettore joined #sbcl
2015-10-19T07:06:50Z nikki93_ joined #sbcl
2015-10-19T07:09:39Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-19T07:13:15Z nikki93 joined #sbcl
2015-10-19T07:16:09Z nikki93_ quit (Ping timeout: 255 seconds)
2015-10-19T07:17:09Z nikki93_ joined #sbcl
2015-10-19T07:18:19Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-19T07:19:16Z nikki93_ quit (Remote host closed the connection)
2015-10-19T07:19:18Z salva joined #sbcl
2015-10-19T07:40:06Z Shinmera joined #sbcl
2015-10-19T07:43:14Z nikki93 joined #sbcl
2015-10-19T07:47:32Z nikki93 quit (Ping timeout: 246 seconds)
2015-10-19T07:48:02Z sjl joined #sbcl
2015-10-19T07:52:36Z sjl quit (Ping timeout: 255 seconds)
2015-10-19T08:39:02Z Cymew joined #sbcl
2015-10-19T08:43:11Z Ettore1 joined #sbcl
2015-10-19T08:43:37Z Ettore quit (Ping timeout: 252 seconds)
2015-10-19T08:48:53Z sjl joined #sbcl
2015-10-19T08:53:23Z sjl quit (Ping timeout: 244 seconds)
2015-10-19T09:37:01Z scymtym joined #sbcl
2015-10-19T09:54:11Z attila_lendvai joined #sbcl
2015-10-19T09:54:11Z attila_lendvai quit (Changing host)
2015-10-19T09:54:11Z attila_lendvai joined #sbcl
2015-10-19T10:08:02Z Ettore1 quit (Ping timeout: 272 seconds)
2015-10-19T10:10:46Z sjl joined #sbcl
2015-10-19T10:31:17Z DGASAU quit (Read error: Connection reset by peer)
2015-10-19T10:33:09Z DGASAU joined #sbcl
2015-10-19T10:33:26Z Ettore joined #sbcl
2015-10-19T10:33:45Z DGASAU quit (Remote host closed the connection)
2015-10-19T10:34:06Z DGASAU joined #sbcl
2015-10-19T10:55:34Z Shinmera: Is there a way to READ-SEQUENCE from a stream, but only as much as is currently available? As in, simply return if no further data is available, perhaps with a timeout.
2015-10-19T11:16:34Z dougk2 joined #sbcl
2015-10-19T11:17:17Z dougk2: anyone around?
2015-10-19T11:48:18Z Fare joined #sbcl
2015-10-19T11:57:43Z sjl quit (Ping timeout: 260 seconds)
2015-10-19T12:15:24Z sjl joined #sbcl
2015-10-19T12:17:32Z scymtym: dougk2: more or less
2015-10-19T12:18:07Z dougk2: hi!
2015-10-19T12:19:08Z dougk2: I just pushed the change which is the "only" important one to review for correctness regarding the new interpreter, in the sense that it's the only one that could cause regressions if it's wrong.
2015-10-19T12:19:40Z dougk2: the rest of the changes will at worst be wrong in code that is only in the new interpreter
2015-10-19T12:21:14Z Fare quit (Ping timeout: 246 seconds)
2015-10-19T12:22:26Z scymtym: i usually try to at least skim through all commits. at which one should we take a closer look?
2015-10-19T12:26:20Z Fare joined #sbcl
2015-10-19T12:32:12Z dougk2: "update compiler and runtime ..."
2015-10-19T12:45:44Z Fare quit (Ping timeout: 246 seconds)
2015-10-19T12:46:02Z scymtym: thanks. i will try to read it carefully
2015-10-19T12:47:30Z stassats joined #sbcl
2015-10-19T12:50:26Z zacts quit (Read error: Connection reset by peer)
2015-10-19T12:53:37Z zacts joined #sbcl
2015-10-19T13:02:00Z eudoxia joined #sbcl
2015-10-19T13:24:07Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-19T13:25:18Z rpg joined #sbcl
2015-10-19T13:25:18Z rpg quit (Remote host closed the connection)
2015-10-19T13:41:21Z sjl quit (Ping timeout: 255 seconds)
2015-10-19T14:07:59Z hydan` joined #sbcl
2015-10-19T14:12:41Z Fare joined #sbcl
2015-10-19T14:17:36Z pkhuong: will do.
2015-10-19T14:40:08Z Fare: if I want to expose the cflags and ldflags into Lisp, what's the correct way?
2015-10-19T14:40:23Z stassats: what for?
2015-10-19T14:40:40Z Fare: have them around when I link a new runtime with extensions
2015-10-19T14:41:12Z stassats: that doesn't happen from lisp
2015-10-19T14:41:13Z sjl joined #sbcl
2015-10-19T14:41:17Z stassats: save a makefile or something
2015-10-19T14:41:20Z Fare: -lm -lz etc. flags would also be nice
2015-10-19T14:41:32Z Fare: ok
2015-10-19T14:41:42Z stassats: pkg_config?
2015-10-19T14:41:48Z Fare: something like that
2015-10-19T14:41:52Z Fare: yes
2015-10-19T14:42:07Z Fare: I'll try to follow the format
2015-10-19T14:42:28Z stassats: a makefile to include could be simpler for now
2015-10-19T14:42:35Z Fare: still gotta find myself some Windows and MacOS and Linux 32-on-64 dev environments.
2015-10-19T14:43:07Z Fare: stassats: well, the cffi extension I wrote currently calls the C compiler directly
2015-10-19T14:43:59Z stassats: sure, but i don't want things i'm not going to use to be in the image
2015-10-19T14:44:40Z Fare: it could parse a pkg-config file.
2015-10-19T14:45:06Z dougk2 joined #sbcl
2015-10-19T14:45:21Z Fare: parsing a .pc file is painful, but doable.
2015-10-19T14:45:35Z stassats: then not making into pkg-config
2015-10-19T14:46:00Z Fare: or calling pkg-config to do the parsing.
2015-10-19T14:50:19Z Fare quit (Ping timeout: 240 seconds)
2015-10-19T14:52:15Z Fare joined #sbcl
2015-10-19T15:00:00Z dougk_ quit (Remote host closed the connection)
2015-10-19T15:14:18Z dougk_ joined #sbcl
2015-10-19T15:16:33Z dougk_: hi everybody! new interpreter is committed
2015-10-19T15:22:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-19T15:23:59Z Fare quit (Ping timeout: 240 seconds)
2015-10-19T15:26:54Z hydan` quit (Ping timeout: 260 seconds)
2015-10-19T15:27:29Z Cymew quit (Ping timeout: 265 seconds)
2015-10-19T15:34:14Z sjl quit (Ping timeout: 260 seconds)
2015-10-19T15:37:34Z dougk2 joined #sbcl
2015-10-19T15:40:36Z stassats: do we need the old one?
2015-10-19T15:42:03Z Fare joined #sbcl
2015-10-19T15:42:17Z stassats: i'm in the process of getting rid of pa_alloc
2015-10-19T15:44:04Z dougk_: stassats: I'd like to think not.
2015-10-19T15:50:19Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-19T15:56:41Z stassats: testing on mips, with qemu, is quite slow
2015-10-19T16:17:28Z dougk2 joined #sbcl
2015-10-19T16:29:00Z failproofshark quit (Ping timeout: 264 seconds)
2015-10-19T16:29:00Z yauz_2 quit (Ping timeout: 264 seconds)
2015-10-19T16:29:01Z p_l quit (Ping timeout: 264 seconds)
2015-10-19T16:30:16Z failproofshark joined #sbcl
2015-10-19T16:30:26Z p_l joined #sbcl
2015-10-19T16:30:32Z yauz joined #sbcl
2015-10-19T16:41:10Z Fare quit (Remote host closed the connection)
2015-10-19T16:51:19Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-19T16:55:51Z dougk2 joined #sbcl
2015-10-19T17:04:21Z eudoxia_ joined #sbcl
2015-10-19T17:05:22Z ASau joined #sbcl
2015-10-19T17:05:50Z eudoxia quit (Ping timeout: 240 seconds)
2015-10-19T17:05:53Z eudoxia_ quit (Client Quit)
2015-10-19T17:09:38Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-19T17:10:57Z stassats: discovered a couple of simple ARM bugs which i missed due to my local-target-features being outdated
2015-10-19T17:16:14Z stassats: ok, no more pa_alloc
2015-10-19T17:16:28Z stassats: seems to work on mips: ship it
2015-10-19T17:27:57Z dougk2 joined #sbcl
2015-10-19T17:33:22Z gucore quit (Ping timeout: 260 seconds)
2015-10-19T17:35:19Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-19T17:48:19Z ASau quit (Quit: ERC (IRC client for Emacs 24.5.1))
2015-10-19T17:48:51Z ASau joined #sbcl
2015-10-19T18:05:11Z fridim_ joined #sbcl
2015-10-19T18:09:56Z dougk2 joined #sbcl
2015-10-19T18:39:35Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-19T18:49:21Z dougk2 joined #sbcl
2015-10-19T19:10:17Z scymtym quit (Remote host closed the connection)
2015-10-19T19:12:09Z attila_lendvai: is it considered a bug if a source file makes sb-c:ir1-convert blow up the stack?
2015-10-19T19:15:22Z attila_lendvai: it's most probably due to an invalid program
2015-10-19T19:17:23Z attila_lendvai: unless macroexpand is called in this stage and a user macro is buggy, which may very well be the case here
2015-10-19T19:20:13Z gingerale quit (Remote host closed the connection)
2015-10-19T19:21:00Z reb: Running tests on HEAD sbcl ... Linux / x86 64-bit has some new failures: threads.impure.lisp and foreign.test.sh
2015-10-19T19:21:06Z reb: ... just noticed.
2015-10-19T19:32:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-19T19:37:14Z dougk2 joined #sbcl
2015-10-19T19:56:04Z lisp joined #sbcl
2015-10-19T19:56:16Z lisp quit (Client Quit)
2015-10-19T19:57:04Z clique joined #sbcl
2015-10-19T20:03:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-19T20:08:32Z attila_lendvai: is this a bug in shadow, or I just fail to understand CLHS? http://paste.lisp.org/+3D97
2015-10-19T20:12:53Z Xof: "If it is accessible as an internal symbol via use-package, it is first imported into package, then exported."
2015-10-19T20:12:59Z Xof: (CLHS EXPORT)
2015-10-19T20:13:17Z Xof: so since the symbol is exported from B, it is present (not inherited) in B, so shadow has no effect
2015-10-19T20:13:28Z stassats: attila_lendvai: you need to use the :shadow option in defpackage
2015-10-19T20:13:57Z dougk2 joined #sbcl
2015-10-19T20:14:36Z attila_lendvai: ooh. managed to code in cl for a decade without having a clue... thanks guys!
2015-10-19T20:14:54Z stassats: yeah, the order is important
2015-10-19T20:14:56Z stassats: clhs defpackage
2015-10-19T20:14:56Z specbot: http://www.lispworks.com/reference/HyperSpec/Body/m_defpkg.htm
2015-10-19T20:15:13Z stassats: right before the Examples section it lists the order
2015-10-19T20:24:05Z heddwch joined #sbcl
2015-10-19T20:25:36Z PuercoPop: is the new fast interpreted going to be included in the default fancy features or so I need to explicitly require it?
2015-10-19T20:26:51Z stassats: the latter
2015-10-19T20:28:34Z PuercoPop nods
2015-10-19T20:36:59Z dougk2 quit (Ping timeout: 259 seconds)
2015-10-19T20:41:37Z ASau quit (Remote host closed the connection)
2015-10-19T20:42:02Z dougk2 joined #sbcl
2015-10-19T20:43:42Z ASau joined #sbcl
2015-10-19T20:47:11Z clique quit (Quit: Page closed)
2015-10-19T21:14:06Z fridim_ quit (Ping timeout: 240 seconds)
2015-10-19T21:58:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-19T22:06:03Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-19T22:12:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-19T22:19:45Z dougk2 joined #sbcl
2015-10-19T22:45:17Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-19T22:52:27Z dougk2 joined #sbcl
2015-10-19T23:12:54Z Ettore quit (Ping timeout: 268 seconds)
2015-10-19T23:14:39Z stassats quit (Ping timeout: 240 seconds)
2015-10-19T23:15:00Z nikki93 joined #sbcl
2015-10-19T23:24:15Z nikki93 quit (Remote host closed the connection)
2015-10-19T23:46:14Z salva quit (Ping timeout: 260 seconds)
2015-10-19T23:58:32Z hydan` joined #sbcl
2015-10-20T00:00:37Z salva joined #sbcl
2015-10-20T00:02:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-20T00:06:24Z stassats joined #sbcl
2015-10-20T00:13:30Z stassats quit (Ping timeout: 240 seconds)
2015-10-20T01:08:20Z heddwch quit (Remote host closed the connection)
2015-10-20T01:33:15Z dougk2 joined #sbcl
2015-10-20T01:35:52Z karswell quit (Ping timeout: 272 seconds)
2015-10-20T02:43:34Z dougk2 quit (Ping timeout: 260 seconds)
2015-10-20T03:21:23Z nikki93 joined #sbcl
2015-10-20T05:44:46Z nikki93 quit (Remote host closed the connection)
2015-10-20T05:45:31Z nikki93 joined #sbcl
2015-10-20T05:58:15Z gingerale joined #sbcl
2015-10-20T06:04:24Z Shinmera joined #sbcl
2015-10-20T06:17:43Z fridim_ joined #sbcl
2015-10-20T06:24:26Z ASau quit (Ping timeout: 240 seconds)
2015-10-20T06:46:51Z nikki93_ joined #sbcl
2015-10-20T06:49:51Z nikki93 quit (Ping timeout: 244 seconds)
2015-10-20T06:59:17Z nikki93 joined #sbcl
2015-10-20T07:01:56Z nikki93_ quit (Ping timeout: 246 seconds)
2015-10-20T07:27:19Z Cymew joined #sbcl
2015-10-20T07:55:09Z Posterdati quit (Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/)
2015-10-20T08:03:45Z Posterdati joined #sbcl
2015-10-20T08:16:28Z nikki93 quit (Remote host closed the connection)
2015-10-20T09:58:44Z Ettore joined #sbcl
2015-10-20T10:29:12Z attila_lendvai joined #sbcl
2015-10-20T10:30:07Z stassats joined #sbcl
2015-10-20T11:58:49Z Posterdati quit (Ping timeout: 244 seconds)
2015-10-20T12:11:40Z Posterdati joined #sbcl
2015-10-20T12:16:46Z eudoxia joined #sbcl
2015-10-20T12:45:06Z hydan` quit (Ping timeout: 255 seconds)
2015-10-20T12:45:21Z dougk2 joined #sbcl
2015-10-20T13:05:46Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-20T13:06:57Z dougk2 joined #sbcl
2015-10-20T13:24:37Z dougk2 quit (Ping timeout: 256 seconds)
2015-10-20T13:30:05Z sjl joined #sbcl
2015-10-20T13:56:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-20T14:21:20Z nyef joined #sbcl
2015-10-20T14:25:17Z whiteline quit (Read error: Connection reset by peer)
2015-10-20T14:26:23Z whiteline joined #sbcl
2015-10-20T14:57:36Z dougk2 joined #sbcl
2015-10-20T14:58:56Z Cymew quit (Ping timeout: 272 seconds)
2015-10-20T15:17:17Z psy_ joined #sbcl
2015-10-20T15:27:41Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-20T15:29:15Z sjl quit (Ping timeout: 260 seconds)
2015-10-20T15:36:43Z sjl joined #sbcl
2015-10-20T15:36:57Z hydan` joined #sbcl
2015-10-20T15:38:21Z dougk2 joined #sbcl
2015-10-20T15:39:20Z psy_ quit (Ping timeout: 252 seconds)
2015-10-20T15:45:42Z sjl quit (Ping timeout: 265 seconds)
2015-10-20T15:47:52Z sjl joined #sbcl
2015-10-20T15:55:20Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-20T15:57:43Z DeadTrickster quit (Read error: No route to host)
2015-10-20T16:21:06Z dougk2 joined #sbcl
2015-10-20T16:34:32Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-20T16:37:06Z mprelude joined #sbcl
2015-10-20T16:39:26Z mprelude quit (Client Quit)
2015-10-20T16:44:50Z Ettore quit (Ping timeout: 260 seconds)
2015-10-20T16:45:46Z mprelude joined #sbcl
2015-10-20T16:49:30Z Ettore1 joined #sbcl
2015-10-20T16:49:51Z Ettore1 left #sbcl
2015-10-20T16:52:01Z psy_ joined #sbcl
2015-10-20T16:55:23Z dougk2 joined #sbcl
2015-10-20T16:58:36Z mprelude quit (Quit: WeeChat 1.3)
2015-10-20T17:06:20Z eudoxia_ joined #sbcl
2015-10-20T17:07:38Z eudoxia_ quit (Client Quit)
2015-10-20T17:08:50Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-20T17:10:01Z eudoxia quit (Ping timeout: 250 seconds)
2015-10-20T17:15:37Z phoe_krk joined #sbcl
2015-10-20T17:16:54Z Ettore joined #sbcl
2015-10-20T17:30:42Z sjl quit (Ping timeout: 250 seconds)
2015-10-20T17:35:30Z dougk2 joined #sbcl
2015-10-20T17:44:13Z attila_lendvai joined #sbcl
2015-10-20T17:44:13Z attila_lendvai quit (Changing host)
2015-10-20T17:44:13Z attila_lendvai joined #sbcl
2015-10-20T18:06:57Z ASau joined #sbcl
2015-10-20T18:30:43Z Ettore quit (Quit: Leaving.)
2015-10-20T18:33:39Z hydan quit (Ping timeout: 250 seconds)
2015-10-20T18:41:25Z clique joined #sbcl
2015-10-20T18:50:08Z gingerale quit (Remote host closed the connection)
2015-10-20T18:50:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-20T19:09:02Z Bicyclidine joined #sbcl
2015-10-20T19:09:22Z scymtym joined #sbcl
2015-10-20T19:23:51Z DeadTrickster joined #sbcl
2015-10-20T19:53:37Z Ettore joined #sbcl
2015-10-20T19:57:05Z clique quit (Quit: Page closed)
2015-10-20T20:11:20Z clique joined #sbcl
2015-10-20T20:14:22Z fridim_ quit (Ping timeout: 260 seconds)
2015-10-20T20:27:42Z clique quit (Ping timeout: 246 seconds)
2015-10-20T20:37:27Z sjl joined #sbcl
2015-10-20T20:39:33Z clique joined #sbcl
2015-10-20T20:50:42Z Ettore quit (Quit: Leaving.)
2015-10-20T20:54:17Z sjl quit (Read error: Connection reset by peer)
2015-10-20T20:55:04Z sjl joined #sbcl
2015-10-20T20:56:08Z clique quit (Quit: Page closed)
2015-10-20T20:59:38Z Ettore joined #sbcl
2015-10-20T21:01:30Z hydan` quit (Ping timeout: 260 seconds)
2015-10-20T21:03:20Z sjl quit (Ping timeout: 268 seconds)
2015-10-20T21:05:23Z sjl joined #sbcl
2015-10-20T21:10:55Z karswell joined #sbcl
2015-10-20T21:12:53Z hydan joined #sbcl
2015-10-20T21:30:07Z whiteline quit (Quit: Leaving)
2015-10-20T21:31:20Z Ettore quit (Quit: Leaving.)
2015-10-20T22:09:21Z sjl quit (Ping timeout: 256 seconds)
2015-10-20T22:12:08Z sjl joined #sbcl
2015-10-20T22:20:22Z psy_ quit (Ping timeout: 244 seconds)
2015-10-20T22:21:42Z psy_ joined #sbcl
2015-10-20T22:28:29Z eudoxia joined #sbcl
2015-10-20T22:28:37Z antoszka quit (Read error: Connection reset by peer)
2015-10-20T22:39:13Z karswell quit (Read error: Connection reset by peer)
2015-10-20T22:41:01Z antoszka joined #sbcl
2015-10-20T22:41:35Z antoszka quit (Client Quit)
2015-10-20T22:41:50Z antoszka joined #sbcl
2015-10-20T22:46:24Z dougk2 joined #sbcl
2015-10-20T22:58:46Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-20T23:10:56Z eudoxia quit (Remote host closed the connection)
2015-10-20T23:12:44Z Bicyclidine quit (Quit: leaving)
2015-10-20T23:12:55Z karswell joined #sbcl
2015-10-20T23:28:27Z phoe_krk quit (Remote host closed the connection)
2015-10-20T23:48:06Z sjl quit (Ping timeout: 240 seconds)
2015-10-20T23:50:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-21T00:01:26Z scymtym quit (Read error: Connection reset by peer)
2015-10-21T00:21:05Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-21T00:38:23Z dougk2 joined #sbcl
2015-10-21T00:47:12Z dougk2_ joined #sbcl
2015-10-21T00:48:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-21T00:56:15Z dougk2 joined #sbcl
2015-10-21T00:57:59Z dougk2_ quit (Ping timeout: 240 seconds)
2015-10-21T01:07:17Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-21T01:18:13Z nikki93 joined #sbcl
2015-10-21T01:24:28Z nikki93 quit (Remote host closed the connection)
2015-10-21T01:29:01Z dougk2 joined #sbcl
2015-10-21T01:38:55Z nikki93 joined #sbcl
2015-10-21T01:45:24Z nikki93_ joined #sbcl
2015-10-21T01:46:06Z nikki93__ joined #sbcl
2015-10-21T01:46:23Z nikki93 quit (Ping timeout: 264 seconds)
2015-10-21T01:49:38Z nikki93_ quit (Ping timeout: 246 seconds)
2015-10-21T01:54:53Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-21T02:11:39Z nikki93__ quit (Remote host closed the connection)
2015-10-21T02:37:23Z psy_ quit (Ping timeout: 256 seconds)
2015-10-21T02:59:26Z nicdev` quit (Remote host closed the connection)
2015-10-21T03:11:54Z nikki93 joined #sbcl
2015-10-21T03:26:56Z stassats quit (Ping timeout: 246 seconds)
2015-10-21T03:29:45Z nikki93_ joined #sbcl
2015-10-21T03:33:52Z nikki93 quit (Ping timeout: 272 seconds)
2015-10-21T03:42:45Z nikki93_ quit (Remote host closed the connection)
2015-10-21T03:44:22Z psy_ joined #sbcl
2015-10-21T03:45:45Z hydan` joined #sbcl
2015-10-21T03:53:55Z psy_ quit (Remote host closed the connection)
2015-10-21T04:59:11Z luis quit (Quit: ZNC - http://znc.sourceforge.net)
2015-10-21T05:00:17Z |3b| quit (Remote host closed the connection)
2015-10-21T05:47:38Z |3b| joined #sbcl
2015-10-21T05:58:39Z Shinmera joined #sbcl
2015-10-21T06:35:39Z nikki93 joined #sbcl
2015-10-21T06:37:30Z nikki93_ joined #sbcl
2015-10-21T06:40:19Z nikki93 quit (Ping timeout: 240 seconds)
2015-10-21T06:46:52Z fridim_ joined #sbcl
2015-10-21T06:51:27Z nikki93_ quit (Remote host closed the connection)
2015-10-21T06:51:52Z nikki93 joined #sbcl
2015-10-21T07:04:13Z karswell` joined #sbcl
2015-10-21T07:04:45Z karswell quit (Remote host closed the connection)
2015-10-21T07:07:53Z Quadrescence quit (Ping timeout: 244 seconds)
2015-10-21T07:14:30Z nikki93 quit (Remote host closed the connection)
2015-10-21T07:15:15Z ASau quit (Ping timeout: 255 seconds)
2015-10-21T07:58:55Z Cymew joined #sbcl
2015-10-21T08:10:19Z whiteline joined #sbcl
2015-10-21T09:00:09Z brucem quit (Quit: ZNC - http://znc.sourceforge.net)
2015-10-21T09:39:58Z attila_lendvai joined #sbcl
2015-10-21T09:39:58Z attila_lendvai quit (Changing host)
2015-10-21T09:39:58Z attila_lendvai joined #sbcl
2015-10-21T09:42:46Z scymtym joined #sbcl
2015-10-21T09:51:59Z brucem joined #sbcl
2015-10-21T09:52:55Z brucem quit (Changing host)
2015-10-21T09:52:55Z brucem joined #sbcl
2015-10-21T10:08:03Z failproofshark quit (Remote host closed the connection)
2015-10-21T10:42:38Z zacts quit (Read error: Connection reset by peer)
2015-10-21T10:43:53Z zacts joined #sbcl
2015-10-21T11:32:05Z stassats joined #sbcl
2015-10-21T12:01:00Z Quadrescence joined #sbcl
2015-10-21T12:25:01Z Ettore joined #sbcl
2015-10-21T12:27:20Z abeaumont quit (Ping timeout: 250 seconds)
2015-10-21T12:58:39Z anthracite joined #sbcl
2015-10-21T13:02:54Z sjl joined #sbcl
2015-10-21T13:23:26Z _iwc joined #sbcl
2015-10-21T13:27:41Z eudoxia joined #sbcl
2015-10-21T13:36:18Z _iwc quit (Killed (Sigyn (Spam is off topic on freenode.)))
2015-10-21T14:07:20Z stassats` joined #sbcl
2015-10-21T14:09:14Z stassats quit (Ping timeout: 260 seconds)
2015-10-21T14:20:52Z psy_ joined #sbcl
2015-10-21T14:23:57Z reb quit (Ping timeout: 244 seconds)
2015-10-21T14:52:26Z reb joined #sbcl
2015-10-21T15:05:06Z eudoxia quit (Quit: Leaving)
2015-10-21T15:16:06Z gingerale joined #sbcl
2015-10-21T15:31:30Z sjl quit (Ping timeout: 240 seconds)
2015-10-21T15:48:24Z anthracite quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2015-10-21T16:58:48Z Cymew quit (Ping timeout: 252 seconds)
2015-10-21T17:09:07Z Cymew joined #sbcl
2015-10-21T17:18:50Z clique joined #sbcl
2015-10-21T17:38:24Z hydan quit (Ping timeout: 252 seconds)
2015-10-21T17:49:14Z abeaumont joined #sbcl
2015-10-21T17:54:50Z Cymew quit (Read error: Connection reset by peer)
2015-10-21T17:55:08Z ASau joined #sbcl
2015-10-21T18:15:56Z angavrilov quit (Ping timeout: 244 seconds)
2015-10-21T18:31:55Z angavrilov joined #sbcl
2015-10-21T18:33:42Z gingerale quit (Remote host closed the connection)
2015-10-21T18:56:15Z clique quit (Ping timeout: 246 seconds)
2015-10-21T19:38:58Z sjl joined #sbcl
2015-10-21T19:40:48Z clique joined #sbcl
2015-10-21T19:45:42Z sjl__ joined #sbcl
2015-10-21T19:48:03Z sjl quit (Ping timeout: 268 seconds)
2015-10-21T19:51:37Z scymtym: dougk_: i'm done reviewing the "Update compiler and runtime …" commit. found nothing major, jsut a few nitpicks: http://paste.lisp.org/display/157258. the only actual error i found is not using DEFN in the INSPECTED-PARTS method for SB-EVAL:INTERPRETED-FUNCTION. all other changes may be a matter of preference.
2015-10-21T19:52:03Z clique quit (Quit: Page closed)
2015-10-21T19:52:56Z scymtym: s/error/most likely mistake/
2015-10-21T20:09:51Z reb: dougk_: FYI: ./run-tests.sh foreign.test.sh generates an interpreter-type-error.  Might be a bug in the test.
2015-10-21T20:14:25Z dougk_: reb: I suspect that failure and the fact that I had to #- out some tests in alien.impure is no coincidence.
2015-10-21T20:18:51Z clique joined #sbcl
2015-10-21T20:42:41Z clique quit (Quit: Page closed)
2015-10-21T21:01:02Z rj-code quit (Quit: :q)
2015-10-21T21:14:21Z Bicyclidine joined #sbcl
2015-10-21T21:28:09Z hydan joined #sbcl
2015-10-21T21:35:15Z fridim_ quit (Ping timeout: 256 seconds)
2015-10-21T21:38:24Z rszeno joined #sbcl
2015-10-21T21:49:04Z dougk_: scymtym: thanks, i'll incorporate those
2015-10-21T22:00:29Z pchrist quit (Remote host closed the connection)
2015-10-21T22:12:12Z scymtym: dougk_: great, thanks
2015-10-21T22:43:32Z Ettore quit (Quit: Leaving.)
2015-10-21T22:57:40Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-21T23:02:42Z karswell` quit (Read error: Connection reset by peer)
2015-10-21T23:02:55Z karswell` joined #sbcl
2015-10-21T23:06:49Z Bicyclidine quit (Quit: leaving)
2015-10-21T23:14:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-22T00:30:27Z stassats` quit (Ping timeout: 265 seconds)
2015-10-22T00:33:19Z scymtym quit (Ping timeout: 240 seconds)
2015-10-22T00:36:30Z sjl__ quit (Ping timeout: 240 seconds)
2015-10-22T01:00:13Z karswell` quit (Ping timeout: 252 seconds)
2015-10-22T01:13:42Z abbe joined #sbcl
2015-10-22T01:39:54Z whiteline quit (Ping timeout: 260 seconds)
2015-10-22T01:40:21Z rszeno quit (Quit: Leaving.)
2015-10-23T09:29:26Z ccl-logbot joined #sbcl
2015-10-23T09:29:26Z 
2015-10-23T09:29:26Z names: ccl-logbot sigjuice whiteline_ whiteline heddwch loke Cymew psy_ fridim_ Shinmera hydan zacts chris2 fikusz dougk_ Quadrescence brucem foom frankS2 PuercoPop White__Flame scymtym__ nzambe mood pchrist nyef abbe angavrilov abeaumont reb |3b| hydan` antoszka DeadTrickster Posterdati salva yauz p_l DGASAU kanru jlarocco_ ym Bike specbot minion jdz alchemis7 christoph_debian jackdaniel drmeister jackc-_ pkhuong psilord pootler jozip dim les` hzp Ober ferada
2015-10-23T09:29:26Z names: joshe fe[nl]ix Blkt flip214 carvite jsnell dustinm` jlarocco irsol dsp_
2015-10-23T09:34:50Z attila_lendvai joined #sbcl
2015-10-23T10:00:20Z jackc- joined #sbcl
2015-10-23T10:01:06Z pkhuong_ joined #sbcl
2015-10-23T10:02:25Z sjl joined #sbcl
2015-10-23T10:02:41Z psilord1 joined #sbcl
2015-10-23T10:02:42Z p_l_ joined #sbcl
2015-10-23T10:03:50Z pchrist_ joined #sbcl
2015-10-23T10:07:23Z p_l quit (Disconnected by services)
2015-10-23T10:07:30Z p_l_ is now known as p_l
2015-10-23T10:07:44Z loke quit (*.net *.split)
2015-10-23T10:07:44Z pchrist quit (*.net *.split)
2015-10-23T10:07:45Z psilord quit (*.net *.split)
2015-10-23T10:07:45Z pkhuong quit (*.net *.split)
2015-10-23T10:07:45Z jackc-_ quit (*.net *.split)
2015-10-23T10:07:45Z drmeister quit (*.net *.split)
2015-10-23T10:07:45Z jackdaniel quit (*.net *.split)
2015-10-23T10:09:30Z jackdani1l joined #sbcl
2015-10-23T10:10:41Z jackdani1l left #sbcl
2015-10-23T10:12:45Z p_l quit (Ping timeout: 244 seconds)
2015-10-23T10:13:37Z p_l joined #sbcl
2015-10-23T10:14:44Z loke joined #sbcl
2015-10-23T10:15:11Z sjl quit (Ping timeout: 246 seconds)
2015-10-23T10:15:23Z sjl_ joined #sbcl
2015-10-23T10:15:40Z drmeister joined #sbcl
2015-10-23T10:22:35Z sjl_ quit (Ping timeout: 260 seconds)
2015-10-23T10:27:18Z sjl joined #sbcl
2015-10-23T10:36:30Z hydan quit (Ping timeout: 240 seconds)
2015-10-23T10:56:15Z stassats joined #sbcl
2015-10-23T11:08:21Z man213 joined #sbcl
2015-10-23T11:17:20Z abeaumont quit (Ping timeout: 244 seconds)
2015-10-23T11:33:44Z attila_lendvai quit (Quit: Leaving.)
2015-10-23T11:33:50Z attila_lendvai1 joined #sbcl
2015-10-23T11:33:50Z attila_lendvai1 is now known as attila_lendvai
2015-10-23T11:33:50Z attila_lendvai quit (Changing host)
2015-10-23T11:33:50Z attila_lendvai joined #sbcl
2015-10-23T13:04:46Z ZabaQ joined #sbcl
2015-10-23T14:11:54Z jackdaniel joined #sbcl
2015-10-23T14:13:33Z lnostdal joined #sbcl
2015-10-23T14:14:25Z lnostdal: everyone hates sourceforge these days; https://i.imgur.com/aATLDSs.png ...... :)
2015-10-23T14:14:38Z eudoxia joined #sbcl
2015-10-23T14:15:02Z jackdaniel: lnostdal: heheh
2015-10-23T14:22:17Z stassats: pretty stupid, if you ask me
2015-10-23T14:52:32Z Cymew quit (Ping timeout: 268 seconds)
2015-10-23T15:06:38Z lnostdal quit (Remote host closed the connection)
2015-10-23T15:08:05Z scymtym joined #sbcl
2015-10-23T15:17:23Z foom: Well, sourceforge *was* adding unwanted software to installers hosted there.
2015-10-23T15:17:29Z foom: (or is still? not sure)
2015-10-23T15:18:05Z foom: You could suggest that they only block files named ".exe", though. :)
2015-10-23T15:18:18Z stassats: a blanket ban is not a solution
2015-10-23T15:19:10Z stassats: maybe it should just block the internet
2015-10-23T15:19:18Z stassats: after all, you can find all sorts of malware there
2015-10-23T15:19:34Z foom: Well, that's why people download and install uBlock
2015-10-23T15:19:47Z foom: To block malware on the internet.
2015-10-23T15:20:00Z stassats: and not advertisements?
2015-10-23T15:20:18Z foom: Those too
2015-10-23T15:21:10Z stassats: if something has too many false positives then it doesn't improve security
2015-10-23T15:21:49Z stassats: but oh well, the internet mob can only accept things that are exactly perfect
2015-10-23T15:39:24Z chris2 quit (Ping timeout: 268 seconds)
2015-10-23T15:45:10Z man2131 joined #sbcl
2015-10-23T15:46:10Z man213 quit (Ping timeout: 250 seconds)
2015-10-23T16:08:05Z chris2 joined #sbcl
2015-10-23T16:10:43Z gingerale joined #sbcl
2015-10-23T16:13:08Z hydan joined #sbcl
2015-10-23T16:35:26Z ZabaQ quit (Ping timeout: 240 seconds)
2015-10-23T16:38:33Z dra joined #sbcl
2015-10-23T16:42:50Z dra quit (Ping timeout: 244 seconds)
2015-10-23T16:53:24Z fridim_ quit (Ping timeout: 265 seconds)
2015-10-23T16:58:41Z man2131 quit (Quit: Leaving.)
2015-10-23T17:01:16Z White__Flame is now known as White_Flame
2015-10-23T17:04:07Z psy_ quit (Read error: No route to host)
2015-10-23T17:18:38Z eudoxia_ joined #sbcl
2015-10-23T17:21:56Z eudoxia quit (Ping timeout: 250 seconds)
2015-10-23T17:41:05Z psy_ joined #sbcl
2015-10-23T17:44:14Z eudoxia_ quit (Quit: Leaving)
2015-10-23T17:44:24Z eudoxia joined #sbcl
2015-10-23T17:51:11Z ASau joined #sbcl
2015-10-23T18:15:00Z pchrist_ is now known as pchrist
2015-10-23T18:40:17Z nyef: Hrm... Okay, just threw together a quick test program for calling pow() without getting constant-folded, and I had to pass -lm to the compiler to get it to link... And the result was 7fe00000 0000008d.
2015-10-23T18:40:37Z nyef: For pow(2.0d, 1023.0d).
2015-10-23T18:40:56Z nyef: But, constant-folded, it's known to be 7fe00000 00000000.
2015-10-23T18:41:15Z stassats: pushed some mips disassembler changes earlier today
2015-10-23T18:41:27Z sjl quit (Ping timeout: 260 seconds)
2015-10-23T18:46:05Z nyef: So, is pow() doing something wrong, is this permissible variation, or is my FPU state misconfigured?
2015-10-23T18:46:27Z nyef: (Or, alternately, is some FPU emulation somehow broken?)
2015-10-23T18:55:11Z eudoxia quit (Quit: Leaving)
2015-10-23T18:57:13Z nyef: ... No effect from tweaking the rounding-mode control, presuming that it's actually implemented...
2015-10-23T18:58:57Z pkhuong_: could definitely be a shitty libm
2015-10-23T18:59:19Z pkhuong_: gcc's emulation can be more accurate than the real deal
2015-10-23T18:59:20Z nyef: Not implausible, under the circumstances.
2015-10-23T19:02:25Z pkhuong_: yeah. so your libm implements pow as exp . log
2015-10-23T19:02:51Z pkhuong_: (exp (* 1023d0 (log 2d0))) => 8.98846567431186d307; (format t "~x~%" (sb-impl::double-float-high-bits 8.98846567431186d307)) => 7FE00000; (format t "~x~%" (sb-impl::double-float-low-bits 8.98846567431186d307)) => 8D
2015-10-23T19:03:18Z nyef: http://paste.lisp.org/display/157347
2015-10-23T19:03:45Z pkhuong_: bad libm.
2015-10-23T19:03:49Z nyef: Okay, is that a legitimate implementation choice?
2015-10-23T19:04:00Z pkhuong_: I think it's technically allowed by the standard
2015-10-23T19:04:11Z pkhuong_: it earns a frowny face from my end, fwiw.
2015-10-23T19:04:11Z nyef: Damn.
2015-10-23T19:05:24Z pkhuong_: plus 754R wasn't ratified until very recently
2015-10-23T19:06:10Z nyef: Let's pretend that my hardware is almost old enough to vote, shall we?
2015-10-23T19:06:27Z pkhuong_: you're looking at a software problem.
2015-10-23T19:06:34Z pkhuong_: a newer libm might just work
2015-10-23T19:06:37Z nyef: That's good, I guess.
2015-10-23T19:06:50Z nyef: ... Except that, apparently, I installed this thing three months ago.
2015-10-23T19:08:13Z gargaml joined #sbcl
2015-10-23T19:11:11Z stassats: i have glibc-2.13
2015-10-23T19:11:32Z nyef: I can't help but wonder if I might be better off "just" writing a suitable implementation of pow() for SBCL?
2015-10-23T19:12:39Z stassats: to bring in more bugs?
2015-10-23T19:12:50Z nyef: Something like that, yes.
2015-10-23T19:20:29Z Posterdati quit (Ping timeout: 246 seconds)
2015-10-23T19:23:06Z nyef: I have glibc-2.20, apparently?
2015-10-23T19:33:27Z stassats: do you have a real fpu?
2015-10-23T19:34:26Z nyef: Yes.
2015-10-23T19:34:38Z nyef: Standard on this hardware era.
2015-10-23T19:35:02Z nyef: Doesn't necessarily mean that it's getting used, though.
2015-10-23T19:38:45Z dra joined #sbcl
2015-10-23T20:00:13Z fridim_ joined #sbcl
2015-10-23T20:05:14Z pkhuong_ is now known as pkhuong
2015-10-23T20:05:55Z pkhuong: "just"?
2015-10-23T20:12:08Z nyef: Well, I'm full well aware that I haven't much of a clue as to what I'm doing with floating-point stuff, so yeah, "just".
2015-10-23T20:17:41Z sjl joined #sbcl
2015-10-23T20:18:33Z nyef: stassats: Is your glibc-2.13 on real MIPS hardware, or under qemu?
2015-10-23T20:19:16Z nyef: Because if it's under qemu, we might be looking at the x86oid 80-bit internal FPU representation tidying things up behind the scenes.
2015-10-23T20:20:01Z sjl quit (Read error: Connection reset by peer)
2015-10-23T20:21:46Z heddwch is now known as agent_nchambers
2015-10-23T20:22:27Z agent_nchambers is now known as heddwch
2015-10-23T20:25:03Z psy_ quit (Ping timeout: 260 seconds)
2015-10-23T20:30:02Z stassats: qemu
2015-10-23T20:32:00Z nyef: Hrm. And I can confirm that I get all of the same wrong answers on x86-64 as I do on MIPS if I treat (EXPT B E) as (EXP (* E (LOG B))).
2015-10-23T20:44:17Z Bicyclidine joined #sbcl
2015-10-23T21:02:45Z gargaml quit (Quit: WeeChat 1.1.1)
2015-10-23T21:04:59Z hydan quit (Remote host closed the connection)
2015-10-23T21:18:55Z dra quit (Ping timeout: 252 seconds)
2015-10-23T21:32:20Z dra joined #sbcl
2015-10-23T21:39:53Z hydan joined #sbcl
2015-10-23T21:53:57Z nyef: Preliminary conclusion: This pow() thing is an issue...  And I have a very hard time following what's going on with the glibc implementation.
2015-10-23T21:58:24Z stassats: noticed that it takes four instructions to fix up CODE on arm64
2015-10-23T21:58:55Z stassats: yet it could be done in one instruction within a megabyte, that would cover almost all functions
2015-10-23T22:00:02Z stassats: now i'm wondering, what's pinning down the code object before CODE is being set up?
2015-10-23T22:00:57Z stassats: lexenv-offset?
2015-10-23T22:01:31Z stassats: lexenv-tn, that is
2015-10-23T22:02:47Z nyef: lexenv can't do so for closures.
2015-10-23T22:03:23Z nyef: Isn't CODE also used for the closure-fun / simple-fun-self during the funcall sequence?
2015-10-23T22:03:31Z stassats: but the closure refers to the simple-fun
2015-10-23T22:04:33Z nyef: But you have an interior-pointer, so you need a direct, tagged pointer in the register set.
2015-10-23T22:05:01Z nyef: The closure doesn't do that, because it's at a single level of indirection.
2015-10-23T22:05:57Z stassats: is that the reason for the closure tramp?
2015-10-23T22:06:46Z stassats: doesn't seem like
2015-10-23T22:07:00Z stassats: still don't know what closure_tramp does
2015-10-23T22:08:25Z Bicyclidine quit (Quit: nap)
2015-10-23T22:09:16Z hydan quit (Remote host closed the connection)
2015-10-23T22:09:41Z hydan joined #sbcl
2015-10-23T22:10:29Z stassats: dang, something went with my merging
2015-10-23T22:10:33Z stassats: lost some changes
2015-10-23T22:24:00Z nyef: closure_tramp is similar to undefined_tramp, in that it gets stuffed into the entry point slot of an fdefinition for NAMED calls.
2015-10-23T22:25:15Z nyef: Basically, a non-NAMED call has a function object, either a closure, a funcallable-instance, or a SIMPLE-FUN. They treat that as the LEXENV, grab the funcallable-instance-fun/closure-fun/simple-fun-self, and jump to the appropriate entry point.
2015-10-23T22:25:44Z stassats: looks i'll have to redo some stuff
2015-10-23T22:26:07Z nyef: A NAMED call has an FDEFINITION, and it loads the function (a LEXENV, and could even be NIL) then jumps through the "code" pointer, which is the entry point for a simple-fun, or an appropriate trampoline.
2015-10-23T22:26:31Z nyef: There's some other wrinkle with funcallable-instances, I forget quite what.
2015-10-23T22:26:46Z stassats: found some *vc-diff* buffer with some of the stuff
2015-10-23T22:27:35Z nyef: But basically, in order to speed up the common case with a named call, it only loads LEXENV and jumps, so if there's any sort of complicated processing (that is, it's not a SIMPLE-FUN) then the real entry point has to be grovelled from the LEXENV.
2015-10-23T22:29:07Z nyef: Otherwise you're looking at another memory read for each named funcall to ground the entry point properly for non-SIMPLE-FUN cases, plus larger FDEFINITION objects, or hairier logic for dealing with FDEFNs in the GC, plus still having to have an undefined-tramp...
2015-10-23T22:30:13Z stassats: how is the code object being held down anyway?
2015-10-23T22:30:25Z stassats: before CODE is set up
2015-10-23T22:30:57Z nyef: The funcall sequence loads CODE with the correct SIMPLE-FUN, which is embedded within the code object.
2015-10-23T22:31:07Z nyef: That holds it until the "real" value for CODE can be found.
2015-10-23T22:33:39Z stassats: doesn't do it here
2015-10-23T22:33:43Z stassats: for call-named
2015-10-23T22:34:46Z hydan quit (Remote host closed the connection)
2015-10-23T22:34:59Z scymtym quit (Ping timeout: 264 seconds)
2015-10-23T22:35:10Z hydan joined #sbcl
2015-10-23T22:36:54Z stassats: first thing the calle sets up LR, which is the function object
2015-10-23T22:37:43Z stassats: how are jumps interrupted?
2015-10-23T22:38:18Z stassats: i bet the threads will unravel a lot of issues
2015-10-23T22:42:01Z stassats: i guess i can insert do_pending_interrupt just before LR is being set up
2015-10-23T22:45:09Z stassats: ha
2015-10-23T22:45:20Z stassats: PC-OFFSET not in code object
2015-10-23T22:45:57Z stassats: gotcha!
2015-10-23T22:46:13Z stassats: nyef: so, this thing is broken, and it looks the same on arm32
2015-10-23T22:47:37Z stassats: i was happy to catch this, but now i'm sad because i have to fix it
2015-10-23T22:48:43Z stassats: let's take a look at ppc
2015-10-23T22:50:25Z jdz quit (Ping timeout: 256 seconds)
2015-10-23T22:50:41Z stassats: ppc sets up $CODE first thing, but that's still not enough
2015-10-23T22:51:02Z stassats: it has threads, i can try a real world test
2015-10-23T22:55:38Z jdz joined #sbcl
2015-10-23T22:56:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-23T22:59:36Z stassats: maybe i'm just not seeing how ppc solves it, but my test doesn't yet crash
2015-10-23T23:02:07Z stassats: i guess i should try the pending interrupt trick on ppc first
2015-10-23T23:07:30Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-23T23:07:45Z nyef: For an FDEFN, the FDEFN-FUN slot contains a value suitable to pin the entry point.
2015-10-23T23:07:52Z nyef: Unless the entry point is a trampoline.
2015-10-23T23:08:25Z stassats: the problem demonstrates itself only when i access a CODE constant
2015-10-23T23:09:09Z nyef: ?
2015-10-23T23:10:04Z stassats: it seems to return to the function, but CODE is hosed
2015-10-23T23:10:31Z nyef: ARM, like SPARC, is a special case: The entry points in the FDEFN are tagged pointers.
2015-10-23T23:10:41Z stassats: this is arm64
2015-10-23T23:10:52Z nyef: So arm64 is plausibly hosed.
2015-10-23T23:11:47Z nyef: On PPC, in a WITHOUT-SCHEDULING, the named-call sequence loads FUNCTION and then ENTRY-POINT from the FDEFN.
2015-10-23T23:12:19Z nyef: ... And FUNCTION is in a descriptor-reg, and thus grounds the code object and interior pointers.
2015-10-23T23:12:59Z nyef: There's your bug. I'm not seeing the corresponding noise for ARM64.
2015-10-23T23:16:13Z stassats: i like finding bugs by looking at the code and asking "what will happen when"
2015-10-23T23:16:24Z stassats: as opposed to seeing things crashing
2015-10-23T23:16:48Z nikki93 joined #sbcl
2015-10-23T23:18:15Z gingerale quit (Remote host closed the connection)
2015-10-23T23:19:01Z stassats: so, having to load the function, there's not much of a point in having a raw-addr, is there?
2015-10-23T23:21:05Z stassats: on arm64, i can do LDP to load both in one shot
2015-10-23T23:21:46Z stassats: or can i, ldp can't do tagged pointers
2015-10-23T23:22:53Z nyef: Hence why ARM and SPARC use tagged "raw addr"s: They can apply the displacement in the jump instruction.
2015-10-23T23:23:22Z nyef: x86oids do their own thing, and every other platform has a raw-addr that has to be loaded to an interior pointer *after* loading the FDEFN-FUN.
2015-10-23T23:23:23Z stassats: but if i have two loads, i can just use an addition for the second load?
2015-10-23T23:23:52Z stassats: or is it variable? i don't remember
2015-10-23T23:23:54Z nyef: No, because FDEFN-FUN might be a closure or funcallable-instance, not a SIMPLE-FUN.
2015-10-23T23:24:13Z stassats: ok, that seems to have fixed
2015-10-23T23:24:19Z stassats: let's revert back and see if it breaks it again
2015-10-23T23:26:15Z nyef: This sort of thing is one reason why I did the ARM backend piecemeal: I always knew which bits still needed to be done, and I had to take the time to actually understand each piece in turn.
2015-10-23T23:28:20Z stassats: understanding a working system is easier
2015-10-23T23:30:44Z stassats: ok, that fixes it
2015-10-23T23:31:25Z stassats: and i just wanted to add disassembly annotations to MIPS
2015-10-23T23:31:39Z stassats: ended up getting a clue on how to make complicated aliases on ARM
2015-10-23T23:31:57Z stassats: and discovered this discrepancy during staring the disassembly output
2015-10-23T23:32:41Z stassats: and i found a good way to test gc safety
2015-10-23T23:33:11Z stassats: insert (when *x* (inst brk pending-interrupt-trap)) where you want into the vop, set *x* to T during compilation
2015-10-23T23:33:21Z stassats: and then set *gc-pending* to T
2015-10-23T23:37:23Z dra quit (Remote host closed the connection)
2015-10-23T23:37:32Z stassats: ok, ppc does CODE set up in one instruction
2015-10-23T23:37:36Z stassats: i want too
2015-10-23T23:39:08Z stassats: i can reclaim more than enough space lost to loading of the tagged pointer
2015-10-23T23:41:52Z nyef: ... One instruction? What?
2015-10-23T23:42:28Z nyef: No, it's a one-or-three instruction sequence.
2015-10-23T23:42:45Z stassats: it's one instruction on any given day
2015-10-23T23:42:57Z stassats: i mean, who has 1MB code objects?
2015-10-23T23:43:07Z nyef: You do, at times, IIRC.
2015-10-23T23:44:48Z stassats: it's 1MB in one instruction, 32GB in three instructions
2015-10-23T23:45:10Z stassats: oughta be enough for everyone
2015-10-23T23:45:21Z nyef: I still like the ARM version of INST COMPUTE-CODE.
2015-10-23T23:45:29Z nyef: Even if it IS always three instructions.
2015-10-23T23:46:10Z stassats: it's translation into arm64 ended up with 4 instructions
2015-10-23T23:46:53Z stassats: no widetag tricks possible
2015-10-23T23:47:38Z nyef: Oh?
2015-10-23T23:48:06Z stassats: see the comment in the arm version of compute-code
2015-10-23T23:48:31Z nyef: I suppose arranging for the SIMPLE_FUN_WIDETAG to have the right high bits to strip off the FUN_POINTER_LOWTAG might be a bit much...
2015-10-23T23:49:11Z stassats: i already have 1-3 instruction code for compute-lra, so this is trivial to change
2015-10-23T23:50:48Z nyef: Hang on, really? If anything, you could get another couple of bits of displacement, couldn't you?
2015-10-23T23:51:07Z stassats: what do you mean?
2015-10-23T23:51:18Z nyef: Count in words on 64-bit means a three-bit shift, which means your low bit is set given widetags of 2A and 36, right?
2015-10-23T23:51:55Z stassats: i don't want to use that approach anyway
2015-10-23T23:51:57Z nyef: So you compensate by tweaking the displacement in the SUB instruction...
2015-10-23T23:52:04Z nyef: Fair enough, I guess...
2015-10-23T23:52:59Z stassats: it does a memory load, patching up the PC relative address is three instructions for 35 bits
2015-10-24T00:15:36Z nikki93 quit (Remote host closed the connection)
2015-10-24T00:16:15Z nikki93 joined #sbcl
2015-10-24T00:17:26Z stassats: and my disassembly of ADR is broken
2015-10-24T00:21:11Z nikki93 quit (Ping timeout: 264 seconds)
2015-10-24T00:25:11Z stassats: and i learned that i can find the code header length by looking at the ppc disassembly, so, it's full of accidental discoveries
2015-10-24T00:27:41Z dougk2 joined #sbcl
2015-10-24T00:27:56Z karswell joined #sbcl
2015-10-24T00:36:39Z dougk2: scymtym__: i'm pushing your edits. They LGTM.
2015-10-24T00:47:24Z nyef: Hrm. Messing around with (sb-kernel:%pow 2.0d0 X) for various values of X gets me a first bit of damage at 12d0.
2015-10-24T00:47:38Z nyef: (%pow 2.0d0 11d0) => 2048.0d0.
2015-10-24T00:48:02Z nyef: But (%pow 2.0d0 12d0) => 4096.000000000001d0.
2015-10-24T00:51:06Z stassats: other backends calculate the offsets from LIP, i can just do it from the PC
2015-10-24T00:51:23Z stassats: which should be the same? not all ISAs have access to PC, maybe that's why
2015-10-24T00:51:36Z nyef: stassats: Just don't clobber whatever register is backing PC as an interior pointer as you do so.
2015-10-24T00:52:08Z stassats: i either load straight to CODE or mumble over LRA and TMP
2015-10-24T00:54:03Z stassats: precise GC is hard
2015-10-24T00:54:06Z stassats: let's go home
2015-10-24T00:54:37Z stassats: moving x86-64 to a precise GC would be akin to making a new backend
2015-10-24T00:55:41Z stassats: ok, this CODE trick saves a good chunk of core space
2015-10-24T00:56:19Z stassats: a megabyte or so
2015-10-24T00:58:51Z stassats: should be faster too
2015-10-24T00:59:05Z stassats: three fewer instructions, one less memory access
2015-10-24T00:59:39Z stassats: (that ought to make the fewer-less nazis unhappy)
2015-10-24T01:00:49Z nyef: ... because they have little ground to complain about it?
2015-10-24T01:02:37Z stassats: ok, everything is primed for the threads
2015-10-24T01:02:58Z stassats: still waiting for it for the next month, to give myself a mental break
2015-10-24T01:07:55Z stassats: x86-64 core is 44M, ARM64 is 51M
2015-10-24T01:09:26Z stassats: there are still some optimizations not enabled, some modular stuff, but that shouldn't impact things greatly
2015-10-24T01:12:00Z stassats: and one instruction CODE setup is much easier to read
2015-10-24T01:15:42Z stassats: i think i still don't fully understand how LRA works
2015-10-24T01:17:09Z nyef: Oh?
2015-10-24T01:17:28Z nyef: The register, or the objects with RETURN_PC_HEADER_WIDETAG ?
2015-10-24T01:17:39Z stassats: return-pc
2015-10-24T01:17:57Z nyef: What's not to understand?
2015-10-24T01:18:22Z stassats: all the "why isn't it done another way" questions
2015-10-24T01:19:21Z nyef: They are aligned, they have a widetag, thus they can be shown to the GC and not have it pitch a fit. Their header value points to the start of the code object, thus the GC can find the code object and in turn deal with relocating references and whatnot.
2015-10-24T01:27:19Z stassats: i think my other ideas depend on precise backtraces
2015-10-24T01:27:45Z stassats: something like saving the return address in special stack location
2015-10-24T01:28:25Z nyef: ... The return address would at least be aligned, right?
2015-10-24T01:28:50Z stassats: half-world
2015-10-24T01:28:55Z stassats: but a fixnum, yes
2015-10-24T01:29:20Z stassats: before the new CODE is setup it would be paired with CODE
2015-10-24T01:29:40Z stassats: now, if there were an interior control stack location
2015-10-24T01:31:32Z stassats: presumably precisish backtraces are easier on (not c-stack-is-control-stack)
2015-10-24T01:39:27Z dougk2: (constant-form-value '(if t 1 ((any (random (garbage (u) (want))))))) => T
2015-10-24T01:39:30Z dougk2: is that allowed ?
2015-10-24T01:39:41Z dougk2: sorry, i meant => 1
2015-10-24T01:40:44Z stassats: clhs constantp
2015-10-24T01:40:44Z specbot: http://www.lispworks.com/reference/HyperSpec/Body/f_consta.htm
2015-10-24T01:43:00Z dougk2: ok, (constantp '(if 1 2 3 4 5 6)) => signals error.
2015-10-24T01:43:06Z dougk2: is THAT allowed? spec says no exceptional conditions
2015-10-24T01:44:05Z stassats: Exceptional Situations is usually quite inconsistent
2015-10-24T01:46:31Z psy_ joined #sbcl
2015-10-24T01:47:03Z psy_ quit (Max SendQ exceeded)
2015-10-24T01:47:24Z psy_ joined #sbcl
2015-10-24T02:26:19Z whiteline_ quit (Ping timeout: 250 seconds)
2015-10-24T02:26:55Z whiteline quit (Ping timeout: 268 seconds)
2015-10-24T02:34:52Z psy_ quit (Ping timeout: 250 seconds)
2015-10-24T03:05:47Z mood quit (Ping timeout: 265 seconds)
2015-10-24T03:05:54Z mood joined #sbcl
2015-10-24T03:26:31Z ccl-logbot joined #sbcl
2015-10-24T03:26:31Z 
2015-10-24T03:26:31Z names: ccl-logbot mood karswell dougk2 jdz hydan fridim_ ASau chris2 jackdaniel stassats drmeister loke p_l pchrist psilord1 pkhuong jackc- sigjuice heddwch zacts fikusz dougk_ Quadrescence brucem foom frankS2 PuercoPop White_Flame scymtym__ nzambe nyef abbe angavrilov reb |3b| hydan` antoszka DeadTrickster salva yauz DGASAU kanru jlarocco_ ym Bike specbot minion alchemis7 christoph_debian pootler jozip hzp Ober ferada joshe fe[nl]ix Blkt flip214 carvite
2015-10-24T03:26:31Z names: jsnell dsp_ irsol jlarocco dustinm` les` dim
2015-10-24T03:35:45Z nyef: "Mathematically impeccable, computationally inexcusable"?
2015-10-24T04:15:20Z jlarocco_ quit (Remote host closed the connection)
2015-10-24T04:19:03Z jlarocco_ joined #sbcl
2015-10-24T04:26:22Z stassats: i liked *compile-print*
2015-10-24T04:27:01Z rszeno joined #sbcl
2015-10-24T04:58:22Z stassats: found a way to save 672KB on x86-64
2015-10-24T04:59:07Z stassats: by removing the compiled-debug-fun-nfp slot
2015-10-24T05:02:53Z stassats: and 572KB on ARM64
2015-10-24T05:11:48Z stassats: see an opportunity for some more compaction
2015-10-24T05:20:19Z stassats: sc-offset is 27 bits
2015-10-24T05:20:39Z stassats: i can pack two offsets and the 8-bit flags into a single fixnum on #+64-bit
2015-10-24T05:22:16Z stassats: that saves a further 672KB, for a total of 1344KB
2015-10-24T05:22:20Z stassats: now, that's not bad
2015-10-24T05:32:06Z stassats: isn't as dramatic on ARM64
2015-10-24T05:33:30Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-24T05:36:35Z dougk2 joined #sbcl
2015-10-24T05:42:57Z dougk2 quit (Ping timeout: 256 seconds)
2015-10-24T06:09:22Z zacts quit (Read error: Connection reset by peer)
2015-10-24T06:11:12Z zacts joined #sbcl
2015-10-24T06:21:00Z whiteline joined #sbcl
2015-10-24T06:23:32Z whiteline_ joined #sbcl
2015-10-24T06:32:18Z stassats quit (Ping timeout: 260 seconds)
2015-10-24T06:52:14Z fridim_ quit (Ping timeout: 260 seconds)
2015-10-24T07:00:11Z gingerale joined #sbcl
2015-10-24T07:09:46Z nikki93 joined #sbcl
2015-10-24T07:36:05Z rszeno quit (Quit: Leaving.)
2015-10-24T07:46:24Z nikki93 quit (Remote host closed the connection)
2015-10-24T07:47:05Z Shinmera joined #sbcl
2015-10-24T07:53:07Z nikki93 joined #sbcl
2015-10-24T08:05:37Z nikki93 quit (Remote host closed the connection)
2015-10-24T08:21:45Z fridim_ joined #sbcl
2015-10-24T08:30:03Z jlarocco_ quit (Remote host closed the connection)
2015-10-24T08:35:48Z jlarocco_ joined #sbcl
2015-10-24T08:45:10Z Posterdati joined #sbcl
2015-10-24T09:44:19Z clique joined #sbcl
2015-10-24T09:45:33Z clique left #sbcl
2015-10-24T09:47:39Z S161No3 joined #sbcl
2015-10-24T09:53:40Z attila_lendvai joined #sbcl
2015-10-24T09:53:40Z attila_lendvai quit (Changing host)
2015-10-24T09:53:40Z attila_lendvai joined #sbcl
2015-10-24T09:53:47Z S161No3 left #sbcl
2015-10-24T10:13:15Z scymtym joined #sbcl
2015-10-24T10:15:00Z edgar-rft joined #sbcl
2015-10-24T10:23:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-24T11:08:31Z psy_ joined #sbcl
2015-10-24T11:09:08Z psy_ quit (Max SendQ exceeded)
2015-10-24T11:09:41Z psy_ joined #sbcl
2015-10-24T12:26:16Z hydan quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2015-10-24T12:32:05Z hydan joined #sbcl
2015-10-24T12:35:52Z stassats joined #sbcl
2015-10-24T12:40:43Z scymtym quit (Ping timeout: 252 seconds)
2015-10-24T13:05:45Z dougk2 joined #sbcl
2015-10-24T13:07:16Z hydan quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2015-10-24T13:13:57Z hydan joined #sbcl
2015-10-24T13:47:33Z _iwc joined #sbcl
2015-10-24T13:49:10Z nyef: ... Stepping through my little test program shows calling pow() in C winds up in powl() in libm.
2015-10-24T13:53:51Z nyef: ... Might be explained by NO_LONG_DOUBLE...?
2015-10-24T13:54:41Z scymtym__: dougk2: the changes to tests/* seem to arrange for tests to mostly work when *EVALUATOR-MODE* already is :INTERPRET. has there also been a change to automatically run tests in that mode? i'm asking because merely building with sb-fasteval and without sb-eval seems to cause that. see https://ci.cor-lab.org/job/sbcl-master/featureset=8,label=ubuntu_trusty_64bit/1261/consoleFull
2015-10-24T13:55:59Z fridim_ quit (Ping timeout: 264 seconds)
2015-10-24T13:58:30Z dougk2: scymtym__: the tests run as interpreted (except ones that call COMPILE) when sb-fasteval is present. Ideally they'd all run both ways.
2015-10-24T13:58:32Z nyef: Okay, NO_LONG_DOUBLE causes an alias of pow() to powl().
2015-10-24T13:59:35Z scymtym__: dougk2: that's what i though. thanks for confirming. should i continue to automatically run the tests that way or would there be too many expected errors?
2015-10-24T14:00:09Z _iwc quit (Killed (Sigyn (Spam is off topic on freenode.)))
2015-10-24T14:00:21Z dougk2: scymtym__: it would premature to suppose that the testing in the interpreter is reliable.  Too many things are #-'ed out.
2015-10-24T14:00:46Z dougk2: so I suppose what we should do is one of two things: either not default the REPL to :interpret, or change it for the test run.
2015-10-24T14:03:51Z scymtym__:  maybe even one those two to ensure that building with sb-fasteval doesn't break the rest of the system and then a separate run in :interpret mode to track sb-fasteval progress?
2015-10-24T14:04:06Z scymtym__: s/one those/one of those/
2015-10-24T14:07:50Z nyef: ... There's an "FPR" bit in CP0 SR that affects the layout of the FPU registers. Lovely.
2015-10-24T14:07:53Z dougk2: i could make the test-driver process each file twice.
2015-10-24T14:08:33Z dougk2: As a separate issue, some tests on an ad-hoc basis decide to call compile and eval on the same thing to try it both ways. That's not a god way to got coverage, it's just random and ugly.
2015-10-24T14:11:19Z nyef: So, if the code is compiled for sixteen registers, does it matter if there actually are 32 registers?
2015-10-24T14:13:52Z nyef: ... And I can't quite tell.
2015-10-24T14:15:43Z scymtym__: dougk2: i agree re. eval vs. compile issue, but wouldn't it be easier to keep the test driver mostly unchanged and run it twice? for our hacked version of the test driver (for jenkins integration), that would mean producing one result file for each mode which would be presented as two independent reports of the same "shape" as reports for builds of other feature sets.
2015-10-24T14:17:54Z dougk2: scymtym__: yes, two independent reports would be great.
2015-10-24T14:18:54Z scymtym__: dougk2: ok, i will try to implement that
2015-10-24T14:20:31Z stassats: hm, sbcl keeps not getting the desired mmaping on arm64
2015-10-24T14:20:42Z stassats: not always, but often enough to be annoying
2015-10-24T14:20:47Z stassats: maybe that's just the android kernel
2015-10-24T14:29:05Z nyef: Hrm... I'm not sure, but I think that this code might have been compiled for 16-register mode.
2015-10-24T14:30:47Z stassats: doesn't that use 16-bit instructions?
2015-10-24T14:31:04Z nyef: All instructions are 32 bits wide.
2015-10-24T14:31:19Z nyef: I'm on MIPS, not ARM.
2015-10-24T14:31:41Z nyef: Old-school MIPS, at that.
2015-10-24T14:32:04Z stassats: https://en.wikipedia.org/wiki/MIPS_instruction_set#MIPS16e ?
2015-10-24T14:32:52Z stassats: but there's only 8 registers available
2015-10-24T14:33:39Z nyef: Floating-point registers, not general registers.
2015-10-24T14:33:55Z stassats: and you are probably talking about... ^that
2015-10-24T14:36:19Z stassats: arm64 has an UBFM instruction which is basically LDB
2015-10-24T14:36:26Z stassats: now, how to channel LDB into that
2015-10-24T14:36:47Z nyef: Same hack as on PPC for rlwinm?
2015-10-24T14:37:41Z stassats: looks like
2015-10-24T14:38:14Z stassats: there's also EXTR, which may be some sort of DPB
2015-10-24T14:39:05Z stassats: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0802a/EXTR.html
2015-10-24T14:45:02Z stassats: so it's basically (subseq (concatenate bits-1 bits-2) lsb)
2015-10-24T14:45:18Z stassats: not much room for DPb
2015-10-24T14:45:58Z stassats: if any
2015-10-24T15:17:48Z DeadTrickster quit (Read error: No route to host)
2015-10-24T15:26:42Z scymtym__: dougk2: initial report for sb-fasteval on x86_64: https://ci.cor-lab.org/job/sbcl-master/featureset=8,label=ubuntu_trusty_64bit/1262/testReport/
2015-10-24T15:28:22Z dougk2: is the Jenkins test driver pushing :INTERPRETER onto *features*?  A lot of those are marked as :skipped-on :interpreter
2015-10-24T15:29:25Z dougk2: scymtym__: for example (with-test (:name :delete-to-delq-opt :skipped-on :interpreter) should have been skipped
2015-10-24T15:29:49Z scymtym__: i may have missed some changes when updating the hacked test driver
2015-10-24T15:29:59Z scymtym__: i will have another look
2015-10-24T15:30:41Z DeadTrickster joined #sbcl
2015-10-24T16:12:41Z loke quit (Ping timeout: 250 seconds)
2015-10-24T16:24:47Z chris2 quit (Ping timeout: 264 seconds)
2015-10-24T16:25:41Z attila_lendvai joined #sbcl
2015-10-24T16:25:41Z attila_lendvai quit (Changing host)
2015-10-24T16:25:41Z attila_lendvai joined #sbcl
2015-10-24T16:30:20Z nyef sighs.
2015-10-24T16:30:32Z nyef: Install debian chroot, run test program in debain chroot, results as expected.
2015-10-24T16:32:10Z nyef: So I'm looking at it _probably_ being the libm still.
2015-10-24T16:33:55Z Posterdati quit (Ping timeout: 265 seconds)
2015-10-24T16:36:48Z stassats: can you load it via LD_LIBRARY_PATH?
2015-10-24T16:37:29Z nyef: What would that tell me?
2015-10-24T16:38:41Z stassats: you can just use it that way
2015-10-24T16:39:03Z nyef: Hrm. That might get me a build, at least, but doesn't fix my host environment.
2015-10-24T16:45:24Z nyef: I wonder if I'm looking at the effects of a cross-build compiler bug?
2015-10-24T16:46:41Z Posterdati joined #sbcl
2015-10-24T16:53:22Z chris2 joined #sbcl
2015-10-24T17:11:06Z hydan quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2015-10-24T17:11:23Z hydan joined #sbcl
2015-10-24T17:27:11Z dougk2 quit (Ping timeout: 264 seconds)
2015-10-24T17:27:48Z stassats: ok, done with LDB=>UBFM
2015-10-24T17:27:57Z stassats: now, DPB should be done with BFM
2015-10-24T17:28:34Z nyef: Better Fixnum Math ?
2015-10-24T17:28:50Z stassats: bitflield move
2015-10-24T17:29:40Z nyef: Ooh. Build survived to genesis.
2015-10-24T17:29:44Z stassats: UBFM is unsigned bitfield move, it zeros the remaining bits
2015-10-24T17:29:49Z stassats: SBFM sign extends
2015-10-24T17:30:00Z stassats: and BFM keeps them
2015-10-24T17:34:16Z heddwch is now known as pickled_pigs_foo
2015-10-24T17:34:28Z pickled_pigs_foo is now known as heddwch
2015-10-24T17:44:07Z nyef: Two contribs failed to build, sb-gmp and sb-sprof.
2015-10-24T17:44:47Z nyef: ... This is not a fast build host.
2015-10-24T18:23:25Z dougk2 joined #sbcl
2015-10-24T18:27:55Z fridim_ joined #sbcl
2015-10-24T18:31:30Z clique joined #sbcl
2015-10-24T18:35:57Z scymtym__: dougk2: test run with compile and interpret mode: https://ci.cor-lab.org/job/sbcl-master/1266/featureset=8,label=ubuntu_trusty_64bit/consoleFull. results with both reports: https://ci.cor-lab.org/job/sbcl-master/1266/featureset=8,label=ubuntu_trusty_64bit/testReport/. does that look more plausible?
2015-10-24T18:36:15Z stassats: nyef: sb-sprof fails here too
2015-10-24T18:36:21Z stassats: that may be due to the slowness
2015-10-24T18:36:46Z nyef: That's fair. 800 MHz just isn't that fast these days.
2015-10-24T18:37:21Z dougk2: scymtym__: yup
2015-10-24T18:42:10Z scymtym__: dougk2: great
2015-10-24T18:42:46Z stassats: nyef: depends on the hertz
2015-10-24T18:43:42Z stassats: 800MHz might be not so bad for an out of order superscalar
2015-10-24T18:45:53Z nyef: That's what I'm hoping, yes.
2015-10-24T18:45:55Z pkhuong: and a modern memory bus
2015-10-24T18:46:07Z nyef: I don't know about that latter bit.
2015-10-24T18:46:34Z nyef: Might only be DDR.
2015-10-24T18:46:38Z nyef: Might not even be DDR.
2015-10-24T18:46:46Z stassats: should beat qemu
2015-10-24T18:46:56Z stassats: which is about 150 slower than the host
2015-10-24T18:47:11Z nyef: Of course, I still have to get the secondary CPUs online on this thing.
2015-10-24T18:47:33Z stassats: at 4300MHz, that's about 30MHz
2015-10-24T18:48:01Z nyef: ... No wonder it takes all afternoon for me to do a MIPS build on QEMU.
2015-10-24T18:48:15Z nyef: Makes a 68k port more viable-sounding, though. (-:
2015-10-24T18:49:24Z stassats: it took me 5 hours 10 minutes
2015-10-24T18:50:09Z stassats: i wonder if output redirection would help, since it's a single core "machine"
2015-10-24T18:51:20Z scymtym__: dougk2: foreign.test.sh seems to run in :interpret mode in both cases. would it make sense to expose the evaluator mode of the current test run in an environment variable and have shell scripts pass it to sbcl subprocesses for such cases?
2015-10-24T18:51:45Z dougk2: scymtym__: oh, that's why it fails in both. huh.
2015-10-24T18:51:54Z dougk2: thanks for figuring it out
2015-10-24T18:51:59Z clique left #sbcl
2015-10-24T18:52:09Z dougk2: i hadn't the time to look into it.
2015-10-24T18:52:22Z dougk2: so, whatever you think is the right solution is the right solution.
2015-10-24T18:53:25Z scymtym__: ok, i will try that after the freeze
2015-10-24T18:55:47Z dougk2: well, if you're feeling ambitious, the test failure itself is something to tackle as well. I hadn't a clue.
2015-10-24T18:58:08Z scymtym__: i should probably finish my fix for https://bugs.launchpad.net/sbcl/+bug/1498644 first
2015-10-24T19:07:14Z stassats: DPB is quite more compact when done using BFM
2015-10-24T19:07:22Z stassats: more of a win compared to LDB to UBFM
2015-10-24T19:26:30Z attila_lendvai quit (Quit: Leaving.)
2015-10-24T19:26:32Z attila_lendvai1 joined #sbcl
2015-10-24T19:26:32Z attila_lendvai1 is now known as attila_lendvai
2015-10-24T19:26:32Z attila_lendvai quit (Changing host)
2015-10-24T19:26:32Z attila_lendvai joined #sbcl
2015-10-24T19:33:45Z whiteline quit (Read error: Connection reset by peer)
2015-10-24T19:34:06Z whiteline_ quit (Quit: Leaving)
2015-10-24T19:34:27Z whiteline joined #sbcl
2015-10-24T19:44:48Z nyef: Today I am grateful for... hot-swap SCSI drives.
2015-10-24T20:22:09Z scymtym joined #sbcl
2015-10-24T20:37:35Z prxq joined #sbcl
2015-10-24T20:45:24Z duikboot joined #sbcl
2015-10-24T21:05:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-24T21:14:03Z duikboot quit (Ping timeout: 252 seconds)
2015-10-24T21:14:05Z stassats: argh, :load-if is driving me crazy
2015-10-24T21:14:31Z stassats: trying to load a constant only it has a certain value, but in the generator tn-value is NIL, not the immediate in question
2015-10-24T21:15:02Z stassats: i don't want to be loading all immediates manually
2015-10-24T21:16:14Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-24T21:20:28Z duikboot joined #sbcl
2015-10-24T21:20:37Z prxq quit (Remote host closed the connection)
2015-10-24T21:27:00Z stassats: i can't even load things manually
2015-10-24T21:27:15Z duikboot quit (Ping timeout: 252 seconds)
2015-10-24T21:30:53Z stassats: for some reason sc-case doesn't catch an immediate case
2015-10-24T21:31:56Z stassats: and the sc is #<SC :NAME IMMEDIATE>
2015-10-24T21:31:59Z stassats: what the hell
2015-10-24T21:33:20Z stassats: oh ha, i'm modifying the IF-EQ VOP
2015-10-24T21:33:28Z stassats: so on the second recompilation it breaks itself
2015-10-24T21:43:07Z dougk2 joined #sbcl
2015-10-24T21:58:59Z psy_ quit (Remote host closed the connection)
2015-10-24T22:06:42Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-24T22:17:50Z edgar-rft quit (Quit: edgar-rft)
2015-10-24T22:19:18Z fridim_ quit (Ping timeout: 255 seconds)
2015-10-24T22:26:09Z stassats: trying to optimize loading immediates, i guess i need to look up what clang does
2015-10-24T22:31:11Z gingerale quit (Remote host closed the connection)
2015-10-24T22:33:10Z dougk2 joined #sbcl
2015-10-24T22:44:42Z phoe_krk_ joined #sbcl
2015-10-24T22:57:22Z stassats: looks like arm a57 does uop fusion for MOVZ/MOVK, so any fancy encoding is not likely to be faster, just more compact
2015-10-24T23:25:31Z scymtym: shouldn't line 39 of tests/full-eval.impure.lisp say sb-vm:n-word-*bits*? with sb-fasteval, the test fails on x86 because 42 is not of type (unsigned-byte 4)
2015-10-24T23:32:20Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-24T23:35:49Z stassats: at that rate, it should be sb-ext:word
2015-10-24T23:53:56Z scymtym: thanks, will commit after the freeze
2015-10-25T00:00:39Z stassats: scymtym: it's a test, commit now
2015-10-25T00:01:03Z scymtym: stassats: ok
2015-10-25T00:03:22Z revolve joined #sbcl
2015-10-25T00:09:29Z stassats: how badly do we need .cvsignore ?
2015-10-25T00:10:55Z akkad joined #sbcl
2015-10-25T00:13:10Z scymtym: i wanted to delete them for some time, but i didn't want to bother everybody with asking whether we need them :)
2015-10-25T00:24:41Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-25T00:50:46Z scymtym quit (Ping timeout: 240 seconds)
2015-10-25T00:55:41Z stassats quit (Ping timeout: 244 seconds)
2015-10-25T01:46:27Z dougk2 joined #sbcl
2015-10-25T01:58:52Z dougk2 quit (Ping timeout: 250 seconds)
2015-10-25T02:21:53Z dougk2 joined #sbcl
2015-10-25T03:28:33Z dougk2 quit (Ping timeout: 256 seconds)
2015-10-25T03:54:29Z dougk2 joined #sbcl
2015-10-25T04:13:27Z akkad: the latest package for netbsd, 1.2.7, generates this error.   Don't know how to REQUIRE SB-POSIX.
2015-10-25T04:13:28Z akkad:  
2015-10-25T04:15:35Z dougk2 quit (Ping timeout: 256 seconds)
2015-10-25T04:23:30Z akkad: got it working.
2015-10-25T04:41:50Z dougk2 joined #sbcl
2015-10-25T05:38:12Z phoe_krk joined #sbcl
2015-10-25T05:40:47Z phoe_krk_ quit (Ping timeout: 265 seconds)
2015-10-25T05:44:00Z Quadrescence: Does SBCL write out all of the compilation output somewhere?
2015-10-25T05:45:52Z psy_ joined #sbcl
2015-10-25T05:57:11Z Quadrescence: (During compilation of SBCL*.)
2015-10-25T06:12:31Z karswell` joined #sbcl
2015-10-25T06:12:32Z karswell quit (Read error: Connection reset by peer)
2015-10-25T07:02:49Z karswell` quit (Read error: Connection reset by peer)
2015-10-25T07:03:03Z karswell` joined #sbcl
2015-10-25T07:06:33Z karswell` quit (Remote host closed the connection)
2015-10-25T07:06:48Z karswell` joined #sbcl
2015-10-25T07:16:43Z Shinmera joined #sbcl
2015-10-25T07:51:43Z zacts quit (Ping timeout: 250 seconds)
2015-10-25T07:53:14Z gingerale joined #sbcl
2015-10-25T08:40:36Z phoe_krk_ joined #sbcl
2015-10-25T08:43:16Z phoe_krk quit (Ping timeout: 244 seconds)
2015-10-25T09:09:18Z scymtym joined #sbcl
2015-10-25T09:17:00Z karswell` quit (Remote host closed the connection)
2015-10-25T09:48:39Z nikki93 joined #sbcl
2015-10-25T09:57:26Z nikki93 quit (Remote host closed the connection)
2015-10-25T10:01:35Z edgar-rft joined #sbcl
2015-10-25T10:06:51Z fridim_ joined #sbcl
2015-10-25T10:19:14Z stassats joined #sbcl
2015-10-25T10:21:35Z duikboot joined #sbcl
2015-10-25T10:21:48Z duikboot quit (Remote host closed the connection)
2015-10-25T10:22:05Z duikboot joined #sbcl
2015-10-25T10:28:22Z nikki93 joined #sbcl
2015-10-25T10:28:55Z nikki93 quit (Remote host closed the connection)
2015-10-25T10:43:17Z duikboot quit (Ping timeout: 252 seconds)
2015-10-25T12:02:28Z clique joined #sbcl
2015-10-25T12:07:55Z clique left #sbcl
2015-10-25T12:54:57Z Mark- joined #sbcl
2015-10-25T12:55:02Z Mark- left #sbcl
2015-10-25T13:01:39Z Xof joined #sbcl
2015-10-25T13:03:53Z clique joined #sbcl
2015-10-25T13:13:29Z clique quit (Quit: Page closed)
2015-10-25T13:22:47Z stassats quit (Ping timeout: 250 seconds)
2015-10-25T13:37:25Z hydan quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2015-10-25T14:01:23Z nikki93 joined #sbcl
2015-10-25T14:22:39Z stassats joined #sbcl
2015-10-25T14:52:47Z hydan joined #sbcl
2015-10-25T15:02:02Z nikki93_ joined #sbcl
2015-10-25T15:06:11Z nikki93 quit (Ping timeout: 264 seconds)
2015-10-25T15:21:45Z heddwch quit (Ping timeout: 265 seconds)
2015-10-25T15:24:38Z heddwch joined #sbcl
2015-10-25T15:51:24Z nikki93_ quit (Remote host closed the connection)
2015-10-25T15:51:43Z heddwch quit (Ping timeout: 265 seconds)
2015-10-25T15:52:53Z heddwch joined #sbcl
2015-10-25T15:58:56Z nzambe quit (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
2015-10-25T16:18:35Z stassats quit (Remote host closed the connection)
2015-10-25T16:19:44Z stassats joined #sbcl
2015-10-25T16:22:27Z dougk2 quit (Ping timeout: 268 seconds)
2015-10-25T16:23:28Z dougk2 joined #sbcl
2015-10-25T16:25:22Z attila_lendvai joined #sbcl
2015-10-25T16:25:22Z attila_lendvai quit (Changing host)
2015-10-25T16:25:22Z attila_lendvai joined #sbcl
2015-10-25T16:28:46Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-25T16:38:39Z dougk2 joined #sbcl
2015-10-25T16:42:37Z ASau` joined #sbcl
2015-10-25T16:46:20Z ASau quit (Ping timeout: 272 seconds)
2015-10-25T16:50:18Z dougk2 quit (Ping timeout: 260 seconds)
2015-10-25T17:13:01Z dougk2 joined #sbcl
2015-10-25T17:16:42Z ASau` is now known as ASau
2015-10-25T17:37:18Z dougk2 quit (Ping timeout: 260 seconds)
2015-10-25T17:47:29Z ASau quit (Changing host)
2015-10-25T17:47:30Z ASau joined #sbcl
2015-10-25T18:04:04Z dougk2 joined #sbcl
2015-10-25T18:14:16Z edgar-rft quit (Quit: edgar-rft)
2015-10-25T18:39:57Z dougk2 quit (Ping timeout: 252 seconds)
2015-10-25T18:44:51Z karswell` joined #sbcl
2015-10-25T18:47:40Z Posterdati quit (Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/)
2015-10-25T18:48:02Z Posterdati joined #sbcl
2015-10-25T18:49:12Z dougk2 joined #sbcl
2015-10-25T19:07:19Z Posterdati quit (Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/)
2015-10-25T19:09:27Z Posterdati joined #sbcl
2015-10-25T19:09:38Z karswell` quit (Read error: Connection reset by peer)
2015-10-25T19:09:54Z karswell` joined #sbcl
2015-10-25T19:16:56Z jlarocco_ quit (Quit: Leaving)
2015-10-25T19:33:49Z karswell` quit (Read error: Connection reset by peer)
2015-10-25T19:35:17Z duikboot joined #sbcl
2015-10-25T19:51:25Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-25T19:59:31Z psy_ quit (Ping timeout: 268 seconds)
2015-10-25T20:02:59Z dougk2 joined #sbcl
2015-10-25T20:04:16Z psy_ joined #sbcl
2015-10-25T20:04:49Z psy_ quit (Max SendQ exceeded)
2015-10-25T20:05:42Z psy_ joined #sbcl
2015-10-25T20:09:41Z psy_ quit (Read error: No route to host)
2015-10-25T20:18:55Z psy_ joined #sbcl
2015-10-25T20:21:32Z prxq joined #sbcl
2015-10-25T20:41:19Z duikboot quit (Ping timeout: 252 seconds)
2015-10-25T20:46:12Z duikboot joined #sbcl
2015-10-25T21:01:51Z duikboot quit (Ping timeout: 252 seconds)
2015-10-25T21:04:38Z antoszka is now known as jowtoszka
2015-10-25T21:05:09Z phoe_krk_ quit (Quit: phoe_krk_)
2015-10-25T21:06:11Z phoe_krk joined #sbcl
2015-10-25T21:11:10Z jowtoszka is now known as antoszka
2015-10-25T21:30:26Z nikki93 joined #sbcl
2015-10-25T21:37:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-25T21:38:13Z nyef: Bleh. I thought I had some unused PATA desktop drives kicking around, but I can't seem to find any. /-:
2015-10-25T21:49:40Z prxq quit (Remote host closed the connection)
2015-10-25T21:54:33Z fridim_ quit (Ping timeout: 250 seconds)
2015-10-25T21:59:58Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-25T22:01:45Z dougk2 joined #sbcl
2015-10-25T22:25:05Z gingerale quit (Remote host closed the connection)
2015-10-25T22:28:50Z stassats: what would it take to get rid of CODE altogether?
2015-10-25T22:29:27Z stassats: make LEXENV always leave and figure PC from that?
2015-10-25T22:29:31Z stassats: live
2015-10-25T22:30:26Z stassats: assuming CODE is only needed for avoiding PC from floating away
2015-10-25T22:31:01Z stassats: constants can be accessed via PC-relative loads
2015-10-25T22:31:19Z nyef: Alter scavenge_interrupt_context() to use search_dynamic_space() to find the enclosing object whenever the program counter is in dynamic space.
2015-10-25T22:31:44Z nyef: And then turn around and also try to use that to base any other registers with the interior-pointer nature.
2015-10-25T22:31:47Z stassats: assuming it's slow (does it matter?)
2015-10-25T22:32:19Z stassats: we load LEXENV anyhow, though that leaves LEXENV being set up to the new function
2015-10-25T22:32:27Z nyef: Actually, not just program counter, you'd also need to cover NPC, LINK, and CTR.
2015-10-25T22:33:24Z stassats: CODE is determined from PC, then everything paired with that
2015-10-25T22:34:02Z stassats: if that can be avoided, and LRA somehow magically removed, the call overhead could be really reduced
2015-10-25T22:34:46Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-25T22:34:47Z nyef: I bet that you want to do that across the board, don't you?
2015-10-25T22:35:11Z stassats: not really, can't test everywhere
2015-10-25T22:35:37Z nyef: Neither can I, but between the two of us we should be able to get decent coverage.
2015-10-25T22:36:02Z stassats: what about PC-relative loads?
2015-10-25T22:36:04Z stassats: that's not universal
2015-10-25T22:36:26Z nyef: Right, it's not.
2015-10-25T22:36:33Z stassats: and i have no good solution for LRA
2015-10-25T22:36:54Z nyef: At least, not without precise backtrace?
2015-10-25T22:39:21Z dougk2 joined #sbcl
2015-10-25T22:39:24Z nyef: Hrm... alpha, arm, arm64, hppa, mips, ppc, sparc, x86, x86-64.
2015-10-25T22:39:55Z nyef: Nine arches. Of which I can cover seven, if necessary.
2015-10-25T22:42:22Z nyef: (Well, maybe seven. Might only be five, depending on how well some of my hardware works.)
2015-10-25T22:45:46Z stassats: at the very least, this has to be addresses before any work on getting x86-64 to have a precise gc
2015-10-25T22:45:53Z stassats: don't want to make it slower just for some precision
2015-10-25T22:59:26Z heddwch is now known as sammwch
2015-10-25T22:59:31Z sammwch is now known as heddwch
2015-10-25T23:02:23Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-25T23:16:54Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-25T23:18:38Z Posterdati quit (Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/)
2015-10-25T23:18:50Z Posterdati joined #sbcl
2015-10-25T23:32:10Z whiteline quit (Quit: Leaving)
2015-10-25T23:33:29Z dougk2 joined #sbcl
2015-10-26T00:39:58Z stassats quit (Remote host closed the connection)
2015-10-26T01:00:10Z hydan quit (Quit: Textual IRC Client: www.textualapp.com)
2015-10-26T01:00:31Z hydan joined #sbcl
2015-10-26T01:05:41Z nikki93_ joined #sbcl
2015-10-26T01:05:48Z nikki93 quit (Ping timeout: 255 seconds)
2015-10-26T01:14:06Z scymtym quit (Ping timeout: 240 seconds)
2015-10-26T01:15:20Z hydan quit (Quit: Textual IRC Client: www.textualapp.com)
2015-10-26T01:19:09Z stassats joined #sbcl
2015-10-26T01:33:18Z nikki93_ quit (Remote host closed the connection)
2015-10-26T01:35:16Z nikki93 joined #sbcl
2015-10-26T01:49:40Z heddwch is now known as sammwch
2015-10-26T01:49:51Z sammwch is now known as heddwch
2015-10-26T01:49:58Z heddwch is now known as porkmonger
2015-10-26T01:50:35Z porkmonger is now known as heddwch
2015-10-26T01:59:03Z heddwch is now known as abcdefghijabcdef
2015-10-26T01:59:17Z abcdefghijabcdef is now known as heddwch
2015-10-26T02:03:34Z nikki93 quit (Remote host closed the connection)
2015-10-26T02:22:23Z stassats quit (Ping timeout: 264 seconds)
2015-10-26T02:27:16Z loke joined #sbcl
2015-10-26T02:34:19Z psy_ quit (Ping timeout: 244 seconds)
2015-10-26T02:43:33Z nikki93 joined #sbcl
2015-10-26T02:50:48Z nikki93 quit (Remote host closed the connection)
2015-10-26T03:18:22Z nicdev joined #sbcl
2015-10-26T03:48:40Z zacts joined #sbcl
2015-10-26T04:44:34Z psy_ joined #sbcl
2015-10-26T04:45:00Z psy_ quit (Max SendQ exceeded)
2015-10-26T04:46:01Z psy_ joined #sbcl
2015-10-26T05:07:17Z heddwch quit (Ping timeout: 265 seconds)
2015-10-26T05:16:21Z heddwch joined #sbcl
2015-10-26T05:50:12Z psy_ quit (Ping timeout: 255 seconds)
2015-10-26T06:07:30Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-26T06:18:27Z ASau quit (Ping timeout: 260 seconds)
2015-10-26T06:22:46Z dougk2 joined #sbcl
2015-10-26T06:31:55Z igajsin joined #sbcl
2015-10-26T07:14:51Z Shinmera joined #sbcl
2015-10-26T07:20:25Z gingerale joined #sbcl
2015-10-26T07:38:03Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-26T07:43:36Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-26T07:57:39Z duikboot joined #sbcl
2015-10-26T08:10:09Z dougk2 joined #sbcl
2015-10-26T08:14:56Z duikboot quit (Quit: Leaving)
2015-10-26T08:20:25Z ZabaQ joined #sbcl
2015-10-26T08:24:03Z Guest51183 joined #sbcl
2015-10-26T08:28:21Z dougk2 quit (Ping timeout: 265 seconds)
2015-10-26T08:55:11Z dougk2 joined #sbcl
2015-10-26T09:04:08Z fridim_ joined #sbcl
2015-10-26T09:26:27Z scymtym__ quit (Read error: Connection reset by peer)
2015-10-26T09:27:33Z scymtym joined #sbcl
2015-10-26T10:00:27Z heddwch quit (Read error: Connection reset by peer)
2015-10-26T10:01:53Z rszeno joined #sbcl
2015-10-26T10:04:42Z heddwch joined #sbcl
2015-10-26T10:19:29Z sjl joined #sbcl
2015-10-26T11:15:54Z stassats joined #sbcl
2015-10-26T11:34:07Z ZabaQ_ joined #sbcl
2015-10-26T11:37:46Z ZabaQ quit (Ping timeout: 240 seconds)
2015-10-26T11:48:26Z sjl quit (Ping timeout: 240 seconds)
2015-10-26T11:54:58Z phoe_krk quit (Ping timeout: 246 seconds)
2015-10-26T12:23:13Z sjl joined #sbcl
2015-10-26T13:00:11Z loke quit (Ping timeout: 264 seconds)
2015-10-26T13:11:40Z eudoxia joined #sbcl
2015-10-26T13:12:39Z loke joined #sbcl
2015-10-26T13:20:51Z dougk2 quit (Ping timeout: 252 seconds)
2015-10-26T13:40:02Z nzambe joined #sbcl
2015-10-26T13:52:20Z Guest51183 is now known as attila_lendvai
2015-10-26T14:02:06Z sjl quit (Ping timeout: 240 seconds)
2015-10-26T14:02:49Z heddwch quit (Ping timeout: 265 seconds)
2015-10-26T14:05:28Z heddwch joined #sbcl
2015-10-26T14:13:21Z akkad quit (Excess Flood)
2015-10-26T14:15:39Z eudoxia quit (Quit: Leaving)
2015-10-26T14:18:59Z rszeno quit (Quit: Leaving.)
2015-10-26T14:19:40Z akkad joined #sbcl
2015-10-26T14:32:37Z dougk2 joined #sbcl
2015-10-26T14:46:06Z myrkraverk joined #sbcl
2015-10-26T14:46:07Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T14:47:45Z myrkraverk: I'm writing a wrapper for some C++ code, using the sbcl specific alien interface.  Since I'm also wrapping the C++ code up in C functions, is there a way for the C function to throw lisp errors/exceptions/something, when there's a C++ exception?
2015-10-26T14:48:31Z stassats: callbacks
2015-10-26T14:48:41Z stassats: just cffi for ffi
2015-10-26T14:48:45Z stassats: just use
2015-10-26T14:50:49Z myrkraverk: Hmm, I'll take a look, thanks.
2015-10-26T14:52:56Z psy_ joined #sbcl
2015-10-26T14:53:22Z psy_ quit (Max SendQ exceeded)
2015-10-26T14:53:56Z psy_ joined #sbcl
2015-10-26T15:00:20Z sjl joined #sbcl
2015-10-26T15:03:21Z dougk2 joined #sbcl
2015-10-26T15:17:33Z myrkraverk: Callbacks with CFFI are easy enough, at least.
2015-10-26T15:17:58Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T15:18:44Z myrkraverk: Maybe some day I'll look into throwing SBCL errors with C code, but today is not it.
2015-10-26T15:33:04Z psy_ quit (Ping timeout: 250 seconds)
2015-10-26T15:38:41Z dougk2 joined #sbcl
2015-10-26T15:56:07Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T16:22:33Z dougk2 joined #sbcl
2015-10-26T16:23:25Z whiteline joined #sbcl
2015-10-26T16:56:41Z dougk2_ joined #sbcl
2015-10-26T16:58:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-26T17:06:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-26T17:08:39Z psy_ joined #sbcl
2015-10-26T17:09:04Z psy_ quit (Max SendQ exceeded)
2015-10-26T17:12:25Z dougk2_ quit (Ping timeout: 246 seconds)
2015-10-26T17:20:46Z ZabaQ_ quit (Ping timeout: 240 seconds)
2015-10-26T17:39:53Z dougk2 joined #sbcl
2015-10-26T17:56:35Z nikki93 joined #sbcl
2015-10-26T17:57:16Z nikki93 quit (Client Quit)
2015-10-26T18:00:22Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T18:08:41Z nyef: Okay, contrib build failures on sb-gmp, sb-md5, and sb-sprof. Fun and games.
2015-10-26T18:10:08Z dougk2 joined #sbcl
2015-10-26T18:14:22Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T18:14:24Z nyef: ... wha? The x86oid backends support big-endian operation?
2015-10-26T18:15:56Z stassats: do they?
2015-10-26T18:16:44Z nyef: Well, there are a couple of backend-byte-order hooks in their code.
2015-10-26T18:17:22Z dougk2 joined #sbcl
2015-10-26T18:20:15Z nyef: Heh. And sb-md5 on my MIPS? 235 out of 236 tests failed.
2015-10-26T18:20:53Z stassats: which is the lucky one?
2015-10-26T18:20:58Z Xof: oops
2015-10-26T18:21:01Z nyef: I have no idea?
2015-10-26T18:21:18Z stassats: building-contribs doesn't say?
2015-10-26T18:21:27Z nyef: No, it lists all of the FAILING tests.
2015-10-26T18:21:27Z Xof: I'm somewhat surprised that a mips build even vaguely works
2015-10-26T18:21:48Z stassats: output/building-contrib.sb-md5 that is
2015-10-26T18:22:09Z nyef: Xof: Well, my previous build failures were all due to a blown libm pow() implementation...
2015-10-26T18:24:08Z nyef: Xof: MIPS mostly-working isn't that much of a surprise, it occasionally gets some amount of maintenance attention.
2015-10-26T18:24:20Z stassats: now, alpha
2015-10-26T18:24:21Z nyef: Xof: Alpha and HPPA mostly-working would be a surprise.
2015-10-26T18:24:25Z stassats: or pa-risc
2015-10-26T18:32:34Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T18:32:36Z Quadrescence joined #sbcl
2015-10-26T18:35:41Z dougk2 joined #sbcl
2015-10-26T18:49:34Z stassats: new sparc chips coming out
2015-10-26T18:50:40Z ASau joined #sbcl
2015-10-26T18:51:30Z stassats: looking at www.infoworld.com/article/2997552/processors/oracle-just-made-its-biggest-sparc-announcement-since-buying-sun.html
2015-10-26T18:51:53Z stassats: the memory protection thing sounds like some marketing bullshit
2015-10-26T18:52:47Z stassats: or the journalist just doesn't know what he's saying
2015-10-26T18:52:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-26T18:54:21Z igajsin left #sbcl
2015-10-26T18:56:46Z nyef: Well, it looks like my DS10L works, at least somewhat.
2015-10-26T19:00:20Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-10-26T19:00:21Z stassats: looks like it's just tagged pointers
2015-10-26T19:02:41Z Bicyclidine joined #sbcl
2015-10-26T19:03:36Z stassats: i wonder if it can be abused
2015-10-26T19:05:05Z nyef: It almost certainly CAN be abused, the question is if it can be abused to good effect?
2015-10-26T19:05:57Z stassats: the details are unclear, is it in the low bits or high bits?
2015-10-26T19:06:18Z stassats: it says something about alignment, so low bits are possible
2015-10-26T19:06:51Z stassats: but how does it manage "widetags", that's unclear
2015-10-26T19:07:40Z stassats: from http://i.imgur.com/7Oec0aj.png
2015-10-26T19:08:51Z stassats: there's also http://i.imgur.com/aTISZf5.png
2015-10-26T19:09:01Z stassats: i believe arm64 has that too
2015-10-26T19:09:23Z stassats: not particularly useful
2015-10-26T19:10:54Z stassats: and i don't believe that this tagging scheme is adding anything for security, i mean, you can just fake those tags
2015-10-26T19:10:59Z stassats: a bit harder, but that's it
2015-10-26T19:12:02Z stassats: probably not so easy to get this newfangled new cpu for sbcl development
2015-10-26T19:12:09Z stassats: without it costing an arm and a leg
2015-10-26T19:13:17Z stassats: "the upper four bits of each memory pointer are used to indicate the memory version, or "color." During the process of memory allocation, this four-bit pattern is written to the memory. When this memory is accessed by any pointer, the four-bit pattern of the pointer attempting the access and the four-bit pattern of the memory being accessed are compared. If there is a match between these two patterns, the access is legal; if there is no
2015-10-26T19:13:18Z stassats: match, the memory reference error is caught immediately."
2015-10-26T19:13:22Z stassats: so, it's hightags
2015-10-26T19:13:53Z stassats: but four bits is what sbcl needs
2015-10-26T19:15:17Z stassats: so, sparc based lisp machines?
2015-10-26T19:20:07Z dougk2 joined #sbcl
2015-10-26T19:44:28Z gingerale quit (Remote host closed the connection)
2015-10-26T19:59:20Z ym quit (Remote host closed the connection)
2015-10-26T19:59:44Z nyef: ... Wha? sb-gmp failed, but I have dev-libs/gmp installed?
2015-10-26T20:00:08Z stassats: what's the failure mode?
2015-10-26T20:00:18Z nyef: Ah. Memory fault while running tests.
2015-10-26T20:00:38Z stassats: that's not good
2015-10-26T20:01:04Z stassats: have you run the tests?
2015-10-26T20:01:10Z nyef: The SBCL tests?
2015-10-26T20:01:13Z stassats: yes
2015-10-26T20:01:15Z nyef: Not yet.
2015-10-26T20:01:41Z stassats: i think it'll take 15 hours on qemu
2015-10-26T20:01:59Z nyef: Figured I'd see if any of the contrib failures were easy enough to fix first, and then I got a little side-tracked onto trying to get Linux installed on my DS10L...
2015-10-26T20:02:00Z stassats: i'm thinking of setting up user mode qemu for mips, should take the kernel out of the picture
2015-10-26T20:02:20Z nyef: Running tests now.
2015-10-26T20:03:43Z nyef: The machine is also playing NFS server, so it's not exactly at its best right now.
2015-10-26T20:04:58Z stassats: doing apt-get install libc6:mips
2015-10-26T20:06:34Z scymtym_ joined #sbcl
2015-10-26T20:10:57Z stassats: well, usermode qemu doesn't seem to work
2015-10-26T20:14:27Z nyef: Should I get you an SGI box?
2015-10-26T20:15:40Z stassats: that smells dusty
2015-10-26T20:15:53Z stassats: imagination seems to be putting out a new mips board soon
2015-10-26T20:16:59Z stassats: sb-sprof fails because of Unhandled TYPE-ERROR: The value NIL is not of type SB-KERNEL:CODE-COMPONENT.
2015-10-26T20:17:39Z stassats: speaking of get_component_from_ptr, it's not guaranteed to work, is it?
2015-10-26T20:18:46Z nyef: I don't know. It might be guaranteed to work as long as the component is in dynamic space or something like that?
2015-10-26T20:19:24Z stassats: an instruction might look like a widetag, can't it?
2015-10-26T20:19:50Z nyef: Doesn't it walk the heap forward?
2015-10-26T20:20:16Z stassats: from pc? what will it find?
2015-10-26T20:20:30Z stassats: oh, i see
2015-10-26T20:20:31Z fridim_ quit (Ping timeout: 256 seconds)
2015-10-26T20:20:36Z stassats: well, that's inefficient
2015-10-26T20:21:04Z nyef: No, from the start of the heap space. Either the start of the current dynamic space (on cheneygc) or from the whatever allocation region thing (on gencgc).
2015-10-26T20:21:13Z nyef: Hrm.
2015-10-26T20:21:39Z nyef: The test run is just sitting here after "// Running /home/nyef/sbcl-mips-2/tests/octets.pure.lisp".
2015-10-26T20:22:49Z stassats: can an object straddle two gc regions?
2015-10-26T20:25:44Z stassats: nyef: spinning?
2015-10-26T20:26:09Z stassats: same thing here, i believe
2015-10-26T20:26:42Z stassats: i may be the one to blame for this
2015-10-26T20:27:00Z stassats: i change the stepping api, and not correctly
2015-10-26T20:27:32Z nyef: An object shouldn't be able to straddle two regions, but the page table will indicate, for any given page, a point from which to start searching forward.
2015-10-26T20:27:34Z nyef: Ah!
2015-10-26T20:27:38Z nyef: It just picked up.
2015-10-26T20:28:21Z nyef: So it sits there for a while and then resumes.
2015-10-26T20:29:05Z stassats: wait, i didn't update mips at all?
2015-10-26T20:29:24Z stassats: one place too few
2015-10-26T20:29:45Z stassats: sorry about that
2015-10-26T20:29:47Z nyef: You mean you updated HPPA and Alpha?
2015-10-26T20:30:11Z stassats: i update mips too, but not comprehensively
2015-10-26T20:30:18Z nyef: Ah.
2015-10-26T20:31:29Z stassats: the fix is easy, will check it
2015-10-26T20:31:33Z prxq joined #sbcl
2015-10-26T20:31:53Z stassats: and cross check other non-tested arches for the same mistake
2015-10-26T20:35:25Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T20:36:48Z stassats: nyef: there's no stepping on alpha and the hppa stepping check was broken in the first place
2015-10-26T20:37:34Z nyef: Okay then.
2015-10-26T20:38:09Z nyef: HPPA is going to be tons of fun to sort out, I'm sure.
2015-10-26T20:38:26Z stassats: well, not any harder than a new port
2015-10-26T20:39:11Z Xof: back in the day, I remember hppa working (very slowly) on pa-risc 1.0 but having obscure memory faults on pa-risc 2.0
2015-10-26T20:39:41Z Xof: I think it depends on how many kernel bugs there are these days
2015-10-26T20:40:09Z nyef: Well, first things first.
2015-10-26T20:40:34Z stassats: but how often can you say "it's not my bug, it's an OS bug?"
2015-10-26T20:40:40Z stassats: "?
2015-10-26T20:40:49Z stassats: or even better, a hardware bug
2015-10-26T20:40:55Z stassats: haven't encountered the latter yet
2015-10-26T20:41:13Z nyef: stassats: Not nearly often enough, given that I hack on the Linux kernel and do some of my own hardware work.
2015-10-26T20:41:23Z stassats: well, a cpu bug, memory causing bugs, yes
2015-10-26T20:42:27Z stassats: my last two OS bugs were with NetBSD
2015-10-26T20:43:00Z Xof: I've been bitten by TCP/IP bugs in the NetBSD stack
2015-10-26T20:44:33Z nyef: Heh. I think it just failed all of the NO-CONSING tests. Or, at least, most of them. (-:
2015-10-26T20:45:10Z stassats: some tests expect disassemble annotation
2015-10-26T20:45:33Z stassats: i recently added CODE access annotation, so that at least take care of inlining checks
2015-10-26T20:45:45Z stassats: the non-consing checks probably didn't expect cheneygc
2015-10-26T20:46:04Z stassats: nobody expects cheneygc
2015-10-26T20:47:28Z nyef: Also, while MIPS was one of the earlier platforms to get some of the DX magic, it wasn't quite done correctly in a few cases.
2015-10-26T20:50:31Z stassats: waiting for make-target-2
2015-10-26T20:50:36Z stassats: it's taking fooooorever
2015-10-26T20:51:50Z stassats: i have a sudden urge to hack on qemu an optimize it, there ought to be a few easy things
2015-10-26T20:53:21Z stassats: i should also try qemulating sparc
2015-10-26T20:54:13Z nyef: ... I should get my SPARC system up and working again... And see if I can source a replacement CPU fan for the other one.
2015-10-26T20:55:00Z stassats: seems like there's a shortage of 32-bit linux sparcs
2015-10-26T20:57:14Z stassats: i don't really need a 32-bit linux, just the ability to run 32-bit code
2015-10-26T21:01:27Z nyef: Doesn't Debian tend to prefer 32-bit userlands?
2015-10-26T21:01:57Z Xof: debian sparc/linux definitely used to be 32-on-64
2015-10-26T21:02:00Z heddwch is now known as agent_nchambers
2015-10-26T21:02:10Z agent_nchambers is now known as HEDDWCH
2015-10-26T21:02:23Z dougk2 joined #sbcl
2015-10-26T21:03:54Z stassats: 22 minutes for make-target-2.sh
2015-10-26T21:04:30Z stassats: compared to 6 seconds on the host
2015-10-26T21:04:43Z Bicyclidine quit (Ping timeout: 256 seconds)
2015-10-26T21:07:17Z nyef: Xof: And I remember it being so last time I checked, as well. Same with Debian MIPS/linux, which is one of the reasons I'm running Gentoo instead.
2015-10-26T21:13:14Z nzambe quit (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
2015-10-26T21:20:25Z stassats: one mips change was missing, and the one that was there was wrong
2015-10-26T21:20:30Z HEDDWCH is now known as heddwch
2015-10-26T21:26:09Z nyef: stassats: So, I might get better build results if I update now?
2015-10-26T21:26:22Z stassats: after i check octets.pure
2015-10-26T21:26:30Z stassats: ok, success
2015-10-26T21:26:35Z stassats: pushed
2015-10-26T21:27:04Z nyef: My current run is making its way through the mop impure-cload tests now.
2015-10-26T21:27:15Z nyef: I'm not sure I should kill it at this point.
2015-10-26T21:28:23Z nyef: Okay, definitely shouldn't kill it.
2015-10-26T21:28:37Z stassats: anything with DEBUG in it is hosed
2015-10-26T21:31:20Z Bicyclidine joined #sbcl
2015-10-26T21:31:29Z nyef: http://paste.lisp.org/display/157516
2015-10-26T21:32:24Z stassats: run-program.impure.lisp / IS-/BIN/ED-INSTALLED? is clear
2015-10-26T21:32:54Z nyef: Right, I'll try to sort that out once I finish installing git.
2015-10-26T21:33:35Z stassats: some of it are due to tests not designed for cheneygc
2015-10-26T21:33:46Z nyef: I'm sure.
2015-10-26T21:37:58Z nyef: Okay, installing git is going to be a _pain_.
2015-10-26T21:41:13Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T21:44:30Z attila_lendvai joined #sbcl
2015-10-26T21:44:30Z attila_lendvai quit (Changing host)
2015-10-26T21:44:30Z attila_lendvai joined #sbcl
2015-10-26T21:50:46Z karswell joined #sbcl
2015-10-26T22:04:45Z beaumonta joined #sbcl
2015-10-26T22:05:05Z prxq quit (Remote host closed the connection)
2015-10-26T22:07:36Z dougk2 joined #sbcl
2015-10-26T22:21:38Z sjl quit (Quit: WeeChat 1.3)
2015-10-26T22:23:13Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-26T22:58:43Z beaumonta is now known as abeaumont_
2015-10-26T22:59:31Z dougk2 joined #sbcl
2015-10-26T23:07:54Z fe[nl]ix: Xof: debian sparc/linux used to be 32-on-64 because that's how Solaris was, so porting was easy
2015-10-26T23:09:19Z stassats: since this sparc m7 thing, i wonder how much of a work would it be to go from lowtags to hightags
2015-10-26T23:09:37Z stassats: let me see how many hightags does arm have available
2015-10-26T23:10:35Z nyef: Hrm. FIXNUMs would have an interesting distribution on a hightag system...
2015-10-26T23:10:37Z stassats: doesn't mess well with fixnum
2015-10-26T23:10:37Z stassats: s
2015-10-26T23:10:41Z stassats: mesh
2015-10-26T23:10:43Z nyef: Heh.
2015-10-26T23:11:19Z nyef: Actually, I think that fixnums might work okay.
2015-10-26T23:11:46Z stassats: positive fixnums could work wonderfully, but the negative ones would have to be sign-extended
2015-10-26T23:12:05Z nyef: The negative-fixnum tag would be all ones.
2015-10-26T23:13:19Z stassats: for four tag bits? that's limiting the range
2015-10-26T23:15:12Z stassats: arvm8 can sign extend from arbitrary lengths
2015-10-26T23:15:43Z nyef: For however many tag bits, although the practical minimum would be two.
2015-10-26T23:18:24Z stassats: sign extension is as cheap as shifting
2015-10-26T23:20:22Z stassats: ok, arm64 has 8 high tag bits
2015-10-26T23:20:55Z stassats: but this is optional
2015-10-26T23:21:41Z stassats: let's see what linux does https://www.kernel.org/doc/Documentation/arm64/tagged-pointers.txt
2015-10-26T23:22:28Z nyef: Tags destroyed by signal handling.
2015-10-26T23:22:49Z stassats: surely not for things stashed in registers?
2015-10-26T23:23:13Z nyef: "Non-zero tags are not preserved when delivering signals."
2015-10-26T23:23:49Z stassats: how can you imagine it would save the registers without tags? how does it know which of the registers are pointers?
2015-10-26T23:23:53Z nyef: Not sure if that applies to the registers in a sigcontext, though.
2015-10-26T23:24:24Z stassats: pc cannot tagged, not big of a deal
2015-10-26T23:25:01Z nyef: Means that structure pointers are aligned, at least, which should simplify slot addressing.
2015-10-26T23:25:03Z stassats: only fault address will end up untagged
2015-10-26T23:25:11Z stassats: the thing we care about
2015-10-26T23:25:34Z nyef: So... you'd have to back-match it to the register.
2015-10-26T23:26:06Z nyef: Or acknowledge that you're on a sparse BIBOP setup.
2015-10-26T23:26:25Z stassats: well, do we need the tag from the fault?
2015-10-26T23:26:42Z stassats: it's either just printed, or gc protects/unprotects things
2015-10-26T23:27:05Z stassats: besides, we don't have tagged fault addresses anyway
2015-10-26T23:27:26Z stassats: would have been great having a tagged PC
2015-10-26T23:31:35Z nyef: What would you tag it with?
2015-10-26T23:31:55Z nyef: Oh, right. LRA?
2015-10-26T23:32:55Z stassats: right, immediate instruction pointers, something like that
2015-10-26T23:33:12Z stassats: well, that's where a lisp machine needs to broken out
2015-10-26T23:33:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-26T23:37:14Z stassats: i really wonder how that sparc widetag thing works, does it have to effectively do two memory loads
2015-10-26T23:37:24Z stassats: not many details yet
2015-10-26T23:38:21Z stassats: was the symbolics lispm cpu a load/store one or ciscy?
2015-10-26T23:39:04Z nyef: Didn't SMBX have something like three separate generations of CPU design?
2015-10-26T23:42:10Z stassats: if we can just get rid of LRA and reg_CODE, then things would be much more reasonable, no fancy new instruction sets needed
2015-10-26T23:42:44Z nyef: I know that the TI system was very CISCy at the macro-level, but at the micro-level, well... Main memory was accessed by writing the address to a "functional register" and then reading or writing data to another "functional register".
2015-10-26T23:42:49Z stassats: but pairing PC via component_ptr_from_pc is not great
2015-10-26T23:43:40Z stassats: it could be paired via LEXENV, but that leaves a window during a function call
2015-10-26T23:44:10Z nyef: If code objects didn't move, things would also be a lot simpler, wouldn't they?
2015-10-26T23:44:19Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-26T23:44:36Z stassats: didn't move while being on the stack?
2015-10-26T23:44:50Z nyef: Didn't move, period.
2015-10-26T23:45:05Z stassats: so, some sort of a new GC then?
2015-10-26T23:45:11Z nyef: If there were no references, they could be freed, but they just Didn't Move.
2015-10-26T23:45:11Z stassats: new gc mode
2015-10-26T23:45:40Z nyef: Well, it wouldn't work with cheneygc, and would likely require some work on the gencgc end...
2015-10-26T23:45:46Z stassats: we could also make the code pages non-writable
2015-10-26T23:45:51Z stassats: W^X
2015-10-26T23:46:07Z nyef: That'd break TRACE :ENCAPSULATE NIL, you realize.
2015-10-26T23:46:08Z stassats: security is in vogue nowadays
2015-10-26T23:46:22Z stassats: how much more broken can it get?
2015-10-26T23:47:11Z nyef: Well, you could catch a segmentation violation to the face for merely trying to SET the breakpoint, not just for hitting it.
2015-10-26T23:48:06Z nyef: Ooh. Finally, git is getting installed.
2015-10-26T23:48:57Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-26T23:50:01Z Bicyclidine quit (Quit: allowed)
2015-10-26T23:50:22Z stassats: ooops, while making a sandwich i realized that my debug info compaction may be broken
2015-10-26T23:50:57Z stassats: no, i think it's alright, but the whole debug info may be broken
2015-10-26T23:51:14Z stassats: (doesn't make it sound any more promising)
2015-10-26T23:51:52Z nyef: The debugger is a festering sore anyway, and has long been on my list of things to figure out how to sort out.
2015-10-26T23:52:17Z stassats: debug info seems to be mostly alright
2015-10-26T23:52:35Z stassats: but the debugger itself is not pretty, yeah
2015-10-26T23:53:25Z stassats: so, offsets in sc-offsets are 21-bit long
2015-10-26T23:54:17Z stassats: 16MB of stack offsets are probably nothing to worry about, ok
2015-10-26T23:55:05Z nyef: And similar for access to the constant vector.
2015-10-26T23:55:10Z stassats: that's when running, what if i compile something like that
2015-10-26T23:55:32Z stassats: SB-C:MAKE-SC-OFFSET doesn't check things and will silently cut them
2015-10-26T23:55:55Z stassats: still, good luck compiling something with 2097151 stack saves
2015-10-26T23:58:35Z stassats: i think i found a problem in the debugger anyway
2015-10-26T23:58:57Z stassats: (defun foo () (let* ((x (the double-float (bar))) (y (+ x 1d0))) (break) y)) Y = 3.3992312471955544d-313
2015-10-26T23:59:03Z stassats: doesn't sound right to me
2015-10-27T00:02:04Z nyef: ... Where did BAR com from?
2015-10-27T00:02:11Z nyef: Err... come from.
2015-10-27T00:02:13Z stassats: outer space
2015-10-27T00:03:06Z nyef: Also, do we still have that issue where D-X closed-over values aren't handled properly in the debugger, and are value cells still invisible to the debugger?
2015-10-27T00:03:11Z stassats: does (stack-frame-offset 2 0) sound weird to me only?
2015-10-27T00:03:46Z nyef: Umm... The function name sounds familiar, to the point where I might have been the one to introduce it, but I don't remember anything about it.
2015-10-27T00:03:54Z stassats: ok, 0 is just an offset for the offset
2015-10-27T00:06:12Z stassats: but the bug is in stack-frame-offset anyway
2015-10-27T00:06:33Z stassats: what's this https://github.com/sbcl/sbcl/blob/master/src/code/debug-int.lisp#L2168 for?
2015-10-27T00:06:50Z stassats: https://github.com/sbcl/sbcl/blob/master/src/code/debug-int.lisp#L2165-L2168
2015-10-27T00:07:05Z nyef: How on earth did I end up with /etc/resolv.conf set chmod 600 ?
2015-10-27T00:07:19Z stassats: bad umask?
2015-10-27T00:07:49Z stassats: huh, that data-width is only on x86oids
2015-10-27T00:07:54Z nyef: Heh.
2015-10-27T00:08:00Z stassats: taken out, i get my 2d0
2015-10-27T00:08:11Z stassats: time to blame
2015-10-27T00:08:50Z nyef: Oh! You're looking at the horrid mess that is %sub-access-debug-var-slot ?
2015-10-27T00:09:00Z stassats: where else
2015-10-27T00:09:28Z nyef: Sorry, I had mostly repressed my memory of that festering sore.
2015-10-27T00:09:38Z nyef: No wonder stack-frame-offset sounded familiar.
2015-10-27T00:09:50Z stassats: so, previously it was just data-width, but then it got changed to (1- data-width)
2015-10-27T00:10:01Z scymtym_ quit (Ping timeout: 256 seconds)
2015-10-27T00:10:02Z stassats: so, most things have with 1, making it 0
2015-10-27T00:10:08Z stassats: but not double-float
2015-10-27T00:10:15Z nyef: What the hell?
2015-10-27T00:10:25Z stassats: now, how does it decide that double-float is 2-word wide on both x86 and x86-64?
2015-10-27T00:10:54Z nyef: I'm vaguely remembering something about the width being significant depending on which way the stack grows...
2015-10-27T00:11:22Z nyef: Hah! If you think that's bad... Are complex floats still screwed up?
2015-10-27T00:13:03Z stassats: so, the stack grows downwards on x86oids, but upwards on riscs, but wait, there's more. the number stack usually grows downwards on riscs
2015-10-27T00:14:21Z stassats: considering this, there should be a problem with offsets, not with widths?
2015-10-27T00:14:38Z stassats: best demonstrated with complex doubles, yes
2015-10-27T00:16:05Z stassats: so, #C(2d0 3d0) shows up as #C(3.0d0 3.397267908183727d-313) with data-width removed, #C(2.0d0 6.9531892602245446d-310) with it intact
2015-10-27T00:16:14Z stassats: total clusterfuck
2015-10-27T00:16:26Z nyef: So, who last touched it?
2015-10-27T00:17:44Z stassats: it'd be good idea to merge sub-access-debug-var-slot and sub-set-access-debug-var-slot
2015-10-27T00:17:55Z nyef: Not it!
2015-10-27T00:18:06Z nyef: I had enough trouble merging the x86oid and non-x86oid versions.
2015-10-27T00:18:35Z stassats: found meself a project after the freeze
2015-10-27T00:18:44Z stassats: fix the float access mess
2015-10-27T00:19:37Z nyef: Let me guess... commit 952d16ab5880823c1864eb9105bb269e2e00760d did the damage?
2015-10-27T00:20:13Z stassats: i think it was broken already, it just swept things under the floor
2015-10-27T00:21:03Z nyef: Nope, this is definitely the commit that added the (1- ,data-width) bit.
2015-10-27T00:21:13Z stassats: since most widths are 1, (1- 0) just neutralized
2015-10-27T00:21:29Z stassats: (1- 1) that is
2015-10-27T00:21:40Z stassats: but (1- 2) (for double floats) didn't get fixed
2015-10-27T00:21:43Z nyef: Yeah, yeah... The delta was applied to the wrong place.
2015-10-27T00:21:53Z nyef: It's clearly an offset, not a width thing.
2015-10-27T00:21:58Z stassats: so, width doesn't matter at all
2015-10-27T00:22:31Z nyef: There was, at one point, a point to the width.
2015-10-27T00:22:39Z stassats: and the direction of the offset matters, for different things too, but wide things are only saved on the number stack, so probably no need in two routines
2015-10-27T00:22:47Z nyef: And it was precisely for things like floats and complexes that take up multiple slots.
2015-10-27T00:22:49Z stassats: for different stacks, that is
2015-10-27T00:23:22Z stassats: sap referrers take care of the width
2015-10-27T00:23:43Z stassats: can't infer from the width which way the stack grows
2015-10-27T00:25:11Z stassats: easy enough to fix, but i don't want to be fixing it twice, so i'll merge sub-access-debug-var-slot and sub-set-access-debug-var-slot
2015-10-27T00:27:06Z stassats: how come #.sb!vm:complex-double-reg-sc-number is #-x86oided
2015-10-27T00:29:09Z nyef: Ugh. I have git, but it wants a password in order to pull, even though I have my ssh keys loaded?
2015-10-27T00:29:37Z nyef: Let me guess... I come up on git blame for that last one?
2015-10-27T00:30:02Z nyef: Oh, wait! I know this one.
2015-10-27T00:30:26Z nyef: x86oid float registers are long-floats (80 bits wide), and we don't support long-floats.
2015-10-27T00:31:29Z stassats: i think it's more of a, can't access escaped FPU registers
2015-10-27T00:31:35Z stassats: but, only on x86
2015-10-27T00:31:58Z nyef: The _reason_ we can't access them is that they're 80 bits wide, not 32 or 64.
2015-10-27T00:32:18Z nyef: Why the heck can't I do a git pull from this box?!?
2015-10-27T00:34:00Z stassats: just use a non-ssh repo url?
2015-10-27T00:34:58Z stassats: so, made the debugger to come down with an escaped complex-double-float
2015-10-27T00:34:59Z stassats: prints WARNING: stub CONTEXT-FLOAT-REGISTER
2015-10-27T00:38:01Z stassats: nyef: cannot 80 bit floats be coerced to the correct format?
2015-10-27T00:38:18Z stassats: anyhow, this mess just got messier
2015-10-27T00:38:20Z nyef: It plausibly can. We just don't.
2015-10-27T00:38:54Z nyef: And, on the other side, why not resurrect the long-float support anyway?
2015-10-27T00:39:18Z stassats: just for the hell of it?
2015-10-27T00:39:34Z stassats: something to do on a rainy day, maybe
2015-10-27T00:39:57Z stassats: how well would it play with SSE?
2015-10-27T00:40:29Z stassats: not a problem on x86, but at this point, who cares about x86?
2015-10-27T00:41:00Z nyef: Okay, finally, got something to work for git pull.
2015-10-27T00:41:07Z nyef: Building a new MIPS build now.
2015-10-27T00:41:29Z stassats: and how well is long double supported by libm?
2015-10-27T00:43:31Z stassats: ok, maybe not such a great idea of unifying sub-access-debug-var-slot and sub-set-access-debug-var-slot
2015-10-27T00:43:39Z stassats: don't want to descend into macro hell
2015-10-27T00:43:45Z stassats: but sharing some bits will be useful
2015-10-27T00:46:04Z stassats: #!+rt #.sb!vm:word-pointer-reg-sc-number is amusing as well
2015-10-27T00:46:58Z stassats: i thin that's the only vestige of RT
2015-10-27T00:48:41Z nyef: Heh. Nice!
2015-10-27T00:49:18Z nyef: I don't think that it's worth trying to dig up a working IBM RT system to try and resurrect either of the CMUCL RT backends.
2015-10-27T00:49:45Z stassats: but more curiously, that bit was added in 1991
2015-10-27T00:50:18Z stassats: with the commit message being "William needs this file."
2015-10-27T00:50:35Z stassats: by Bill Chiles
2015-10-27T00:52:23Z stassats: coincidentally, RT was discontinued in 1991
2015-10-27T00:53:24Z stassats: blast from the past: https://books.google.ru/books?id=33QfOHT69aMC&pg=PA8&hl=en#v=onepage&q&f=false
2015-10-27T00:54:28Z stassats: some video materials http://www.youtube.com/watch?feature=player_detailpage&v=iEou9AbTVT0#t=400
2015-10-27T00:59:46Z stassats: for the price of $16,485
2015-10-27T01:02:12Z nyef: In, what, '86 dollars?
2015-10-27T01:02:17Z stassats: yeah
2015-10-27T01:02:26Z nyef: So, pre-hyperinflation.
2015-10-27T01:03:02Z stassats: some calculater puts it to $35,000 today
2015-10-27T01:03:20Z stassats: well, maybe just enough for some ibm power system
2015-10-27T01:05:41Z stassats: i guess you can't get one without a support contract anyway, so, may not be enough
2015-10-27T01:09:56Z stassats: so many cobol adds in that computerworld from 1986
2015-10-27T01:10:55Z stassats: there's a VAX LISP ad
2015-10-27T01:30:53Z nyef: Genesis, finally.
2015-10-27T01:31:05Z nyef: About 50 minutes.
2015-10-27T01:33:13Z myrkraverk: But there's no lisp for current VMS, is there?
2015-10-27T01:38:36Z nyef: myrkraverk: Would it help or horrify you to know that I'm considering trying to get a VMS hobbyist license?
2015-10-27T01:38:50Z myrkraverk: nyef: it'll make me happy C:
2015-10-27T01:42:23Z zacts quit (Read error: Connection reset by peer)
2015-10-27T01:42:56Z zacts joined #sbcl
2015-10-27T01:43:03Z dougk2 joined #sbcl
2015-10-27T01:46:01Z stassats: do you have an itanium?
2015-10-27T01:47:40Z nyef: I... might have a usable itanic, I might not. I have an Alpha, though!
2015-10-27T01:48:12Z nyef: Okay! Build complete... And sb-md5 still failed.
2015-10-27T01:48:26Z nyef: Call it about 70 minutes end-to-end?
2015-10-27T01:49:10Z nyef: Running the test suite.
2015-10-27T02:04:37Z whiteline quit (Ping timeout: 252 seconds)
2015-10-27T02:07:48Z nyef: I think that my first priority gets to be figuring out the md5 failure, because that really shouldn't be able to break.
2015-10-27T02:15:55Z nyef: Oh, and I should, finally, try to verify a self-build.
2015-10-27T02:33:08Z nyef: http://paste.lisp.org/display/157516#1
2015-10-27T02:33:28Z nyef: Call it about 45 minutes to run the test suite.
2015-10-27T02:47:57Z stassats quit (Ping timeout: 255 seconds)
2015-10-27T03:19:27Z ASau quit (Ping timeout: 255 seconds)
2015-10-27T03:23:12Z ASau joined #sbcl
2015-10-27T03:39:11Z dougk2 quit (Ping timeout: 264 seconds)
2015-10-27T04:05:35Z dougk2 joined #sbcl
2015-10-27T06:04:54Z Quadrescence joined #sbcl
2015-10-27T06:09:17Z psy_ joined #sbcl
2015-10-27T06:23:46Z fridim_ joined #sbcl
2015-10-27T06:33:13Z psy_ quit (Ping timeout: 246 seconds)
2015-10-27T06:50:23Z Shinmera joined #sbcl
2015-10-27T06:55:20Z Guest78532 joined #sbcl
2015-10-27T06:59:58Z gingerale joined #sbcl
2015-10-27T07:21:48Z heddwch quit (Read error: Connection reset by peer)
2015-10-27T07:22:20Z heddwch joined #sbcl
2015-10-27T07:44:34Z Guest78532 left #sbcl
2015-10-27T08:15:11Z ASau quit (Ping timeout: 264 seconds)
2015-10-27T08:16:03Z stassats joined #sbcl
2015-10-27T08:19:44Z heddwch left #sbcl
2015-10-27T08:24:09Z ZabaQ_ joined #sbcl
2015-10-27T08:30:39Z Cymew joined #sbcl
2015-10-27T08:32:12Z fridim_ quit (Ping timeout: 255 seconds)
2015-10-27T08:34:27Z stassats quit (Ping timeout: 255 seconds)
2015-10-27T08:59:54Z Cymew quit (Read error: Connection reset by peer)
2015-10-27T09:07:14Z whiteline joined #sbcl
2015-10-27T09:44:25Z Cymew joined #sbcl
2015-10-27T10:22:01Z Quadrescence quit (Ping timeout: 256 seconds)
2015-10-27T10:24:43Z Quadrescence joined #sbcl
2015-10-27T11:30:02Z myrkraverk: When I have a pointer from a CFFI function, and I want to run a finaziler on said pointer, what's the best way to do it?
2015-10-27T11:30:30Z myrkraverk: I take the 'net to mean that if I simply refer the pointer in the finalizer function itself, it'll never be GCd.
2015-10-27T11:31:15Z myrkraverk: So, I'm wondering, how do I do this?
2015-10-27T11:31:43Z myrkraverk: Can I "wrap" the pointer in a CLOS object, and then attach the finalizer to that instead?
2015-10-27T11:33:51Z jackdaniel: myrkraverk: CFFI returns foreign-porter structure which is backend dependant (in case of sbcl its 'sb-sys:system-area-pointer which is a built-in class
2015-10-27T11:34:02Z jackdaniel: s/foreign-porter/foreign-pointer/
2015-10-27T11:34:23Z jackdaniel: so I believe you can pass it to the finalizer as-is
2015-10-27T11:34:58Z jackdaniel: s/pass to/register with/
2015-10-27T11:35:08Z myrkraverk: Ok, I'll try it.
2015-10-27T11:35:46Z Bike quit (Ping timeout: 272 seconds)
2015-10-27T11:42:26Z dougk2 quit (Ping timeout: 260 seconds)
2015-10-27T11:49:29Z myrkraverk: When I do something like
2015-10-27T11:49:31Z myrkraverk: (sb-ext:finalize str (lambda () (ckytea-delete-string str (callback error-callback))))
2015-10-27T11:49:59Z myrkraverk: I do not see any evidence of ckytea-delete-string being called.
2015-10-27T11:50:36Z myrkraverk: Unless the finalizer is run in some sort of context where STDOUT wouldn't work.
2015-10-27T11:51:18Z jackdaniel: myrkraverk: I'm not familiar enough with the sbcl internals to say anything what wouldn't be a guess, sorry
2015-10-27T11:51:29Z jackdaniel: maybe try to increment some *value*
2015-10-27T11:51:35Z jackdaniel: and read it from somewhere else
2015-10-27T12:20:08Z Xach joined #sbcl
2015-10-27T12:40:20Z dougk2 joined #sbcl
2015-10-27T12:45:25Z pkhuong: myrkraverk: the finalizer callback has a reference to the finalized object (str) so the object will never be GCed. anyone on #lisp could have helped you out.
2015-10-27T12:48:50Z pkhuong: and finalizing a SAP is useless/dangerous. it's the same as finalising a bignum
2015-10-27T12:52:49Z psilord1 quit (Quit: Leaving.)
2015-10-27T13:00:06Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-27T13:18:28Z dougk2 joined #sbcl
2015-10-27T13:19:01Z stassats joined #sbcl
2015-10-27T13:24:18Z Xach left #sbcl
2015-10-27T13:27:54Z dougk2 quit (Ping timeout: 260 seconds)
2015-10-27T13:29:28Z eudoxia joined #sbcl
2015-10-27T13:42:38Z myrkraverk: pkhuong: Yes, I thought so.
2015-10-27T13:43:21Z myrkraverk: I'm taking a break, but I think this is the basics of what I want to do.
2015-10-27T13:43:27Z myrkraverk: http://pvk.ca/Blog/Lisp/finalizing_foreign_pointers_just_late_enough.html
2015-10-27T13:44:05Z myrkraverk: Though I think I'll do my wrapping with a CLOS object.
2015-10-27T13:44:08Z myrkraverk: I hope that'll work too.
2015-10-27T13:45:23Z nzambe joined #sbcl
2015-10-27T13:49:42Z psilord joined #sbcl
2015-10-27T13:51:57Z karswell quit (Read error: Connection reset by peer)
2015-10-27T13:52:07Z karswell joined #sbcl
2015-10-27T13:52:28Z dougk2 joined #sbcl
2015-10-27T14:02:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-27T14:10:52Z fridim_ joined #sbcl
2015-10-27T14:11:25Z dougk2 joined #sbcl
2015-10-27T14:12:31Z _iwc joined #sbcl
2015-10-27T14:25:09Z _iwc quit (Killed (Sigyn (Spam is off topic on freenode.)))
2015-10-27T14:27:37Z Ober quit (Ping timeout: 268 seconds)
2015-10-27T14:28:00Z Ober joined #sbcl
2015-10-27T14:29:13Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-27T14:30:46Z dougk2 joined #sbcl
2015-10-27T15:23:50Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-27T15:29:53Z dougk2 joined #sbcl
2015-10-27T15:47:21Z Bike joined #sbcl
2015-10-27T16:07:22Z karswell quit (Remote host closed the connection)
2015-10-27T16:07:38Z karswell joined #sbcl
2015-10-27T16:17:18Z dougk2 quit (Ping timeout: 260 seconds)
2015-10-27T16:19:00Z dougk2 joined #sbcl
2015-10-27T16:37:59Z dougk2 quit (Ping timeout: 264 seconds)
2015-10-27T16:39:12Z dougk2 joined #sbcl
2015-10-27T16:42:59Z nyef: sb-md5 damage: http://paste.lisp.org/display/157516#2
2015-10-27T16:45:05Z nyef: ... This... doesn't look like it works AT ALL correctly for string inputs.
2015-10-27T16:45:39Z nyef: SIMPLE-BASE-STRING, yes, but the reader doesn't seem to return those by default.
2015-10-27T16:50:42Z stassats: le me try
2015-10-27T16:50:57Z Xof: I wouldn't expect that to work, though
2015-10-27T16:51:31Z Xof: copy-to-buffer is internal-only
2015-10-27T16:52:02Z Xof: but yes, it should probably be octet-buffer-only
2015-10-27T16:52:30Z nyef: I'm tracing through from update-md5-state, which is exposed.
2015-10-27T16:53:03Z Cymew quit (Ping timeout: 252 seconds)
2015-10-27T16:53:18Z Xof: ah.  I think that should only get called (in our code) with octets
2015-10-27T16:53:40Z nyef: Hrm. Guess my next test is something against update-md5-block, then.
2015-10-27T16:56:48Z stassats: same thing here
2015-10-27T16:57:17Z nyef: stassats: Are you on mips or mipsle?
2015-10-27T16:57:23Z stassats: mips
2015-10-27T16:57:29Z Xof: I don't really understand the difference between machines, mind you
2015-10-27T16:57:31Z nyef: It occurs to me that we might see different effects between big and little endian.
2015-10-27T16:57:33Z Xof: so maybe you should just carry on
2015-10-27T16:57:42Z Xof: ah yes
2015-10-27T16:57:52Z nyef: And since I don't have any little-endian mips hardware...
2015-10-27T17:01:07Z nyef: Okay, update-md5-block is busted.
2015-10-27T17:01:07Z stassats: (let ((x (make-string 10))) (sb-kernel:ub8-bash-copy "from" 0 x 0 1) x) => ""
2015-10-27T17:01:10Z stassats: on mips
2015-10-27T17:01:19Z stassats: yet (let ((x (make-string 10))) (sb-kernel:ub8-bash-copy "from" 0 x 0 16) x) => "from"
2015-10-27T17:01:25Z stassats: so, it is endianness
2015-10-27T17:01:51Z nyef: That bit is, yes.
2015-10-27T17:02:16Z stassats: but ppc is big endian too, and there's no failures there
2015-10-27T17:02:17Z nyef: But Xof points out that it may not be an issue in practice.
2015-10-27T17:03:32Z nyef: And I'm getting more substantial garbage from update-md5-block.
2015-10-27T17:04:03Z stassats: which test actually fails?
2015-10-27T17:04:31Z nyef: Every test other than sb-md5.md5sum-sequence.error.0
2015-10-27T17:04:57Z stassats: sb-md5 didn't fail here
2015-10-27T17:05:02Z nyef: Interesting.
2015-10-27T17:05:10Z stassats: (sb-md5:md5sum-string "a") => #(12 193 117 185 192 241 182 168 49 195 153 226 105 119 38 97)
2015-10-27T17:05:10Z stassats:  
2015-10-27T17:05:29Z nyef: #(65 133 9 76 77 92 107 12 33 254 195 15 77 194 251 24)
2015-10-27T17:06:30Z nyef: (sb-md5:update-md5-block (sb-md5::initial-md5-regs) (make-array 16 :element-type '(unsigned-byte 32) :initial-element 0))
2015-10-27T17:06:34Z stassats: what about (sb-md5:md5sum-sequence (coerce #(1 2 3) '(vector (unsigned-byte 8)))) ?
2015-10-27T17:07:05Z nyef: I get #(1698611957 3225902988 435152416 746092991) on MIPS and #(52370860 1856343760 531326903 1949398929) on x86-64.
2015-10-27T17:07:15Z stassats: i get #(52370860 1856343760 531326903 1949398929) on "mips"
2015-10-27T17:07:36Z nyef: What emulated CPU?
2015-10-27T17:07:49Z stassats: MIPS 24Kc V0.0  FPU V0.0
2015-10-27T17:07:59Z stassats: malta
2015-10-27T17:08:02Z nyef: Hrm. What ISA level is that?
2015-10-27T17:08:36Z stassats: mips32?
2015-10-27T17:08:46Z nyef: What isa does /proc/cpuinfo report?
2015-10-27T17:09:06Z stassats: no such thing
2015-10-27T17:09:10Z nyef: Err?
2015-10-27T17:09:55Z nyef: Let me guess, that's only in newer linux kernels. /-:
2015-10-27T17:10:38Z stassats: i assume this v6
2015-10-27T17:12:21Z stassats: and it doesn't really matter, since it may be wrong
2015-10-27T17:12:55Z nyef: Right, so, without reference to what you're running, I need to track down what's going on with update-md5-block.
2015-10-27T17:13:46Z nyef: Probably going to be some 32-bit assumption that gets badly savaged somewhere along the line.
2015-10-27T17:14:10Z nyef: To the point where single-stepping the code might mask the problem!
2015-10-27T17:14:44Z scymtym_ joined #sbcl
2015-10-27T17:17:00Z stassats: it seems like some newer mips are more attractive, no branch delay slots
2015-10-27T17:17:27Z nyef: ... Oh, good grief.
2015-10-27T17:17:39Z stassats: for new branches, of course
2015-10-27T17:19:20Z stassats: callce BC, Branch Compact
2015-10-27T17:19:54Z stassats: and with 26-bit offsets, not 16-bit
2015-10-27T17:22:18Z stassats: were delay slots only useful for in-order execution?
2015-10-27T17:22:53Z pkhuong: yes.
2015-10-27T17:23:04Z pkhuong: and even then didn't help that much for deeply pipelined machines
2015-10-27T17:25:14Z karswell quit (Read error: Connection reset by peer)
2015-10-27T17:29:46Z ZabaQ_ quit (Ping timeout: 240 seconds)
2015-10-27T17:32:19Z sjl joined #sbcl
2015-10-27T17:35:00Z eudoxia_ joined #sbcl
2015-10-27T17:36:50Z eudoxia_ quit (Client Quit)
2015-10-27T17:38:55Z eudoxia quit (Ping timeout: 256 seconds)
2015-10-27T17:46:11Z dougk2 quit (Ping timeout: 250 seconds)
2015-10-27T17:48:21Z psy_ joined #sbcl
2015-10-27T17:55:29Z dougk2 joined #sbcl
2015-10-27T17:58:39Z fiddlerwoaroof joined #sbcl
2015-10-27T18:10:15Z dougk2 quit (Ping timeout: 268 seconds)
2015-10-27T18:18:26Z nyef: Found it! sb-md5::*t* is wrong.
2015-10-27T18:18:45Z stassats: the float bit?
2015-10-27T18:19:19Z nyef: Yeah.
2015-10-27T18:20:44Z nyef: So, new plan: I'm going to leave sb-md5 for a while and work on some of the other egregious failures.
2015-10-27T18:22:21Z stassats: like what?
2015-10-27T18:22:42Z nyef: Like the float.pure.lisp test failures.
2015-10-27T18:22:52Z stassats: so, same thing?
2015-10-27T18:23:02Z nyef: Yeah, though possibly more targetted.
2015-10-27T18:24:04Z stassats: was thinking about component_ptr_from_pc, code header could contain a sequence of bytes that can never be an instruction
2015-10-27T18:24:11Z stassats: that way PC could be walked backwards
2015-10-27T18:25:48Z stassats: or something legal, but that we never use, like debug-trap +magic+
2015-10-27T18:25:53Z nyef: Oh, for the... *t* is dereferenced at build-time. /-:
2015-10-27T18:26:34Z nyef: What's the problem with walking forwards from the start of the allocation region?
2015-10-27T18:26:41Z stassats: slow
2015-10-27T18:27:19Z stassats: especially for profiling
2015-10-27T18:28:19Z nyef: How so?
2015-10-27T18:28:27Z nyef: Especially, how so when compared with walking backwards?
2015-10-27T18:28:47Z stassats: PC is closer to the code object
2015-10-27T18:29:35Z nyef: Walking through a region of code objects is fast, because the code objects tend to be fairly large, surely?
2015-10-27T18:30:09Z stassats: how large are the regions?
2015-10-27T18:30:53Z nyef: That, I don't know.
2015-10-27T18:31:36Z stassats: currently, for profiling we can just query reg_CODE where it's present
2015-10-27T18:32:22Z stassats: but that's just for one level of the stack
2015-10-27T18:34:38Z stassats: does component_ptr_from_pc use binary search? can it?
2015-10-27T18:36:19Z dougk2 joined #sbcl
2015-10-27T18:38:11Z stassats: what is allocated in the unboxed region?
2015-10-27T18:41:20Z stassats: looks like nothing, but later copied to it?
2015-10-27T18:42:28Z nyef: Yeah, I'm not quite impressed with some aspects of our GC.
2015-10-27T18:43:14Z stassats: and code objects are both
2015-10-27T18:44:19Z nyef: Heh. Now you get to dig into the mess that is CODE_PAGE_FLAG, right?
2015-10-27T18:44:40Z nyef: ... while I puzzle over a division-by-zero at compile-time.
2015-10-27T18:44:46Z stassats: say we leave CODE, but save it on the stack, use it to pair LRA
2015-10-27T18:44:52Z stassats: and for quick backtraces
2015-10-27T18:45:55Z nyef: So... a two-part LRA?
2015-10-27T18:46:22Z stassats: well, no lra object embedded at all
2015-10-27T18:46:37Z stassats: i want normal calling conventions
2015-10-27T18:47:16Z stassats: like branch-link or CALL
2015-10-27T18:48:23Z stassats: ok, and distancing from CODE by using the already established LEXENV ABI, but saving it on the stack
2015-10-27T18:48:57Z stassats: that requires for backtraces to always know where to find LEXENV and return-pc
2015-10-27T18:50:10Z stassats: if C doesn't push anything to the control stack, the old stack frames can be easily determined
2015-10-27T18:50:17Z stassats: now for the transition to the new frame
2015-10-27T18:51:48Z stassats: i need a pencil and paper for this
2015-10-27T18:52:45Z stassats: i need to study stack scavenging routines
2015-10-27T18:53:42Z stassats: and nothing like that can be done with PC-relative addressing, so, another split, not good, but whatcha gonna do
2015-10-27T18:53:43Z nyef: Ours, or other people's?
2015-10-27T18:54:01Z stassats: sbcl's
2015-10-27T18:54:24Z nyef: Ah.
2015-10-27T18:54:41Z nyef: It's a linear array of tagged words.
2015-10-27T18:54:44Z stassats: and the debugger
2015-10-27T18:55:30Z stassats: nyef: my scheme would have to break the tagged part
2015-10-27T18:58:05Z stassats: and to figure exactly how slow or fast component_ptr_from_pc can be
2015-10-27T18:58:56Z nyef: You might find it useful to arrange, forcefully, that code objects are always in "code pages".
2015-10-27T18:59:06Z sjl quit (Ping timeout: 272 seconds)
2015-10-27T18:59:40Z stassats: at this point, just one more stack store ain't gonna hurt
2015-10-27T19:00:04Z stassats: time to stare at some arm disassembly
2015-10-27T19:01:20Z nyef: ... Okay, every single failure in gc.impure.lisp is from some gencgcism. ONE of the four failures is because the test doesn't expect the heap to have been purified.
2015-10-27T19:01:30Z nyef: The other three are from using gencgc-specific functions.
2015-10-27T19:02:11Z nyef: I've got backtraces that don't work through an escaped frame...
2015-10-27T19:03:39Z nyef: Ugh.
2015-10-27T19:09:11Z ASau joined #sbcl
2015-10-27T19:34:11Z nyef: Hrm. And my MIPS ASM-fu is surprisingly rusty.
2015-10-27T19:40:33Z stassats: "Call for Amiga Developers Kickstarting PowerPC and 68k software development "
2015-10-27T19:40:45Z stassats: porting sbcl to 68k amiga?
2015-10-27T19:41:46Z stassats: from http://amigadeveloper.com/
2015-10-27T19:45:23Z Shinmera: When will we get SBCL on a dead badger?
2015-10-27T19:45:42Z stassats: honey badgers are hard to kill
2015-10-27T19:47:12Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-27T19:49:13Z dougk2 joined #sbcl
2015-10-27T19:55:21Z reb: I like 68K ... once upon a time I wrote a 68K interpreter and a converter from 68K object code to Sparc.
2015-10-27T19:56:59Z nyef: The 68k was fun.
2015-10-27T20:01:15Z psy_ quit (Ping timeout: 268 seconds)
2015-10-27T20:06:49Z dougk2 quit (Ping timeout: 256 seconds)
2015-10-27T20:15:07Z nyef: ... Alpha build doesn't survive host-1.
2015-10-27T20:15:29Z nyef: I think I'm going to leave that alone for a bit. (-:
2015-10-27T20:24:37Z White_Flame: 68k instruction set is fun.  The instruction timing tables are not
2015-10-27T20:26:48Z dougk2 joined #sbcl
2015-10-27T20:43:35Z nyef: ... I'm being punished by -fomit-frame-pointer, aren't I?
2015-10-27T20:55:09Z dougk2 quit (Ping timeout: 255 seconds)
2015-10-27T20:59:34Z dougk2 joined #sbcl
2015-10-27T21:06:57Z attila_lendvai joined #sbcl
2015-10-27T21:06:57Z attila_lendvai quit (Changing host)
2015-10-27T21:06:57Z attila_lendvai joined #sbcl
2015-10-27T21:11:42Z gingerale quit (Remote host closed the connection)
2015-10-27T21:27:02Z nyef: Almost certainly being punished by -fomit-frame-pointer. Catching a SIGFPE to the face, backtrace stops at call_into_lisp for the handler, and *free-interrupt-context-index* is 0.
2015-10-27T21:27:25Z stassats: ok, sigfpe, is there my fault?
2015-10-27T21:27:38Z stassats: because i removed the sigfpe handler
2015-10-27T21:27:44Z nyef: I fail to see mechanism, given what you were removing.
2015-10-27T21:27:51Z stassats: in c, but it shouldn't natter
2015-10-27T21:28:19Z stassats: can you give me a test case?
2015-10-27T21:28:32Z nyef: (defun foo () (log 123 (eval 0)))
2015-10-27T21:29:03Z stassats: that's 0.0 here
2015-10-27T21:29:45Z nyef: It fails to compile here.
2015-10-27T21:34:03Z nyef: And not being able to get a usable backtrace makes it ever so much better.
2015-10-27T21:34:19Z stassats: can compile
2015-10-27T21:34:30Z stassats: you can't get backtrack from fpu traps?
2015-10-27T21:37:19Z nyef: I think it's an alien trap, and the frame pointer chain is getting lost or something.
2015-10-27T21:37:40Z stassats: my attempts at fp traps seem to work
2015-10-27T21:37:53Z nyef: The undefined-function trap works, and can backtrace through, for example.
2015-10-27T21:38:01Z stassats: incidentally, tried killing sbcl with sigfpe, got "No traps are enabled? How can this be?"
2015-10-27T21:38:18Z stassats: (that's on x86-64)
2015-10-27T21:38:26Z nyef: And, as I said, *free-interrupt-context-index* is zero, indicating that it was delivered while the system was in alien code.
2015-10-27T21:38:45Z stassats: ok, let's try to do that
2015-10-27T21:39:17Z stassats: (sb-kernel:%log 0d0) has a bogus backtrace, yes
2015-10-27T21:39:21Z stassats: even on x86-64 linux
2015-10-27T21:40:29Z stassats: that'd be https://bugs.launchpad.net/sbcl/+bug/1133020
2015-10-27T21:41:27Z stassats: it's understandable on x86-64
2015-10-27T21:41:58Z stassats: is call_into_lisp broken on mips?
2015-10-27T21:42:04Z stassats: or call_into_c
2015-10-27T21:42:36Z stassats: of course no luck trying on arm, no float traps
2015-10-27T21:44:37Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-27T21:46:23Z nyef: Okay, what if the libm is being compiled -fomit-frame-pointer, and my runtime likewise?
2015-10-27T21:46:58Z stassats: but it's on the number stack, so it doesn't matter
2015-10-27T21:48:27Z dougk2 joined #sbcl
2015-10-27T21:49:30Z stassats: so, (read) kill -s SIGFPE `pidof sbcl` gives a broken backtrace too
2015-10-27T21:52:45Z stassats: https://github.com/sbcl/sbcl/blob/master/src/runtime/mips-assem.S#L276
2015-10-27T21:52:54Z stassats: it write LRA to S0
2015-10-27T21:53:14Z stassats: in lisp, it writes MOVE $NL2, $OCFP SW $NL2, $CFP
2015-10-27T21:53:25Z stassats: MOVE $A2, $LRA SW $A2, $CFP[4]
2015-10-27T21:54:21Z stassats: i don't see OCFP saved at all in call_into_c?
2015-10-27T21:55:25Z stassats: that would be hard to trace
2015-10-27T21:58:08Z fridim_ quit (Ping timeout: 252 seconds)
2015-10-27T21:58:59Z scymtym_ quit (Ping timeout: 264 seconds)
2015-10-27T22:02:17Z stassats: rebuilding will take a while for me, so, can't test
2015-10-27T22:02:31Z stassats: but that's clearly it
2015-10-27T22:06:11Z dougk2 quit (Ping timeout: 264 seconds)
2015-10-27T22:06:12Z stassats: or maybe i can without recompiling
2015-10-27T22:08:03Z nyef: Runtime-only change?
2015-10-27T22:08:14Z stassats: yeah, without moving things
2015-10-27T22:08:24Z nyef: Even with moving things this is doable.
2015-10-27T22:09:05Z nyef: Re-run make on the runtime, then sh make-genesis-2.sh, then do the load phase of make-target-2, and you're done.
2015-10-27T22:09:12Z stassats: and the way reg_NFP is being calculated and then discared is dubious
2015-10-27T22:09:15Z nyef: All of the fasls are still good.
2015-10-27T22:09:36Z stassats: nyef: i know, i'm the one who added the --load parameter to make-target.sh
2015-10-27T22:09:40Z stassats: but it's slow
2015-10-27T22:09:54Z nyef: The... Okay, I'll have to look for that one.
2015-10-27T22:10:09Z stassats: make-target-2.sh, that is
2015-10-27T22:10:39Z nyef: Okay, that's neat. Thanks.
2015-10-27T22:10:43Z stassats: compiling the runtime is slow enough already
2015-10-27T22:12:10Z hydan joined #sbcl
2015-10-27T22:12:52Z stassats: all other call_into_c seem to save what is in reg_NFP, not reg_LRA
2015-10-27T22:13:08Z stassats: it used to do this too
2015-10-27T22:13:37Z nyef: We're in code-freeze, right?
2015-10-27T22:13:50Z stassats: right, though nobody would bat an eye for a mips fix
2015-10-27T22:14:03Z stassats: constrained to the mips-only files
2015-10-27T22:14:24Z nyef: Heh. Maybe, but we may as well observe the forms. (-:
2015-10-27T22:14:43Z stassats: nobody runs it as it is
2015-10-27T22:14:43Z nyef: Get things sorted out to commit early post-freeze.
2015-10-27T22:15:06Z stassats: ok, my recompilation didn't succeed
2015-10-27T22:15:09Z nyef: Next you'll be telling me I could commit alpha fixes during freeze as well. d-:
2015-10-27T22:15:12Z stassats: i'll leave call_into_c to you
2015-10-27T22:15:42Z stassats: you can break things, come back half a year later, fix them, nobody will even notice
2015-10-27T22:15:52Z stassats: at least that what i did
2015-10-27T22:16:58Z stassats: in the distant past, call_into_c did save OCFP
2015-10-27T22:18:08Z nyef: I already have a fix for undefined_tramp queued up. (-:
2015-10-27T22:18:50Z stassats: https://github.com/sbcl/sbcl/commit/5e2b057871cf47c795c75106899f5fb05dc3397e#diff-eb8e91ea145c89af274cb79abc0c8548L226
2015-10-27T22:22:27Z stassats: nyef: reg_LIP is coming from JAL, the return address
2015-10-27T22:22:52Z stassats: it saves in the stack just like that, gc no bueno
2015-10-27T22:23:16Z stassats: oh
2015-10-27T22:23:18Z stassats: it save reg_LRA
2015-10-27T22:23:54Z stassats: but LRA is not being setup?
2015-10-27T22:24:16Z dougk2 joined #sbcl
2015-10-27T22:25:37Z stassats: ok, it probably doesn't have to push reg_NFP to the stack
2015-10-27T22:26:04Z stassats: but, that's not going to make the GC happy, will it?
2015-10-27T22:27:24Z stassats: ok, the two problems i've indetified, OCFP is not saved at its location, the return address is saved on the number stack by C
2015-10-27T22:29:29Z stassats: so, it's pretty broken overall
2015-10-27T22:29:48Z stassats: man oh man, is this thing trick or what
2015-10-27T22:29:52Z stassats: tricky
2015-10-27T22:31:23Z psilord quit (Quit: Leaving.)
2015-10-27T22:34:15Z nyef: Heh. I still don't know how the MIPS binaries that are available on the website got built.
2015-10-27T22:37:20Z fe[nl]ix: nyef: the question is who built them...
2015-10-27T22:37:33Z stassats: ths?
2015-10-27T22:38:34Z fe[nl]ix: right
2015-10-27T22:39:37Z fe[nl]ix: I vaguely recall he had to go back to 0.9.* to compile them
2015-10-27T22:39:51Z stassats: found a bug in arm64 call_into_c too
2015-10-27T22:40:05Z stassats: i should be using reg_LR for returning from it, not reg_TMP
2015-10-27T22:40:08Z sjl_ joined #sbcl
2015-10-27T22:40:10Z stassats: this thing keeps on giving
2015-10-27T22:42:32Z stassats: nyef: here's my sketch of a fix http://paste.lisp.org/display/157516#3
2015-10-27T22:43:05Z nyef: Okay, thank you.
2015-10-27T22:43:42Z nyef: I guess I need to do a substantial audit of call-in, call-out, backtrace, unwind, interrupt context handling...
2015-10-27T22:44:16Z stassats: i'm also saving reg_LR on the CS on arm64 in alloc_tramp, but it's under PA, so it's just a fixnum
2015-10-27T22:44:39Z nyef: ... And, at the same time, start to figure out what's going on with the ALPHA backend as well. (-:
2015-10-27T22:46:07Z stassats: one last arm64 patch, i hope
2015-10-27T22:49:29Z stassats: but looking at the code is really helpful
2015-10-27T22:58:52Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-27T23:23:31Z pepol joined #sbcl
2015-10-27T23:31:18Z dougk2 quit (Ping timeout: 260 seconds)
2015-10-27T23:34:34Z pepol quit (Ping timeout: 260 seconds)
2015-10-27T23:35:30Z psilord joined #sbcl
2015-10-27T23:40:46Z stassats: nyef: actually, reg_NFP is just a fixnum, so it doesn't have to be reloaded from the stack
2015-10-27T23:42:54Z dougk2 joined #sbcl
2015-10-27T23:47:58Z pepol joined #sbcl
2015-10-27T23:49:40Z attila_lendvai quit (Quit: Leaving.)
2015-10-27T23:51:52Z attila_lendvai joined #sbcl
2015-10-27T23:55:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-28T00:00:03Z attila_lendvai quit (Quit: Leaving.)
2015-10-28T00:01:20Z hydan quit (Ping timeout: 250 seconds)
2015-10-28T00:02:38Z attila_lendvai joined #sbcl
2015-10-28T00:02:38Z attila_lendvai quit (Changing host)
2015-10-28T00:02:38Z attila_lendvai joined #sbcl
2015-10-28T00:10:13Z reb quit (Ping timeout: 246 seconds)
2015-10-28T00:10:31Z reb joined #sbcl
2015-10-28T00:17:53Z dougk2 joined #sbcl
2015-10-28T00:20:48Z hydan joined #sbcl
2015-10-28T00:37:52Z reb quit (Ping timeout: 246 seconds)
2015-10-28T00:38:10Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-28T00:44:02Z dougk2 joined #sbcl
2015-10-28T00:51:42Z attila_lendvai quit (Quit: Leaving.)
2015-10-28T00:52:42Z attila_lendvai joined #sbcl
2015-10-28T01:10:59Z stassats quit (Ping timeout: 252 seconds)
2015-10-28T01:22:38Z nyef cries. The libm functions sin() and sinf() give different results between his MIPS system and his x86-64 system.
2015-10-28T01:29:27Z nyef: And different results _again_ with the debian jessie libc/libm.
2015-10-28T01:32:40Z nyef: Oh, this is just painful to look at.
2015-10-28T01:43:27Z karswell joined #sbcl
2015-10-28T01:48:34Z nyef: Okay, is there anything stupidly-wrong with my methodology, or is the situation really as bad as it looks? http://paste.lisp.org/display/157625
2015-10-28T01:55:35Z myrkraverk quit (Ping timeout: 265 seconds)
2015-10-28T02:17:00Z karswell` joined #sbcl
2015-10-28T02:19:54Z karswell quit (Ping timeout: 272 seconds)
2015-10-28T02:27:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-28T02:30:39Z Xof quit (Ping timeout: 240 seconds)
2015-10-28T02:32:17Z yvm joined #sbcl
2015-10-28T03:36:21Z myrkraverk joined #sbcl
2015-10-28T03:42:47Z pepol quit (Ping timeout: 264 seconds)
2015-10-28T03:44:59Z myrkraverk quit (Ping timeout: 268 seconds)
2015-10-28T03:46:54Z myrkraverk joined #sbcl
2015-10-28T03:48:16Z pepol joined #sbcl
2015-10-28T04:14:58Z karswell` quit (Read error: Connection reset by peer)
2015-10-28T04:15:12Z karswell` joined #sbcl
2015-10-28T04:18:37Z Quadrescence: nyef, do you get different results if inside the loop you create a float and double variable and assign the int to them?
2015-10-28T04:22:37Z Quadrescence: nyef, this looks awful :(
2015-10-28T04:52:05Z karswell` quit (Read error: Connection reset by peer)
2015-10-28T04:52:21Z karswell` joined #sbcl
2015-10-28T04:55:08Z karswell` quit (Remote host closed the connection)
2015-10-28T04:55:20Z karswell` joined #sbcl
2015-10-28T05:12:54Z karswell` quit (Ping timeout: 250 seconds)
2015-10-28T05:30:39Z zacts quit (Ping timeout: 256 seconds)
2015-10-28T05:36:27Z jdz quit (Ping timeout: 252 seconds)
2015-10-28T05:36:34Z jdz joined #sbcl
2015-10-28T05:47:06Z myrkraverk quit (Ping timeout: 260 seconds)
2015-10-28T05:51:14Z myrkraverk joined #sbcl
2015-10-28T05:57:44Z zacts joined #sbcl
2015-10-28T06:16:36Z dougk2 quit (Ping timeout: 250 seconds)
2015-10-28T06:36:19Z myrkraverk quit (Ping timeout: 240 seconds)
2015-10-28T06:41:21Z fridim_ joined #sbcl
2015-10-28T06:41:44Z myrkraverk joined #sbcl
2015-10-28T06:44:13Z dougk2 joined #sbcl
2015-10-28T06:47:10Z myrkraverk quit (Ping timeout: 272 seconds)
2015-10-28T07:00:47Z Shinmera joined #sbcl
2015-10-28T07:23:57Z akkad quit (Quit: Emacs must have died)
2015-10-28T07:56:46Z Cymew joined #sbcl
2015-10-28T07:59:49Z Cymew quit (Remote host closed the connection)
2015-10-28T08:09:10Z ASau quit (Ping timeout: 240 seconds)
2015-10-28T08:11:35Z pepol quit (Ping timeout: 256 seconds)
2015-10-28T08:13:15Z Cymew joined #sbcl
2015-10-28T08:19:06Z pepol joined #sbcl
2015-10-28T09:40:49Z rszeno joined #sbcl
2015-10-28T09:55:22Z hydan quit (Remote host closed the connection)
2015-10-28T09:59:46Z rszeno quit (Quit: Leaving.)
2015-10-28T10:07:08Z carvite quit (Ping timeout: 250 seconds)
2015-10-28T10:08:05Z carvite joined #sbcl
2015-10-28T10:37:27Z stassats joined #sbcl
2015-10-28T11:20:40Z Xof joined #sbcl
2015-10-28T11:28:34Z sjl_ quit (Ping timeout: 260 seconds)
2015-10-28T12:03:36Z sjl joined #sbcl
2015-10-28T13:17:03Z dougk2 quit (Ping timeout: 250 seconds)
2015-10-28T13:24:14Z dougk2 joined #sbcl
2015-10-28T13:38:54Z rszeno joined #sbcl
2015-10-28T13:47:33Z eudoxia joined #sbcl
2015-10-28T13:54:09Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-28T13:58:24Z myrkraverk joined #sbcl
2015-10-28T13:59:47Z rszeno quit (Quit: Leaving.)
2015-10-28T14:12:38Z eudoxia quit (Quit: Leaving)
2015-10-28T14:25:51Z pkhuong: nyef: wtf?
2015-10-28T14:26:09Z pkhuong: I have tables if you want to roll your own ;)
2015-10-28T14:26:21Z pkhuong: but it looks like they do range reduction w/ too little precision
2015-10-28T14:26:28Z pkhuong: probably single float M_PI
2015-10-28T14:27:06Z nyef: pkhuong: I'm still thinking that it might be something weird with the float hardware.
2015-10-28T14:27:24Z nyef: Like the compiler assuming that it's configured one way, and the kernel configuring it another way.
2015-10-28T14:27:33Z pkhuong: libm assume it's round to even
2015-10-28T14:27:40Z pkhuong: or nearest.
2015-10-28T14:27:49Z pkhuong: anything else will result in ridiculous values
2015-10-28T14:28:54Z nyef: Not rounding mode. More, is it 16 registers, or 32?  And are they paired high/low, or more spread out?
2015-10-28T14:30:10Z pkhuong: that doesn't make sense with the result you get though
2015-10-28T14:30:31Z nyef: Doesn't it? If moving values to a floating-point register can occasionally miss the low half?
2015-10-28T14:30:35Z pkhuong: is it deterministic?
2015-10-28T14:30:42Z nyef: I believe it to be deterministic.
2015-10-28T14:30:57Z pkhuong: so not related to scheduling and saving the FP context
2015-10-28T14:31:39Z pkhuong: I don't see what's special about 5 VS 6 in double float representation
2015-10-28T14:37:04Z dougk2 joined #sbcl
2015-10-28T14:39:52Z nyef: So, the upside is that it's not an SBCL issue. The downside is that it breaks SBCL.
2015-10-28T14:40:34Z nyef: Hrm. Maybe I should try this test on IRIX?
2015-10-28T14:50:39Z ZabaQ joined #sbcl
2015-10-28T15:01:22Z nyef: Okay, I stand corrected: It's not stable from run to run.
2015-10-28T15:11:26Z reb joined #sbcl
2015-10-28T15:15:48Z stassats: is that sin which returns bad results?
2015-10-28T15:16:00Z stassats: have you tried a pure c version?
2015-10-28T15:16:40Z nyef: As in... something other than a stub that calls sinf() and sin() in a loop?
2015-10-28T15:16:43Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-28T15:16:52Z pkhuong: nyef: and if you run two concurrently?
2015-10-28T15:17:11Z stassats: we don't use single float libm routines
2015-10-28T15:17:20Z nyef: It's over too fast to coordinate a concurrent execution.
2015-10-28T15:17:29Z stassats: they were even broken in some glibcs
2015-10-28T15:17:30Z pkhuong: nyef: right, so add a loop around
2015-10-28T15:17:39Z pkhuong: stassats: here it's the double version that's borked
2015-10-28T15:18:12Z nyef: stassats: The single float version allows eyeballing for issues without requiring a test oracle.
2015-10-28T15:19:15Z stassats: oh, it's that wrong?
2015-10-28T15:22:10Z nyef: If they're different in the fourth digit or further left, surely there's an issue?
2015-10-28T15:22:45Z nyef: pkhuong: What would a concurrent test tell us?
2015-10-28T15:22:52Z stassats: https://sourceware.org/bugzilla/show_bug.cgi?id=3976 ?
2015-10-28T15:25:54Z nyef: stassats: Not likely: Still in the default rounding mode, and this is a very recent glibc.
2015-10-28T15:26:16Z stassats: let me try your test program
2015-10-28T15:26:35Z nyef: Also wouldn't explain the differences from run to run.
2015-10-28T15:26:50Z nyef: Sure, it should be at the top of http://paste.lisp.org/display/157625
2015-10-28T15:27:05Z stassats: i know, that was another kind of "let me"
2015-10-28T15:27:12Z nyef: Ah, okay.
2015-10-28T15:29:25Z stassats: nothing unusual
2015-10-28T15:29:36Z stassats: except for some printing disrepancies
2015-10-28T15:29:45Z stassats: 22: -0.008851 -0.00885131
2015-10-28T15:29:51Z stassats: 11: -0.999990 -0.99999
2015-10-28T15:30:18Z stassats: well, it doesn't use a real mips fpu anyway
2015-10-28T15:31:53Z nyef: Yeah, and there's also the possibility of a kernel bug.
2015-10-28T15:32:01Z stassats: nyef: can you try other things? cos, pow?
2015-10-28T15:32:03Z dougk2 joined #sbcl
2015-10-28T15:32:47Z stassats: like unsaved fpu registers on context switches?
2015-10-28T15:32:55Z nyef: Something like that.
2015-10-28T15:33:15Z stassats: that would explain things, though the values are near to each other
2015-10-28T15:33:20Z pkhuong: nyef: yeah, it would exercise saving FPU states in context switches
2015-10-28T15:33:36Z nyef: cos, also affected.
2015-10-28T15:34:21Z stassats: so, need something that dumps all fpu register and to compare the results between runs
2015-10-28T15:36:00Z nyef: Actually, I think that my next step might be to swap an IRIX disk into this machine, make sure that it's not the hardware at fault.
2015-10-28T15:36:19Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-28T15:37:06Z stassats: any luck with call_into_c?
2015-10-28T15:38:46Z stassats: i guess i can now try my patch, now that i have time to wait
2015-10-28T15:42:42Z nyef: I haven't even looked at call_into_c. At this point I'm half convinced that most of the failures that I'm seeing are due to this FPU issue.
2015-10-28T15:43:13Z stassats: i can test it, just will take time
2015-10-28T15:43:18Z stassats: not my time, cpu time
2015-10-28T15:46:22Z stassats: why does mips even use $gp, is there not enough registers?
2015-10-28T15:46:49Z stassats: it's used for storing the address of call_into_c, maybe something else too
2015-10-28T15:47:18Z stassats: for jumping to fixups
2015-10-28T15:51:43Z stassats: 20 minutes for make-target-2.sh
2015-10-28T15:59:26Z Cymew quit (Ping timeout: 272 seconds)
2015-10-28T16:01:11Z stassats: at least make-target-2 works, so it's already a good sign
2015-10-28T16:05:02Z dougk2 joined #sbcl
2015-10-28T16:26:00Z zacts quit (Read error: Connection reset by peer)
2015-10-28T16:26:28Z gingerale joined #sbcl
2015-10-28T16:27:14Z zacts joined #sbcl
2015-10-28T16:41:17Z attila_lendvai joined #sbcl
2015-10-28T16:46:52Z akkad joined #sbcl
2015-10-28T16:48:23Z jdz quit (Ping timeout: 256 seconds)
2015-10-28T16:49:44Z jdz joined #sbcl
2015-10-28T17:04:59Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-28T17:27:15Z dougk2 joined #sbcl
2015-10-28T17:29:34Z stassats: nyef: call_into_c looks traceble now, will update the patch
2015-10-28T17:30:25Z sjl quit (Ping timeout: 246 seconds)
2015-10-28T17:30:38Z stassats: nyef: http://paste.lisp.org/display/157516#4
2015-10-28T17:33:12Z Bicyclidine joined #sbcl
2015-10-28T17:42:26Z ZabaQ quit (Ping timeout: 240 seconds)
2015-10-28T17:56:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-28T18:20:24Z dougk2 joined #sbcl
2015-10-28T18:36:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-28T18:38:42Z fikusz quit (Quit: Leaving)
2015-10-28T18:41:13Z fikusz joined #sbcl
2015-10-28T19:03:06Z Bicyclidine quit (Ping timeout: 260 seconds)
2015-10-28T19:04:40Z dougk2 joined #sbcl
2015-10-28T19:09:51Z Bicyclidine joined #sbcl
2015-10-28T19:25:19Z attila_lendvai quit (Quit: Leaving.)
2015-10-28T19:26:19Z attila_lendvai joined #sbcl
2015-10-28T19:27:33Z gingerale quit (Remote host closed the connection)
2015-10-28T19:32:32Z attila_lendvai quit (Quit: Leaving.)
2015-10-28T19:33:28Z attila_lendvai joined #sbcl
2015-10-28T19:33:35Z attila_lendvai quit (Changing host)
2015-10-28T19:33:35Z attila_lendvai joined #sbcl
2015-10-28T19:38:52Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-28T19:39:05Z attila_lendvai quit (Quit: Leaving.)
2015-10-28T19:40:30Z attila_lendvai joined #sbcl
2015-10-28T19:45:11Z dougk2 joined #sbcl
2015-10-28T19:56:43Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-28T20:00:11Z attila_lendvai quit (Quit: Leaving.)
2015-10-28T20:01:07Z attila_lendvai joined #sbcl
2015-10-28T20:01:07Z attila_lendvai quit (Changing host)
2015-10-28T20:01:07Z attila_lendvai joined #sbcl
2015-10-28T20:06:21Z nikki93 joined #sbcl
2015-10-28T20:17:04Z sjl joined #sbcl
2015-10-28T20:28:32Z Bicyclidine quit (Ping timeout: 250 seconds)
2015-10-28T20:40:20Z Bicyclidine joined #sbcl
2015-10-28T20:54:03Z scymtym_ joined #sbcl
2015-10-28T20:57:53Z Bicyclidine quit (Ping timeout: 252 seconds)
2015-10-28T21:06:04Z hustler74 joined #sbcl
2015-10-28T21:09:09Z hustler74 left #sbcl
2015-10-28T21:36:32Z Bicyclidine joined #sbcl
2015-10-28T21:53:30Z attila_lendvai quit (Quit: Leaving.)
2015-10-28T22:01:37Z nikki93 quit (Remote host closed the connection)
2015-10-28T22:08:46Z ASau joined #sbcl
2015-10-28T22:18:46Z fridim_ quit (Ping timeout: 260 seconds)
2015-10-28T22:35:56Z attila_lendvai joined #sbcl
2015-10-28T22:44:43Z Posterdati quit (Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/)
2015-10-28T22:44:57Z Posterdati joined #sbcl
2015-10-28T22:50:13Z dougk2 joined #sbcl
2015-10-28T23:02:27Z nikki93 joined #sbcl
2015-10-28T23:18:49Z stassats: component-name seems to broken with multiple functions in a component
2015-10-28T23:19:09Z stassats: at least when the component gets to debug-dump
2015-10-28T23:20:17Z stassats: (sb-disassem::fun-code (fdefinition (defun foo () (lambda () t)))) => #<code object (LAMBDA () :IN FOO) {10142713DF}>
2015-10-28T23:21:26Z reb quit (Remote host closed the connection)
2015-10-28T23:21:56Z stassats: curiously (defun foo () (lambda () ))  do not go into the same component, while the lambdas in (defun foo () (lambda () (lambda () t))) do
2015-10-28T23:22:44Z stassats: or (lambda () (lambda () (lambda () t))) uses two code components
2015-10-28T23:22:59Z stassats: as does (lambda () (lambda () (lambda () (lambda () t))))
2015-10-28T23:23:01Z stassats: weird stuff
2015-10-28T23:23:19Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-28T23:23:38Z reb` joined #sbcl
2015-10-28T23:23:57Z stassats: (lambda (x) (lambda () x)) gets one component as well
2015-10-28T23:54:55Z dougk2 quit (Ping timeout: 250 seconds)
2015-10-28T23:59:32Z dougk2 joined #sbcl
2015-10-29T00:08:28Z dougk2 quit (Ping timeout: 268 seconds)
2015-10-29T00:09:09Z dougk2 joined #sbcl
2015-10-29T00:31:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-29T00:32:42Z attila_lendvai joined #sbcl
2015-10-29T00:41:44Z Bicyclidine quit (Quit: m)
2015-10-29T02:11:26Z scymtym_ quit (Ping timeout: 240 seconds)
2015-10-29T02:16:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-29T02:19:37Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-29T02:48:35Z dougk2 joined #sbcl
2015-10-29T03:03:46Z whiteline quit (Ping timeout: 260 seconds)
2015-10-29T03:24:54Z stassats quit (Ping timeout: 260 seconds)
2015-10-29T03:47:31Z jlarocco_ joined #sbcl
2015-10-29T04:02:40Z karswell` joined #sbcl
2015-10-29T04:22:49Z foom quit (Ping timeout: 246 seconds)
2015-10-29T04:35:51Z foom joined #sbcl
2015-10-29T04:46:16Z nikki93 quit (Remote host closed the connection)
2015-10-29T05:04:45Z nikki93 joined #sbcl
2015-10-29T05:33:32Z nikki93 quit (Remote host closed the connection)
2015-10-29T05:35:07Z karswell` is now known as karswell
2015-10-29T05:35:10Z nikki93 joined #sbcl
2015-10-29T05:36:59Z karswell quit (Read error: Connection reset by peer)
2015-10-29T05:37:14Z karswell joined #sbcl
2015-10-29T05:55:34Z nikki93 quit (Remote host closed the connection)
2015-10-29T06:23:45Z nikki93 joined #sbcl
2015-10-29T06:39:27Z nzambe quit (Remote host closed the connection)
2015-10-29T06:56:05Z Shinmera joined #sbcl
2015-10-29T07:58:46Z Cymew joined #sbcl
2015-10-29T08:00:05Z Cymew quit (Read error: Connection reset by peer)
2015-10-29T08:15:06Z fridim_ joined #sbcl
2015-10-29T08:22:10Z ASau quit (Ping timeout: 260 seconds)
2015-10-29T08:23:51Z ZabaQ joined #sbcl
2015-10-29T08:26:31Z karswell` joined #sbcl
2015-10-29T08:27:04Z karswell quit (Read error: Connection reset by peer)
2015-10-29T08:43:03Z whiteline joined #sbcl
2015-10-29T08:44:58Z dougk_ quit (Ping timeout: 246 seconds)
2015-10-29T08:58:03Z dougk_ joined #sbcl
2015-10-29T09:03:50Z nikki93 quit (Remote host closed the connection)
2015-10-29T09:14:43Z Cymew joined #sbcl
2015-10-29T09:44:08Z dsp_ quit (Read error: Connection reset by peer)
2015-10-29T09:44:14Z dsp_ joined #sbcl
2015-10-29T09:51:51Z dead_ joined #sbcl
2015-10-29T09:51:51Z DeadTrickster quit (Read error: Connection reset by peer)
2015-10-29T09:55:52Z scymtym_ joined #sbcl
2015-10-29T10:01:38Z dead_ quit (Quit: Leaving)
2015-10-29T10:01:54Z DeadTrickster joined #sbcl
2015-10-29T10:04:27Z nikki93 joined #sbcl
2015-10-29T10:29:19Z attila_lendvai joined #sbcl
2015-10-29T10:29:19Z attila_lendvai quit (Changing host)
2015-10-29T10:29:19Z attila_lendvai joined #sbcl
2015-10-29T11:15:10Z sjl quit (Ping timeout: 260 seconds)
2015-10-29T11:43:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-29T11:47:57Z dougk2 quit (Ping timeout: 256 seconds)
2015-10-29T11:57:55Z dougk2 joined #sbcl
2015-10-29T12:00:37Z stassats joined #sbcl
2015-10-29T12:04:03Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-29T12:27:56Z sjl joined #sbcl
2015-10-29T12:32:43Z dougk2 joined #sbcl
2015-10-29T12:46:25Z dougk2 quit (Ping timeout: 244 seconds)
2015-10-29T12:48:19Z dougk2 joined #sbcl
2015-10-29T12:56:09Z dougk2 quit (Ping timeout: 265 seconds)
2015-10-29T13:00:23Z dougk2 joined #sbcl
2015-10-29T13:05:48Z psilord quit (Quit: Leaving.)
2015-10-29T13:08:11Z eudoxia joined #sbcl
2015-10-29T13:23:26Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-29T14:01:56Z nzambe joined #sbcl
2015-10-29T14:03:08Z rszeno joined #sbcl
2015-10-29T14:07:28Z psilord joined #sbcl
2015-10-29T14:31:49Z DeadTrickster quit (Read error: No route to host)
2015-10-29T14:34:43Z dougk2 joined #sbcl
2015-10-29T14:55:06Z alchemis7 quit (Ping timeout: 240 seconds)
2015-10-29T15:30:12Z Cymew quit (Ping timeout: 244 seconds)
2015-10-29T15:43:52Z attila_lendvai joined #sbcl
2015-10-29T15:43:52Z attila_lendvai quit (Changing host)
2015-10-29T15:43:52Z attila_lendvai joined #sbcl
2015-10-29T15:52:36Z ZabaQ quit (Read error: Connection reset by peer)
2015-10-29T16:34:40Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-29T16:47:40Z nikki93_ joined #sbcl
2015-10-29T16:47:51Z gingerale joined #sbcl
2015-10-29T16:48:02Z nikki93 quit (Read error: Connection reset by peer)
2015-10-29T16:52:02Z nikki93_ quit (Remote host closed the connection)
2015-10-29T16:52:37Z Intensity joined #sbcl
2015-10-29T17:02:24Z dougk2 joined #sbcl
2015-10-29T17:49:54Z rpg joined #sbcl
2015-10-29T18:15:56Z nikki93 joined #sbcl
2015-10-29T18:19:23Z nyef quit (Ping timeout: 264 seconds)
2015-10-29T18:27:43Z rpg quit (Quit: rpg)
2015-10-29T18:38:34Z eudoxia quit (Read error: Connection reset by peer)
2015-10-29T18:42:30Z rpg joined #sbcl
2015-10-29T19:02:39Z nikki93 quit (Remote host closed the connection)
2015-10-29T19:07:26Z sjl quit (Ping timeout: 240 seconds)
2015-10-29T19:19:22Z Bicyclidine joined #sbcl
2015-10-29T19:32:49Z nikki93 joined #sbcl
2015-10-29T19:32:55Z nikki93 quit (Remote host closed the connection)
2015-10-29T19:33:13Z nikki93 joined #sbcl
2015-10-29T19:33:51Z reb` quit (Remote host closed the connection)
2015-10-29T20:12:57Z rszeno quit (Quit: Leaving.)
2015-10-29T20:20:03Z gingerale quit (Remote host closed the connection)
2015-10-29T20:52:38Z lnostdal joined #sbcl
2015-10-29T20:56:06Z scymtym_ quit (Ping timeout: 265 seconds)
2015-10-29T21:08:22Z scymtym_ joined #sbcl
2015-10-29T21:20:58Z lnostdal quit (Ping timeout: 246 seconds)
2015-10-29T21:22:04Z lnostdal joined #sbcl
2015-10-29T21:47:29Z fridim_ quit (Ping timeout: 256 seconds)
2015-10-29T21:54:17Z ASau joined #sbcl
2015-10-29T21:55:37Z lnostdal quit (Ping timeout: 246 seconds)
2015-10-29T21:59:40Z psilord quit (Quit: Leaving.)
2015-10-29T22:07:06Z lnostdal joined #sbcl
2015-10-29T22:12:25Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-29T22:32:25Z dougk2 joined #sbcl
2015-10-29T22:47:39Z lnostdal quit (Ping timeout: 240 seconds)
2015-10-29T22:53:37Z nikki93 quit
2015-10-29T23:05:32Z rpg quit (Quit: rpg)
2015-10-29T23:15:47Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-29T23:17:07Z dougk2 quit (Ping timeout: 252 seconds)
2015-10-29T23:21:58Z alchemis7 joined #sbcl
2015-10-29T23:23:05Z dougk2 joined #sbcl
2015-10-29T23:27:51Z lnostdal joined #sbcl
2015-10-29T23:28:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-29T23:41:02Z psilord joined #sbcl
2015-10-29T23:41:58Z scymtym_ quit (Ping timeout: 272 seconds)
2015-10-29T23:46:51Z dougk2 joined #sbcl
2015-10-29T23:48:17Z Bicyclidine quit (Ping timeout: 252 seconds)
2015-10-29T23:50:18Z Bicyclidine joined #sbcl
2015-10-29T23:56:26Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-29T23:56:48Z sjl joined #sbcl
2015-10-29T23:59:58Z dougk2 joined #sbcl
2015-10-30T00:00:20Z sjl_ joined #sbcl
2015-10-30T00:02:30Z sjl quit (Ping timeout: 240 seconds)
2015-10-30T00:05:50Z dougk2_ joined #sbcl
2015-10-30T00:08:05Z dougk2 quit (Ping timeout: 252 seconds)
2015-10-30T01:01:19Z stassats quit (Ping timeout: 240 seconds)
2015-10-30T01:07:44Z Bicyclidine quit (Quit: mfw)
2015-10-30T01:16:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-30T01:27:11Z sjl_ quit (Ping timeout: 250 seconds)
2015-10-30T01:35:25Z lnostdal quit (Ping timeout: 246 seconds)
2015-10-30T01:46:09Z karswell` is now known as karswell
2015-10-30T01:47:23Z rpg joined #sbcl
2015-10-30T01:48:42Z lnostdal joined #sbcl
2015-10-30T01:57:36Z nicdev is now known as nicdev_
2015-10-30T02:04:13Z rpg quit (Quit: rpg)
2015-10-30T02:22:02Z karswell quit (Read error: Connection reset by peer)
2015-10-30T02:27:37Z oystewh joined #sbcl
2015-10-30T02:27:48Z oystewh: what's the way to handle quiet underflows? i have single floats turning into "#<SINGLE-FLOAT quiet NaN>" when they approach zero (i think) and they throw exceptions/conditions
2015-10-30T02:29:50Z oystewh: uhh nvm
2015-10-30T02:33:58Z hydan joined #sbcl
2015-10-30T02:45:19Z oystewh quit (Quit: Lost terminal)
2015-10-30T02:53:14Z hydan` quit (Remote host closed the connection)
2015-10-30T03:05:36Z jozip quit (Ping timeout: 264 seconds)
2015-10-30T03:05:36Z pootler quit (Ping timeout: 264 seconds)
2015-10-30T03:06:02Z drmeister quit (Ping timeout: 268 seconds)
2015-10-30T03:06:55Z pootler joined #sbcl
2015-10-30T03:08:11Z jozip joined #sbcl
2015-10-30T03:09:11Z dougk2_ quit (Ping timeout: 264 seconds)
2015-10-30T03:12:21Z drmeister joined #sbcl
2015-10-30T03:19:37Z hydan` joined #sbcl
2015-10-30T03:25:35Z sjl joined #sbcl
2015-10-30T03:29:39Z sjl quit (Ping timeout: 240 seconds)
2015-10-30T04:18:59Z pootler quit (Ping timeout: 240 seconds)
2015-10-30T04:19:19Z dougk_ quit (Ping timeout: 240 seconds)
2015-10-30T04:19:39Z dougk_ joined #sbcl
2015-10-30T04:20:12Z pootler joined #sbcl
2015-10-30T04:26:19Z sjl joined #sbcl
2015-10-30T04:30:57Z sjl quit (Ping timeout: 256 seconds)
2015-10-30T05:05:25Z lnostdal quit (Ping timeout: 246 seconds)
2015-10-30T05:06:21Z lnostdal joined #sbcl
2015-10-30T06:27:02Z sjl joined #sbcl
2015-10-30T06:31:19Z sjl quit (Ping timeout: 240 seconds)
2015-10-30T06:46:39Z Shinmera joined #sbcl
2015-10-30T06:54:47Z fridim_ joined #sbcl
2015-10-30T07:01:40Z igajsin joined #sbcl
2015-10-30T07:19:03Z ASau quit (Ping timeout: 250 seconds)
2015-10-30T07:58:27Z igajsin quit (Remote host closed the connection)
2015-10-30T08:03:08Z hydan quit (Ping timeout: 250 seconds)
2015-10-30T08:27:48Z sjl joined #sbcl
2015-10-30T08:32:26Z sjl quit (Ping timeout: 260 seconds)
2015-10-30T09:28:35Z sjl joined #sbcl
2015-10-30T09:32:49Z sjl quit (Ping timeout: 246 seconds)
2015-10-30T09:38:10Z hydan` quit (Remote host closed the connection)
2015-10-30T09:38:35Z hydan joined #sbcl
2015-10-30T09:50:15Z attila_lendvai joined #sbcl
2015-10-30T09:50:40Z sigjuice quit (Ping timeout: 246 seconds)
2015-10-30T09:50:43Z scymtym quit (Remote host closed the connection)
2015-10-30T09:50:49Z scymtym joined #sbcl
2015-10-30T09:53:18Z sigjuice joined #sbcl
2015-10-30T09:58:27Z hydan quit (Remote host closed the connection)
2015-10-30T10:05:06Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-30T10:06:05Z scymtym__ joined #sbcl
2015-10-30T10:06:20Z scymtym__ is now known as scymtym_
2015-10-30T10:42:20Z hydan joined #sbcl
2015-10-30T11:29:18Z sjl joined #sbcl
2015-10-30T11:33:26Z sjl quit (Ping timeout: 240 seconds)
2015-10-30T11:44:45Z attila_lendvai joined #sbcl
2015-10-30T11:49:45Z dougk2 joined #sbcl
2015-10-30T12:10:59Z lnostdal quit (Ping timeout: 240 seconds)
2015-10-30T12:22:46Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-30T12:25:12Z Gu_______ joined #sbcl
2015-10-30T12:25:31Z Gu_______ quit (Client Quit)
2015-10-30T12:26:42Z anthracite joined #sbcl
2015-10-30T12:32:06Z lnostdal joined #sbcl
2015-10-30T12:36:47Z nyef joined #sbcl
2015-10-30T12:38:14Z scymtym_ quit (Read error: Connection reset by peer)
2015-10-30T12:38:15Z scymtym quit (Remote host closed the connection)
2015-10-30T12:38:46Z eudoxia joined #sbcl
2015-10-30T12:44:40Z scymtym joined #sbcl
2015-10-30T12:45:41Z attila_lendvai joined #sbcl
2015-10-30T12:45:41Z attila_lendvai quit (Changing host)
2015-10-30T12:45:41Z attila_lendvai joined #sbcl
2015-10-30T13:22:42Z stassats joined #sbcl
2015-10-30T13:24:19Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-30T13:30:07Z sjl joined #sbcl
2015-10-30T13:34:36Z sjl quit (Ping timeout: 255 seconds)
2015-10-30T13:43:08Z sjl_ joined #sbcl
2015-10-30T13:47:48Z sjl_ quit (Read error: Connection reset by peer)
2015-10-30T13:48:26Z sjl joined #sbcl
2015-10-30T14:14:46Z lnostdal quit (Ping timeout: 240 seconds)
2015-10-30T14:26:22Z lnostdal joined #sbcl
2015-10-30T14:33:03Z dougk2 joined #sbcl
2015-10-30T14:37:42Z _iwc joined #sbcl
2015-10-30T14:42:12Z jlarocco_ quit (Quit: Leaving)
2015-10-30T14:44:28Z reb joined #sbcl
2015-10-30T14:50:33Z _iwc quit (Killed (Sigyn (Spam is off topic on freenode.)))
2015-10-30T15:17:34Z sjl quit (Read error: Connection reset by peer)
2015-10-30T15:18:14Z sjl joined #sbcl
2015-10-30T15:30:24Z eudoxia quit (Quit: Leaving)
2015-10-30T16:15:19Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-30T16:18:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-30T16:18:46Z attila_lendvai joined #sbcl
2015-10-30T16:18:46Z attila_lendvai quit (Changing host)
2015-10-30T16:18:46Z attila_lendvai joined #sbcl
2015-10-30T16:22:07Z anthracite quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
2015-10-30T16:28:51Z gingerale joined #sbcl
2015-10-30T16:36:19Z whiteline quit (Ping timeout: 268 seconds)
2015-10-30T16:41:51Z dougk2 joined #sbcl
2015-10-30T16:47:19Z guicho joined #sbcl
2015-10-30T16:48:30Z guicho_ joined #sbcl
2015-10-30T16:48:30Z guicho_ quit (Client Quit)
2015-10-30T16:48:52Z guicho quit (Quit: さようなら)
2015-10-30T16:59:42Z sjl quit (Read error: Connection reset by peer)
2015-10-30T17:00:40Z sjl joined #sbcl
2015-10-30T18:03:31Z sjl quit (Ping timeout: 252 seconds)
2015-10-30T18:06:39Z dougk2 quit (Ping timeout: 240 seconds)
2015-10-30T18:07:24Z whiteline joined #sbcl
2015-10-30T18:25:12Z dougk2 joined #sbcl
2015-10-30T18:28:53Z kumbiKambi joined #sbcl
2015-10-30T18:30:16Z kumbiKambi quit (Quit: Leaving)
2015-10-30T18:30:25Z dougk2 quit (Ping timeout: 246 seconds)
2015-10-30T19:19:19Z lnostdal quit (Ping timeout: 240 seconds)
2015-10-30T19:40:46Z lnostdal joined #sbcl
2015-10-30T19:56:40Z dougk2 joined #sbcl
2015-10-30T19:59:28Z dougk2 quit (Remote host closed the connection)
2015-10-30T20:24:11Z gingerale quit (Remote host closed the connection)
2015-10-30T20:29:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-30T20:32:55Z Bicyclidine joined #sbcl
2015-10-30T20:53:49Z sjl joined #sbcl
2015-10-30T21:01:00Z Bicyclidine quit (Quit: leaving)
2015-10-30T21:12:07Z anthracite joined #sbcl
2015-10-30T21:12:18Z anthracite quit (Remote host closed the connection)
2015-10-30T21:15:39Z dougk_ quit (Ping timeout: 240 seconds)
2015-10-30T21:18:25Z scymtym_ joined #sbcl
2015-10-30T22:06:09Z ASau joined #sbcl
2015-10-30T22:13:02Z whiteline quit (Read error: Connection reset by peer)
2015-10-30T22:13:49Z whiteline joined #sbcl
2015-10-30T22:34:58Z attila_lendvai joined #sbcl
2015-10-30T22:38:39Z sjl quit (Ping timeout: 240 seconds)
2015-10-30T22:45:30Z sjl joined #sbcl
2015-10-30T23:09:22Z fridim_ quit (Ping timeout: 246 seconds)
2015-10-30T23:13:33Z lnostdal_ joined #sbcl
2015-10-30T23:16:43Z lnostdal quit (Ping timeout: 246 seconds)
2015-10-30T23:19:26Z attila_lendvai quit (Ping timeout: 240 seconds)
2015-10-30T23:35:07Z akkad quit (Excess Flood)
2015-10-30T23:35:20Z akkad joined #sbcl
2015-10-30T23:44:58Z sjl quit (Read error: Connection reset by peer)
2015-10-30T23:45:33Z sjl joined #sbcl
2015-10-30T23:57:02Z dougk_ joined #sbcl
2015-10-30T23:57:40Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-10-31T00:26:26Z sjl quit (Ping timeout: 260 seconds)
2015-10-31T00:32:31Z sjl joined #sbcl
2015-10-31T00:51:33Z sjl quit (Ping timeout: 256 seconds)
2015-10-31T01:38:31Z sjl joined #sbcl
2015-10-31T02:37:19Z lnostdal_ quit (Ping timeout: 240 seconds)
2015-10-31T02:54:35Z lnostdal_ joined #sbcl
2015-10-31T02:55:07Z scymtym_ quit (Ping timeout: 246 seconds)
2015-10-31T04:45:39Z psy_ joined #sbcl
2015-10-31T04:46:16Z psy_ quit (Max SendQ exceeded)
2015-10-31T04:46:48Z psy_ joined #sbcl
2015-10-31T05:03:47Z psy_ quit (Ping timeout: 264 seconds)
2015-10-31T05:21:26Z eschatologist joined #sbcl
2015-10-31T05:37:19Z Bike quit (Quit: Lost terminal)
2015-10-31T05:37:39Z Bike joined #sbcl
2015-10-31T05:41:13Z eschatologist quit (Quit: Textual IRC Client: www.textualapp.com)
2015-10-31T06:46:07Z gingerale joined #sbcl
2015-10-31T06:47:27Z Shinmera joined #sbcl
2015-10-31T07:03:11Z jozip quit (Ping timeout: 264 seconds)
2015-10-31T07:03:37Z jozip joined #sbcl
2015-10-31T07:06:13Z lnostdal__ joined #sbcl
2015-10-31T07:09:22Z lnostdal_ quit (Ping timeout: 250 seconds)
2015-10-31T08:10:30Z fridim_ joined #sbcl
2015-10-31T08:32:50Z stassats quit (Ping timeout: 240 seconds)
2015-10-31T08:59:44Z scymtym_ joined #sbcl
2015-10-31T09:00:05Z DeadTrickster joined #sbcl
2015-10-31T09:03:19Z hydan quit (Remote host closed the connection)
2015-10-31T09:20:30Z edgar-rft joined #sbcl
2015-10-31T10:54:39Z attila_lendvai joined #sbcl
2015-10-31T11:13:50Z ASau quit (Ping timeout: 240 seconds)
2015-10-31T11:25:29Z sjl quit (Read error: Connection reset by peer)
2015-10-31T11:27:00Z sjl joined #sbcl
2015-10-31T12:28:18Z stassats joined #sbcl
2015-10-31T13:19:00Z stassats` joined #sbcl
2015-10-31T13:20:39Z stassats quit (Ping timeout: 255 seconds)
2015-10-31T13:26:43Z hydan joined #sbcl
2015-10-31T14:08:19Z scymtym_ quit (Ping timeout: 240 seconds)
2015-10-31T15:03:47Z nyef: stassats`: Do you have a built SBCL/MIPS handy, and does it backtrace through alien code?
2015-10-31T15:04:14Z stassats`: it does, but i'm waiting for clasp to build, don't want scew timings
2015-10-31T15:05:22Z nyef: That's fine. Once that's done, could you please see if (defun foo () (log 123 (eval 0))) compiles without error, and if it doesn't can you obtain a backtrace for me?
2015-10-31T15:05:42Z stassats`: it compiled without errors before
2015-10-31T15:06:09Z nyef: That's... disconcerting.
2015-10-31T15:06:33Z stassats`: apply http://paste.lisp.org/display/157516#4 and you'll have backtraces too
2015-10-31T15:07:03Z nyef: Isn't that for call_into_c, not call_into_lisp?
2015-10-31T15:07:27Z stassats`: call_into_lisp is not broken
2015-10-31T15:07:38Z nyef: My backtrace stops at call_into_lisp.
2015-10-31T15:07:55Z stassats`: sure, but the call_into_lisp is coming from call_into_c
2015-10-31T15:08:06Z nyef: Ah, okay. I'll give it a try then.
2015-10-31T15:12:19Z clique joined #sbcl
2015-10-31T15:15:33Z nyef: Wonderful! That defun still fails, but I have a backtrace. Thank you!
2015-10-31T15:17:31Z stassats`: it only took what, 10 years?
2015-10-31T15:18:47Z nyef: Something like that.
2015-10-31T15:18:58Z stassats`: Apr 30, 2005
2015-10-31T15:30:00Z nyef: Do you want to do the commit for this backtrace thing, or should I?
2015-10-31T15:30:22Z stassats`: well, let's wait until the thaw
2015-10-31T15:30:37Z stassats`: i'll commit it no problem
2015-10-31T15:30:48Z stassats`: but that GP is bothering me
2015-10-31T15:31:10Z stassats`: a) why do we touch GP b) can it have bad alignment?
2015-10-31T15:31:20Z clique quit (Quit: q)
2015-10-31T15:32:38Z stassats`: defmethod macro-expander is a forest and i'm lost
2015-10-31T15:33:31Z stassats`: if i change to that m-v-call, maybe we can stop code-walking and just always install call-next-method?
2015-10-31T15:49:14Z Xof: thaw coming soon
2015-10-31T15:50:51Z nyef: ... x86-64 floating-point mode setting is in C? What?
2015-10-31T16:14:04Z nyef: Okay, I think I see why I'm getting hit by a divide-by-zero here.
2015-10-31T16:15:36Z stassats`: 1.3.0, yay
2015-10-31T16:15:52Z stassats`: what will it take to get it to 2.0?
2015-10-31T16:16:04Z nyef: Another four years.
2015-10-31T16:17:02Z stassats`: and there goes my weekend
2015-10-31T16:21:47Z Xof: bah!  I've missed the perfect opportunity to retire from SBCL release maintenance
2015-10-31T16:21:52Z Xof: "Where we're going, we don't need Rhodes"
2015-10-31T16:22:04Z attila_lendvai: :D
2015-10-31T16:27:26Z stassats`: time to produce some new table columns
2015-10-31T16:28:25Z ASau joined #sbcl
2015-10-31T16:33:54Z nyef: Lovely. Looks like MIPS is the odd architecture out for this floating-point thing (having checked the x86, PPC, SPARC, Alpha, and PA-RISC manuals that I have on hand).
2015-10-31T16:34:57Z nyef: Now I just have to figure out how to fix it.
2015-10-31T16:36:06Z stassats`: i'll need to introduce a new feature, number-stack-grows-downward
2015-10-31T16:36:27Z stassats`: but how do we determine that alredy?
2015-10-31T16:36:58Z stassats`: we probably don't, allocate-number-stack is just different for each backend
2015-10-31T16:37:42Z stassats`: there's a C macro i defined some time ago
2015-10-31T16:38:58Z stassats`: only for hppa it's growing up
2015-10-31T16:39:46Z drmeister quit (Ping timeout: 240 seconds)
2015-10-31T16:40:08Z frankS2 quit (Ping timeout: 240 seconds)
2015-10-31T16:40:11Z stassats`: so maybe i'll just do that, only push number-stack-grows-upward for hppa
2015-10-31T16:40:27Z joshe quit (Ping timeout: 240 seconds)
2015-10-31T16:40:46Z hydan quit (Ping timeout: 240 seconds)
2015-10-31T16:40:47Z pchrist quit (Ping timeout: 240 seconds)
2015-10-31T16:41:08Z chris2 quit (Ping timeout: 240 seconds)
2015-10-31T16:42:43Z stassats`: one more column at http://sbcl.org/platform-table.html
2015-10-31T16:42:49Z frankS2 joined #sbcl
2015-10-31T16:43:21Z drmeister joined #sbcl
2015-10-31T16:43:28Z pchrist joined #sbcl
2015-10-31T16:46:09Z akkad quit (Excess Flood)
2015-10-31T16:46:23Z nyef: Hunh. There's also a 32-bit assumption on the floating-point mode control width.
2015-10-31T16:46:39Z nyef: Guess that might explain why the Alpha version is specially coded...
2015-10-31T16:54:11Z akkad joined #sbcl
2015-10-31T16:56:16Z chris2 joined #sbcl
2015-10-31T16:56:22Z nyef: Looks like I need to make up six new serial cables. One to connect my terminal server to my SPARC box (an RJ45 on each end, possible pin permutations), and five for various other boxes (RJ45 on one end, DB9F on the other).
2015-10-31T17:13:45Z nyef: Hrm. Or I could try to get the serial driver working on my O350 and just use null-modem cables for most of the damage...
2015-10-31T17:21:42Z nyef: So, what are our goals for SBCL 1.4 or 2.0?
2015-10-31T17:23:30Z stassats`: no bugs
2015-10-31T17:23:44Z nyef: Hah!
2015-10-31T17:37:42Z stassats`: so, upward growing stack offsets point to the beginning of the word?
2015-10-31T17:38:07Z stassats`: i don't need any special tricks for sub-access-debug-var-slot for that
2015-10-31T17:40:46Z stassats`: so i don't really need number-stack-grows-upward
2015-10-31T17:44:51Z nyef: Okay! Got the compiler.pure.lisp file to complete, even if it DOES have some failures.
2015-10-31T17:45:04Z nyef: Running the full test suite now.
2015-10-31T17:47:26Z stassats`: ok, can access stack floats, but not for escaped floats
2015-10-31T17:51:18Z whiteline quit (Ping timeout: 260 seconds)
2015-10-31T17:53:05Z jackdaniel: congrats the new release :-)
2015-10-31T18:20:00Z stassats`: context-float-register is either  not implemented or implemented in a not suitable way
2015-10-31T18:28:44Z stassats`: got escaped floats on x86-64, that was beyond trivial
2015-10-31T18:29:48Z nyef: Ugh. I'm going to be chasing down bizarre MIPS behavior for a while, aren't I?
2015-10-31T18:35:04Z nyef: Whee... Another build-and-test cycle.
2015-10-31T18:52:52Z stassats`: ok, x87 floats seem to be in some strange format
2015-10-31T18:54:16Z stassats`: short int significand[4] short int exponent
2015-10-31T18:56:03Z stassats`: i suppose i can add some vops for converting this
2015-10-31T18:56:11Z nyef: 80 bits, sounds about right.
2015-10-31T18:56:55Z stassats`: but it's not like xmm, where the lower half has a single float
2015-10-31T18:57:57Z stassats` dusts off the intel manual
2015-10-31T18:58:16Z stassats`: (in portable dust format)
2015-10-31T18:59:49Z stassats`: ok, let's try some vops
2015-10-31T19:00:38Z stassats`: there's sap-ref-long
2015-10-31T19:00:48Z stassats`: what am i going to do with it, though?
2015-10-31T19:01:19Z stassats`: but it gives the right result, so there's that
2015-10-31T19:01:50Z stassats`: without #-long-float
2015-10-31T19:07:50Z gingerale quit (Ping timeout: 240 seconds)
2015-10-31T19:25:19Z sjl quit (Read error: Connection reset by peer)
2015-10-31T19:25:59Z sjl joined #sbcl
2015-10-31T19:32:39Z whiteline joined #sbcl
2015-10-31T19:33:03Z rszeno joined #sbcl
2015-10-31T19:39:15Z stassats`: arm64 doesn't seem to be carrying float registers
2015-10-31T19:40:30Z stassats`: not in sigcontext
2015-10-31T19:44:02Z rpg joined #sbcl
2015-10-31T19:44:31Z stassats`: debugger float access may have become broken on some combinations, but that'll motivate to get it fixed
2015-10-31T19:44:47Z stassats`: and escaped floats are seen rarely
2015-10-31T19:49:55Z sjl quit (Read error: Connection reset by peer)
2015-10-31T19:51:20Z sjl joined #sbcl
2015-10-31T20:03:25Z sjl quit (Ping timeout: 244 seconds)
2015-10-31T20:03:25Z abeaumont_ quit (Read error: Connection reset by peer)
2015-10-31T20:05:12Z sjl joined #sbcl
2015-10-31T20:07:57Z abeaumont_ joined #sbcl
2015-10-31T20:08:30Z jlarocco_ joined #sbcl
2015-10-31T20:08:52Z nyef: http://paste.lisp.org/display/157516#5 <-- new MIPS test results.
2015-10-31T20:28:23Z sjl quit (Ping timeout: 264 seconds)
2015-10-31T20:29:01Z sjl joined #sbcl
2015-10-31T20:33:28Z rszeno quit (Quit: Leaving.)
2015-10-31T20:39:09Z scymtym_ joined #sbcl
2015-10-31T20:44:09Z stassats`: my 900th commit
2015-10-31T20:51:55Z nyef: Whee... ANOTHER build-and-test cycle.
2015-10-31T20:56:08Z nyef: Hrm... I wonder if I can pass WHN for commits by the end of the year?
2015-10-31T20:56:40Z akkad quit (Excess Flood)
2015-10-31T20:58:43Z akkad joined #sbcl
2015-10-31T21:01:07Z sjl quit (Ping timeout: 265 seconds)
2015-10-31T21:02:43Z sjl joined #sbcl
2015-10-31T21:16:56Z nicdev_ quit (K-Lined)
2015-10-31T21:23:08Z nicdev joined #sbcl
2015-10-31T21:40:01Z attila_lendvai: if anyone feels the urge to review some error handling code, this is still waiting and I think it's a patch that can potentially save a lot of developer-debugging-time: https://github.com/sbcl/sbcl/pull/3
2015-10-31T21:40:16Z attila_lendvai: or at least it did for me
2015-10-31T21:41:16Z stassats`: i guess i don't have infinite errors often enough
2015-10-31T21:42:43Z attila_lendvai: me neither, but then it's badly needed. e.g. when a service process dies every few times without much useful printed. or when slime/sldb gets hit in the wrong part.
2015-10-31T21:43:49Z attila_lendvai: I have already helped debug one mysterious slime bug with this patch by producing a backtrace on the bug report
2015-10-31T21:53:39Z sjl quit (Ping timeout: 255 seconds)
2015-10-31T21:58:23Z edgar-rft quit (Quit: edgar-rft)
2015-10-31T22:07:23Z nyef sighs.
2015-10-31T22:07:44Z nyef: Either this sb-sprof thing takes quite a while to run, or it occasionally drops into an infinite loop.
2015-10-31T22:07:56Z nyef: Or it occasionally takes quite a while to run.
2015-10-31T22:08:07Z dougk2 joined #sbcl
2015-10-31T22:09:26Z flip214: it might also drop into an infinite run - that's what I heard.
2015-10-31T22:11:45Z SamSkulls joined #sbcl
2015-10-31T22:26:25Z stassats`: "GNU Hurd 0.7 has been released"
2015-10-31T22:26:32Z stassats`: someone here was porting sbcl to hurd
2015-10-31T22:26:37Z stassats`: i wonder what came out of it
2015-10-31T22:43:24Z nyef: Probably had some trouble with infinite recursion.
2015-10-31T22:52:37Z rpg quit (Quit: rpg)
2015-10-31T23:09:53Z sjl joined #sbcl
2015-10-31T23:12:25Z nyef: lp 486812
2015-10-31T23:12:25Z specbot: https://bugs.launchpad.net/bugs/486812
2015-10-31T23:14:12Z nyef: Hrm. That's odd...
2015-10-31T23:25:49Z nyef: Okay, I found ONE test failure that has an underlying cause that was already on my list to fix.
2015-10-31T23:27:10Z sjl quit (Ping timeout: 272 seconds)
2015-10-31T23:29:58Z sjl joined #sbcl
2015-10-31T23:39:44Z stassats`: fetching floats from context is so different and poorly document, i give up
2015-10-31T23:39:57Z stassats`: just added support for windows and linux x86oids
2015-10-31T23:41:08Z stassats`: somebody more motivated would have to do the rest
2015-10-31T23:43:25Z stassats`: i like https://github.com/torvalds/linux/blob/master/arch/arm64/include/uapi/asm/sigcontext.h#L32 in particular
2015-10-31T23:43:49Z stassats`: how the hell do i access them?
2015-10-31T23:46:08Z stassats`: i have to read kernel sources to figure it out
2015-10-31T23:46:13Z stassats`: great API, what can i say
2015-10-31T23:48:39Z stassats` is particularly discombobulated
2015-10-31T23:50:57Z sjl quit (Read error: Connection reset by peer)
2015-10-31T23:52:24Z sjl joined #sbcl
2015-10-31T23:54:11Z nyef: Okay, what the heck is going on with FLOAT-NAN-P and FLOAT-TRAPPING-NAN-P (code/float.lisp)?
2015-10-31T23:58:17Z nyef: I'm having trouble understanding the logic, but it looks superficially bogus.