2017-03-01T00:07:44Z TruePika joined #sbcl 2017-03-01T00:08:35Z TruePika: Having an issue with http://pastebin.ca/3774186 unable to optimize becasue: could not optimize away %SAP-ALIEN: forced to do runtime allocation of alien-value structure 2017-03-01T00:09:12Z stassats: yes, nothing you can do about it, really 2017-03-01T00:09:13Z Colleen: stassats: drmeister said 9 hours, 25 minutes, 49 seconds ago: I have a working version of trivial-gray-streams - what was the thing that you were trying to get working that had that as a dependency? I pushed changes to testing and here is the changed trivial-gray-streams: https://github.com/clasp-developers/trivial-gray-streams.git 2017-03-01T00:09:37Z TruePika: how would I suppress the note? 2017-03-01T00:09:45Z stassats: except that don't use sb-unix 2017-03-01T00:10:53Z stassats: it's internal and can and will break 2017-03-01T00:11:11Z TruePika: how would you suggest polling a socket then? 2017-03-01T00:11:18Z stassats: iolib 2017-03-01T00:11:38Z TruePika: part of SBCL, or on QL? 2017-03-01T00:11:41Z jamtho quit (Ping timeout: 260 seconds) 2017-03-01T00:11:46Z akkad: ql 2017-03-01T00:11:46Z stassats: the latter 2017-03-01T00:12:02Z akkad: you need to install libfixposix0 2017-03-01T00:23:24Z sjl quit (Ping timeout: 260 seconds) 2017-03-01T00:27:09Z TruePika: stassats: I don't really see anything useful in there 2017-03-01T00:32:14Z cromachina joined #sbcl 2017-03-01T01:05:11Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-01T01:10:15Z TruePika left #sbcl 2017-03-01T02:41:32Z Xof quit (Ping timeout: 260 seconds) 2017-03-01T02:49:36Z karswell joined #sbcl 2017-03-01T02:52:34Z stassats quit (Ping timeout: 264 seconds) 2017-03-01T02:53:28Z fiddlerwoaroof quit (Ping timeout: 255 seconds) 2017-03-01T03:27:44Z salva0 quit (Ping timeout: 260 seconds) 2017-03-01T04:04:05Z salva joined #sbcl 2017-03-01T04:32:50Z bpanthi joined #sbcl 2017-03-01T04:52:24Z jack_rabbit joined #sbcl 2017-03-01T05:01:38Z bpanthi quit (Quit: bpanthi) 2017-03-01T05:30:16Z fiddlerwoaroof joined #sbcl 2017-03-01T06:15:05Z shka_ joined #sbcl 2017-03-01T06:27:59Z nyef` joined #sbcl 2017-03-01T06:29:41Z oleo quit (Quit: Leaving) 2017-03-01T06:29:44Z nyef quit (Ping timeout: 260 seconds) 2017-03-01T06:29:59Z neuronsong quit (Quit: Textual IRC Client: www.textualapp.com) 2017-03-01T06:38:14Z akkad quit (Excess Flood) 2017-03-01T06:44:48Z akkad joined #sbcl 2017-03-01T07:05:25Z shka_ quit (Ping timeout: 246 seconds) 2017-03-01T07:07:16Z jamtho joined #sbcl 2017-03-01T07:12:34Z ym joined #sbcl 2017-03-01T07:13:20Z igajsin joined #sbcl 2017-03-01T07:15:21Z jamtho_ joined #sbcl 2017-03-01T07:17:38Z jamtho quit (Ping timeout: 240 seconds) 2017-03-01T07:19:45Z ym quit (Quit: Leaving) 2017-03-01T07:33:09Z angavrilov joined #sbcl 2017-03-01T07:47:46Z gingerale joined #sbcl 2017-03-01T07:50:38Z jamtho_ quit (Ping timeout: 240 seconds) 2017-03-01T07:51:07Z fiddlerwoaroof quit (Ping timeout: 240 seconds) 2017-03-01T08:17:46Z Xof joined #sbcl 2017-03-01T08:57:10Z edgar-rft joined #sbcl 2017-03-01T09:31:43Z alandipert quit (Ping timeout: 255 seconds) 2017-03-01T09:36:22Z alandipert joined #sbcl 2017-03-01T09:47:52Z jamtho_ joined #sbcl 2017-03-01T09:53:24Z Bike quit (Quit: sleep) 2017-03-01T10:08:48Z jamtho joined #sbcl 2017-03-01T10:10:00Z jamtho_ quit (Ping timeout: 260 seconds) 2017-03-01T10:16:10Z attila_lendvai joined #sbcl 2017-03-01T10:16:10Z attila_lendvai quit (Changing host) 2017-03-01T10:16:10Z attila_lendvai joined #sbcl 2017-03-01T10:17:32Z sjl joined #sbcl 2017-03-01T10:20:24Z jamtho quit (Remote host closed the connection) 2017-03-01T10:20:53Z jamtho joined #sbcl 2017-03-01T10:33:57Z jamtho quit (Remote host closed the connection) 2017-03-01T10:34:26Z jamtho joined #sbcl 2017-03-01T10:59:40Z sjl quit (Read error: Connection reset by peer) 2017-03-01T11:00:21Z sjl joined #sbcl 2017-03-01T11:10:33Z jamtho_ joined #sbcl 2017-03-01T11:11:18Z jamtho quit (Ping timeout: 268 seconds) 2017-03-01T11:14:44Z sjl_ joined #sbcl 2017-03-01T11:16:37Z sjl quit (Ping timeout: 240 seconds) 2017-03-01T11:24:17Z sjl joined #sbcl 2017-03-01T11:25:57Z sjl_ quit (Read error: Connection reset by peer) 2017-03-01T11:26:59Z jamtho_ quit (Remote host closed the connection) 2017-03-01T11:27:18Z jamtho_ joined #sbcl 2017-03-01T11:47:02Z jamtho__ joined #sbcl 2017-03-01T11:49:52Z jamtho_ quit (Ping timeout: 260 seconds) 2017-03-01T11:52:50Z sjl quit (Ping timeout: 260 seconds) 2017-03-01T11:54:58Z jamtho__ quit (Ping timeout: 264 seconds) 2017-03-01T12:27:10Z sjl joined #sbcl 2017-03-01T12:30:57Z salv0 joined #sbcl 2017-03-01T13:09:06Z schjetne joined #sbcl 2017-03-01T13:54:02Z eudoxia joined #sbcl 2017-03-01T14:15:03Z m00natic joined #sbcl 2017-03-01T14:22:10Z cromachina quit (Read error: Connection reset by peer) 2017-03-01T14:27:19Z stassats joined #sbcl 2017-03-01T14:38:54Z angavrilov quit (Remote host closed the connection) 2017-03-01T14:39:19Z angavrilov joined #sbcl 2017-03-01T14:44:00Z DGASAU quit (Read error: Connection reset by peer) 2017-03-01T14:44:26Z DGASAU joined #sbcl 2017-03-01T14:50:50Z sjl quit (Ping timeout: 268 seconds) 2017-03-01T15:23:33Z edgar-rft quit (Quit: edgar-rft) 2017-03-01T15:40:21Z sjl joined #sbcl 2017-03-01T15:45:42Z oleo joined #sbcl 2017-03-01T15:47:55Z sjl quit (Ping timeout: 260 seconds) 2017-03-01T16:04:41Z bpanthi joined #sbcl 2017-03-01T16:05:31Z bpanthi: i get System "sb-rotate-byte" not found error when (ql:quickload :ceramic) 2017-03-01T16:11:55Z scymtym: bpanthi: does the process leading to this error involve saving and restarting an executable or a core? 2017-03-01T16:12:22Z bpanthi: no 2017-03-01T16:13:39Z scymtym: did you update anything while the SBCL process producing the error was running and not restart it after the upgrade? 2017-03-01T16:14:24Z bpanthi: no. 2017-03-01T16:17:26Z scymtym: i don't know ceramic, so i don't have any more specific ideas, sorry 2017-03-01T16:18:17Z bpanthi: strange thing is (require 'sb-rotate-byte) returns ("SB-ROTATE-BYTE") 2017-03-01T16:30:16Z bpanthi: (push #p"/usr/local/lib/sbcl/contrib/" asdf:*central-registry*) solved the problem 2017-03-01T16:42:07Z stassats: bpanthi: is your sbcl really really old? 2017-03-01T16:42:40Z bpanthi: no its 1.3.5 2017-03-01T16:43:24Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-01T16:46:17Z bpanthi: (asdf:clear-configuration) also solves the problem 2017-03-01T16:55:22Z akkad quit (Excess Flood) 2017-03-01T16:57:17Z akkad joined #sbcl 2017-03-01T16:59:54Z bpanthi quit (Quit: bpanthi) 2017-03-01T17:13:11Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-01T17:28:15Z Bike joined #sbcl 2017-03-01T17:49:19Z m00natic quit (Remote host closed the connection) 2017-03-01T17:50:52Z akkad quit (Excess Flood) 2017-03-01T17:55:47Z akkad joined #sbcl 2017-03-01T17:59:29Z akkad quit (Excess Flood) 2017-03-01T18:00:41Z fiddlerwoaroof joined #sbcl 2017-03-01T18:06:47Z akkad joined #sbcl 2017-03-01T18:15:48Z edgar-rft joined #sbcl 2017-03-01T18:32:02Z sjl joined #sbcl 2017-03-01T18:41:26Z shka_ joined #sbcl 2017-03-01T19:34:25Z eudoxia quit (Read error: Connection reset by peer) 2017-03-01T19:47:11Z jibanes quit (Ping timeout: 252 seconds) 2017-03-01T19:48:46Z jibanes joined #sbcl 2017-03-01T20:53:05Z kini quit (*.net *.split) 2017-03-01T20:56:04Z attila_lendvai joined #sbcl 2017-03-01T20:58:05Z sjl quit (Ping timeout: 240 seconds) 2017-03-01T21:09:22Z shka_ quit (Ping timeout: 264 seconds) 2017-03-01T21:09:34Z angavrilov quit (Remote host closed the connection) 2017-03-01T21:14:08Z shka_ joined #sbcl 2017-03-01T21:20:02Z shka_ quit (Read error: Connection reset by peer) 2017-03-01T21:31:57Z igajsin quit (Ping timeout: 240 seconds) 2017-03-01T21:50:27Z salv0 quit (Remote host closed the connection) 2017-03-01T21:50:41Z gingerale quit (Remote host closed the connection) 2017-03-01T22:03:35Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-01T22:05:03Z DGASAU quit (Read error: Connection reset by peer) 2017-03-01T22:05:51Z DGASAU joined #sbcl 2017-03-01T22:28:51Z attila_lendvai joined #sbcl 2017-03-01T22:50:34Z attila_lendvai quit (Quit: Leaving.) 2017-03-01T22:50:40Z attila_lendvai joined #sbcl 2017-03-01T22:50:40Z attila_lendvai quit (Changing host) 2017-03-01T22:50:40Z attila_lendvai joined #sbcl 2017-03-01T22:55:50Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-01T23:47:51Z cromachina joined #sbcl 2017-03-01T23:51:55Z jamtho joined #sbcl 2017-03-02T00:03:24Z jamtho_ joined #sbcl 2017-03-02T00:03:34Z jamtho quit (Read error: Connection reset by peer) 2017-03-02T00:54:29Z salva quit (Ping timeout: 260 seconds) 2017-03-02T01:00:37Z salva joined #sbcl 2017-03-02T01:00:41Z akkad quit (Excess Flood) 2017-03-02T01:09:18Z akkad joined #sbcl 2017-03-02T01:11:28Z akkad quit (Excess Flood) 2017-03-02T01:12:18Z akkad joined #sbcl 2017-03-02T01:38:08Z jamtho_ quit (Ping timeout: 240 seconds) 2017-03-02T01:49:08Z xristos quit (Quit: ZNC - http://znc.in) 2017-03-02T01:50:42Z edgar-rft quit (Quit: edgar-rft) 2017-03-02T02:04:30Z xristos joined #sbcl 2017-03-02T02:50:32Z xristos quit (Quit: ZNC - http://znc.in) 2017-03-02T02:51:00Z specbot quit (Write error: Broken pipe) 2017-03-02T02:51:00Z minion quit (Write error: Broken pipe) 2017-03-02T02:51:31Z minion joined #sbcl 2017-03-02T02:51:31Z specbot joined #sbcl 2017-03-02T03:01:12Z rumbler31 joined #sbcl 2017-03-02T03:02:57Z jack_rabbit joined #sbcl 2017-03-02T03:10:04Z rumbler31 quit (Remote host closed the connection) 2017-03-02T03:49:40Z stassats quit (Ping timeout: 246 seconds) 2017-03-02T04:24:50Z fjl__ quit (Ping timeout: 268 seconds) 2017-03-02T04:26:18Z fjl__ joined #sbcl 2017-03-02T04:38:48Z malm quit (Ping timeout: 260 seconds) 2017-03-02T04:40:56Z malm joined #sbcl 2017-03-02T04:45:19Z malm quit (Ping timeout: 246 seconds) 2017-03-02T04:45:29Z malm joined #sbcl 2017-03-02T04:57:27Z malm quit (Ping timeout: 240 seconds) 2017-03-02T04:57:37Z malm joined #sbcl 2017-03-02T05:04:30Z jamtho_ joined #sbcl 2017-03-02T05:09:36Z jamtho_ quit (Ping timeout: 260 seconds) 2017-03-02T05:29:30Z oleo quit (Quit: Leaving) 2017-03-02T05:36:45Z akkad quit (Excess Flood) 2017-03-02T05:41:22Z akkad joined #sbcl 2017-03-02T06:19:51Z shka_ joined #sbcl 2017-03-02T06:31:48Z shka_ quit (Ping timeout: 240 seconds) 2017-03-02T06:49:20Z reinuseslisp joined #sbcl 2017-03-02T07:06:07Z angavrilov joined #sbcl 2017-03-02T07:21:22Z gingerale joined #sbcl 2017-03-02T07:27:47Z igajsin joined #sbcl 2017-03-02T07:28:26Z reinuseslisp quit (Quit: Leaving) 2017-03-02T07:32:55Z jdz: Hey guys! I have a running SBCL which now has ended up in LDB due to memory exhaustion, and there's valuable data in hash tables. Is there a way to "export" that data? 2017-03-02T07:34:22Z flip214 speculates about doing a core dump, opening that one with GDB later on, and using the available C functions to get the data out .... ouch. 2017-03-02T07:34:59Z jdz: The good thing is that keys and values are strings and lists of strings. 2017-03-02T07:35:22Z jdz: I've never used LDB, which does not help at all. 2017-03-02T07:35:45Z jdz: I'll probably go read the manual. 2017-03-02T07:38:36Z flip214: would be nice if you could change the dynamic space size, to get a few hundred MB free space, to rescue the data 2017-03-02T07:38:57Z flip214: do you already have a non-consing function to write the data out? you should be able to call that. 2017-03-02T07:40:27Z jdz: Nope, no such function. 2017-03-02T07:40:56Z jdz: Another thing is that the heap exhaustion was caused by me running full GC. 2017-03-02T07:41:08Z gingerale quit (Remote host closed the connection) 2017-03-02T07:41:52Z jdz: So the process was taking a bit over 600MB, and now it is 1047MB, and in LDB. 2017-03-02T07:43:10Z jdz: If there is documentation for LDB I don't immediately see it... 2017-03-02T07:43:16Z flip214: hmmm, might be an interesting project to have a core-dump-loader in sbcl... 2017-03-02T07:43:40Z flip214: jdz: http://www.sbcl.org/manual/#Debugger-Entry 2017-03-02T07:44:07Z flip214: ie. some script that takes a core dump, and makes the data therein available via some functions. 2017-03-02T07:44:22Z jdz: flip214: that link is not for LDB, is it? 2017-03-02T07:45:34Z jdz: Another project would be to add an SBCL-dump-frobbing module for RaDaRe2. 2017-03-02T07:47:18Z flip214: hmmm, yeah, sorry... got the wrong link. 2017-03-02T07:49:01Z flip214: jdz: https://github.com/sbcl/sbcl/blob/master/src/runtime/monitor.c#L73 2017-03-02T07:49:56Z jdz: Thanks! 2017-03-02T07:51:23Z flip214: no problem 2017-03-02T07:52:02Z flip214: jdz: crazy idea: 2017-03-02T07:52:27Z flip214: use GDB to attach to the current sbcl, tell it to follow the child process, and do a fork(). 2017-03-02T07:52:35Z flip214: then you have an independent mapping of all the memory pages. 2017-03-02T07:52:43Z flip214: leave that hanging. 2017-03-02T07:52:56Z flip214: start a new SBCL with a bigger dynamic memory area 2017-03-02T07:53:26Z flip214: attach a GDB to that as well, and mmap() the dynamic space of the still-stopped child into the newly started sbcl. 2017-03-02T07:53:50Z flip214: with a bit of luck (and if it's only single-threaded?) you might be able to access the data then... 2017-03-02T07:54:02Z flip214: and if not, there's nothing lost, because it's a COW mapping anyway. 2017-03-02T07:54:27Z jdz: Alas, there are threads. 2017-03-02T07:54:50Z jdz: But an interesting exercise nevertheless. Let's see how far I get. 2017-03-02T07:58:53Z jdz: My last real usage of GDB being more than 15 years ago will not help, either. 2017-03-02T08:06:46Z leah2 quit (Ping timeout: 240 seconds) 2017-03-02T08:10:30Z jack_rabbit: That's nuts. 2017-03-02T08:10:40Z jack_rabbit: But really cool if you can actually get it to work. 2017-03-02T08:24:31Z flip214: My hope is that the size of the dynamic area is in global variables _outside_ the mapped area, so it might just work. 2017-03-02T08:31:06Z edgar-rft joined #sbcl 2017-03-02T08:43:15Z jdz: So I should execsomethingsomething the new SBCL within the forked process, right? 2017-03-02T08:47:05Z jdz: I don't see how to map an address space of one process into an address space of another process. 2017-03-02T09:00:50Z flip214: NOooo.. don't exec. 2017-03-02T09:02:16Z flip214: jdz: you need to "fake" (via the "call" gdb command) the open("/proc//mem", ...) and the mmap() call to the FIXED!! address. 2017-03-02T09:02:37Z flip214: yeah, it's not trivial, but I've done similar things in the past already ;) 2017-03-02T09:03:08Z jdz: Oh, right, /proc. 2017-03-02T09:04:07Z flip214: lots of cools things that can be done, with enough patience. 2017-03-02T09:04:23Z Shinmera: jdz: I'm just curious: why was your data not already persisted, if it's valuable? 2017-03-02T09:04:45Z Bike quit (Quit: leaving) 2017-03-02T09:04:52Z jack_rabbit: That's a good question. 2017-03-02T09:04:58Z Shinmera: flip214: Well, with enough patience you could sift through the memory dump manually and get the data out 2017-03-02T09:06:04Z jdz: Yes, it's not _that_ valuable. But I was just wondering if there's a way to salvage it. 2017-03-02T09:46:35Z attila_lendvai joined #sbcl 2017-03-02T09:46:35Z attila_lendvai quit (Changing host) 2017-03-02T09:46:35Z attila_lendvai joined #sbcl 2017-03-02T10:12:09Z sjl joined #sbcl 2017-03-02T10:13:26Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-02T11:03:08Z scymtym: have the ssh fingerprints of the sourceforge servers changed? i get a mismatch with a fingerprint that is not listed in https://sourceforge.net/p/forge/documentation/SSH%20Key%20Fingerprints/ 2017-03-02T11:25:44Z sjl quit (Read error: Connection reset by peer) 2017-03-02T11:41:50Z m00natic joined #sbcl 2017-03-02T12:11:32Z milanj joined #sbcl 2017-03-02T12:34:10Z xristos joined #sbcl 2017-03-02T12:46:25Z sjl joined #sbcl 2017-03-02T13:03:13Z jamtho joined #sbcl 2017-03-02T13:37:28Z jamtho quit (Ping timeout: 256 seconds) 2017-03-02T13:38:24Z cromachina quit (Read error: Connection reset by peer) 2017-03-02T14:55:44Z gingerale joined #sbcl 2017-03-02T15:04:36Z sjl quit (Ping timeout: 260 seconds) 2017-03-02T15:07:20Z leah2_ joined #sbcl 2017-03-02T15:08:10Z leah2_ is now known as leah2 2017-03-02T15:54:30Z CnTLDude joined #sbcl 2017-03-02T16:09:53Z oleo joined #sbcl 2017-03-02T16:31:20Z sjl joined #sbcl 2017-03-02T16:40:44Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-02T17:14:20Z igajsin quit (Ping timeout: 260 seconds) 2017-03-02T17:19:39Z Bike joined #sbcl 2017-03-02T17:27:34Z igajsin joined #sbcl 2017-03-02T17:53:13Z m00natic quit (Remote host closed the connection) 2017-03-02T18:41:34Z shka_ joined #sbcl 2017-03-02T20:30:05Z angavrilov quit (Remote host closed the connection) 2017-03-02T20:44:50Z stassats joined #sbcl 2017-03-02T21:12:19Z Xach_ joined #sbcl 2017-03-02T21:12:34Z Xach_: Hi friends. Is there a way to expand a pathname #p"~/foo/" without truenaming it at the same time? 2017-03-02T21:13:32Z Xach_ quit (Changing host) 2017-03-02T21:13:32Z Xach_ joined #sbcl 2017-03-02T21:13:35Z stassats: (pathname-directory #p"~/foo/") => (:ABSOLUTE :HOME "foo") 2017-03-02T21:13:45Z stassats: you can use that 2017-03-02T21:13:55Z Xach_: Hmm, ok. Thanks. 2017-03-02T21:17:46Z edgar-rft quit (Quit: edgar-rft) 2017-03-02T21:18:10Z scymtym: stassats: are you using ssh for the sourceforge repository? 2017-03-02T21:18:26Z stassats: scymtym: how do i tell? 2017-03-02T21:19:26Z scymtym: the remote url starts with ssh:// (as opposed to https://) 2017-03-02T21:19:34Z igajsin quit (Ping timeout: 264 seconds) 2017-03-02T21:19:38Z stassats: it's not https:// 2017-03-02T21:20:39Z scymtym: do/did you receive a warning about changed server ssh fingerprints when fetching or pushing? 2017-03-02T21:21:13Z stassats: no, nothing 2017-03-02T21:22:52Z scymtym: ok. sorry to bother you with more questions, but is the remote url of the form ssh://USERNAME@git.code.sf.net/p/sbcl/sbcl ? 2017-03-02T21:23:46Z stassats: yes, looks like that 2017-03-02T21:24:25Z scymtym: i see, but i don't understand what's wrong on my end, then 2017-03-02T21:31:32Z myrkraverk: scymtym: Maybe you /are/ being MITMed. 2017-03-02T21:32:50Z scymtym: myrkraverk: that is an (unlikely) possibility which i'm not taking lightly. however, i suspect a configuration error on my part 2017-03-02T21:40:45Z CnTLDude quit (Read error: Connection reset by peer) 2017-03-02T21:43:58Z scymtym: solved it by looking up the previous remote url in my backup. in case anyone else has this problem: ssh://USERNAME@git.code.sourceforge.net/p/sbcl/sbcl is the old url. it works with the official fingerprints (https://sourceforge.net/p/forge/documentation/SSH%20Key%20Fingerprints/#fingerprint-listing ). the documented urls (https://sourceforge.net/p/forge/documentation/Git/#developer-access-readwrite ) use fingerprints that i could no 2017-03-02T21:43:58Z scymtym: verify anywhere 2017-03-02T21:54:40Z |3b| quit (Read error: Connection reset by peer) 2017-03-02T22:00:15Z |3b| joined #sbcl 2017-03-02T22:00:58Z prxq joined #sbcl 2017-03-02T22:02:29Z shka_ quit (Ping timeout: 258 seconds) 2017-03-02T22:03:26Z Lord_Nightmare quit (Ping timeout: 240 seconds) 2017-03-02T22:07:49Z Xach_ left #sbcl 2017-03-02T22:08:54Z Lord_Nightmare joined #sbcl 2017-03-02T22:42:54Z rumbler3_ joined #sbcl 2017-03-02T22:43:55Z rumbler3_ quit (Remote host closed the connection) 2017-03-02T23:04:49Z milanj quit (Quit: Leaving) 2017-03-02T23:11:27Z attila_lendvai joined #sbcl 2017-03-03T00:03:37Z gingerale quit (Read error: Connection reset by peer) 2017-03-03T00:06:26Z cromachina joined #sbcl 2017-03-03T00:15:29Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-03T00:20:10Z prxq quit (Remote host closed the connection) 2017-03-03T00:27:56Z kini joined #sbcl 2017-03-03T00:43:18Z jamtho joined #sbcl 2017-03-03T00:44:26Z rumbler31 joined #sbcl 2017-03-03T00:48:58Z rumbler31 quit (Ping timeout: 256 seconds) 2017-03-03T01:28:29Z eschatologist quit (Quit: ZNC 1.6.3+deb2 - http://znc.in) 2017-03-03T01:30:09Z eschatologist joined #sbcl 2017-03-03T01:38:44Z jamtho quit (Quit: Leaving) 2017-03-03T02:16:29Z rumbler31 joined #sbcl 2017-03-03T02:29:25Z jack_rabbit joined #sbcl 2017-03-03T02:42:48Z nicdev` is now known as nicdev 2017-03-03T03:48:22Z oleo quit (Read error: Connection reset by peer) 2017-03-03T03:58:11Z neuronsong joined #sbcl 2017-03-03T03:58:17Z neuronsong quit (Client Quit) 2017-03-03T04:20:04Z stassats quit (Ping timeout: 258 seconds) 2017-03-03T04:21:36Z rumbler31 quit (Remote host closed the connection) 2017-03-03T04:37:40Z akkad quit (Excess Flood) 2017-03-03T04:41:33Z akkad joined #sbcl 2017-03-03T04:59:28Z oleo joined #sbcl 2017-03-03T05:44:34Z reinuseslisp joined #sbcl 2017-03-03T05:50:00Z shka_ joined #sbcl 2017-03-03T05:54:22Z akkad quit (Excess Flood) 2017-03-03T06:01:03Z akkad joined #sbcl 2017-03-03T06:45:58Z karswell quit (Ping timeout: 246 seconds) 2017-03-03T06:50:00Z igajsin joined #sbcl 2017-03-03T07:11:32Z shka_ quit (Ping timeout: 260 seconds) 2017-03-03T07:38:13Z reinuseslisp quit (Quit: Leaving) 2017-03-03T07:48:31Z angavrilov joined #sbcl 2017-03-03T08:26:40Z igajsin quit (Ping timeout: 260 seconds) 2017-03-03T08:34:23Z igajsin joined #sbcl 2017-03-03T08:41:56Z akkad quit (Excess Flood) 2017-03-03T08:45:08Z akkad joined #sbcl 2017-03-03T08:53:57Z akkad quit (Excess Flood) 2017-03-03T08:54:07Z akkad joined #sbcl 2017-03-03T09:13:57Z schjetne quit (Ping timeout: 240 seconds) 2017-03-03T09:36:34Z gingerale joined #sbcl 2017-03-03T10:04:30Z attila_lendvai joined #sbcl 2017-03-03T10:04:30Z attila_lendvai quit (Changing host) 2017-03-03T10:04:30Z attila_lendvai joined #sbcl 2017-03-03T10:20:02Z Bike quit (Quit: leaving) 2017-03-03T10:27:13Z schjetne joined #sbcl 2017-03-03T11:15:29Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-03T11:23:19Z rumbler31 joined #sbcl 2017-03-03T11:27:22Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-03T11:45:55Z DGASAU quit (Read error: Connection reset by peer) 2017-03-03T11:46:42Z DGASAU joined #sbcl 2017-03-03T13:37:26Z stassats joined #sbcl 2017-03-03T13:47:47Z dougk_ quit (Ping timeout: 255 seconds) 2017-03-03T13:49:58Z cromachina quit (Read error: Connection reset by peer) 2017-03-03T14:00:37Z dougk_ joined #sbcl 2017-03-03T14:14:46Z schjetne quit (Ping timeout: 264 seconds) 2017-03-03T14:16:00Z edgar-rft joined #sbcl 2017-03-03T14:32:47Z DGASAU quit (Read error: Connection reset by peer) 2017-03-03T14:34:44Z DGASAU joined #sbcl 2017-03-03T14:35:35Z sjl quit (Ping timeout: 240 seconds) 2017-03-03T15:20:52Z schjetne joined #sbcl 2017-03-03T15:23:54Z rumbler31 joined #sbcl 2017-03-03T15:28:10Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-03T15:39:51Z schjetne quit (Read error: Connection reset by peer) 2017-03-03T16:36:40Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-03T16:38:28Z schjetne joined #sbcl 2017-03-03T16:59:24Z Bike joined #sbcl 2017-03-03T17:37:44Z sebboh``` joined #sbcl 2017-03-03T17:37:48Z paroneay` joined #sbcl 2017-03-03T17:37:54Z nicdev` joined #sbcl 2017-03-03T17:40:40Z liead joined #sbcl 2017-03-03T17:42:26Z paroneay` is now known as paroneayea` 2017-03-03T17:42:39Z abruanese_ joined #sbcl 2017-03-03T17:42:44Z foom2 joined #sbcl 2017-03-03T17:43:11Z foom quit (*.net *.split) 2017-03-03T17:43:11Z nicdev quit (*.net *.split) 2017-03-03T17:43:12Z les` quit (*.net *.split) 2017-03-03T17:43:13Z adlai quit (*.net *.split) 2017-03-03T17:43:13Z gko quit (*.net *.split) 2017-03-03T17:43:13Z paroneayea quit (*.net *.split) 2017-03-03T17:43:14Z sebboh`` quit (*.net *.split) 2017-03-03T17:43:14Z abruanese quit (*.net *.split) 2017-03-03T17:43:40Z les joined #sbcl 2017-03-03T17:43:42Z gko joined #sbcl 2017-03-03T17:44:02Z abruanese_ is now known as abruanese 2017-03-03T17:45:02Z SolairePhantom joined #sbcl 2017-03-03T17:49:18Z paroneayea` is now known as paroneayea 2017-03-03T17:49:30Z paroneayea quit (Changing host) 2017-03-03T17:49:30Z paroneayea joined #sbcl 2017-03-03T18:26:57Z sjl joined #sbcl 2017-03-03T18:49:07Z attila_lendvai joined #sbcl 2017-03-03T18:49:07Z attila_lendvai quit (Changing host) 2017-03-03T18:49:07Z attila_lendvai joined #sbcl 2017-03-03T19:06:22Z dra joined #sbcl 2017-03-03T19:21:10Z wsenn joined #sbcl 2017-03-03T19:22:22Z foom2 is now known as foom 2017-03-03T19:26:47Z wsenn: Should I build sbcl from source with --fancy on linux mint 18.1 serena (ubuntu 16.04)? I am getting a failed concurrency check (to complicate things, it's in a vm). 2017-03-03T19:27:42Z wsenn: is there a recommended build command that's skip threads but get everything else --fancy offers? 2017-03-03T19:30:12Z nyef`: You might try --fancy --without-sb-threads, but I don't know if that'll work or not. 2017-03-03T19:30:35Z wsenn: or, alternatively, is make without any arguments good enuf? 2017-03-03T19:30:39Z nyef`: (I don't remember if --without trumps --with or if it's the other way around.) 2017-03-03T19:31:05Z wsenn: i'll give it a go. thx. 2017-03-03T19:32:11Z nyef`: Looks like --fancy is equivalent to --with-sb-core-compression --with-sb-xref-for-internals --with-sb-after-xc-core, and on platforms that support it --with-sb-thread 2017-03-03T19:32:35Z stassats: building without threads doesn't solve thread issues 2017-03-03T19:33:30Z nyef`: ... But does follow the maxim of "never check for an error case that you don't know how to handle." d-: 2017-03-03T19:34:59Z stassats: ignoring the sb-concurrency error would have achieved the same result, but quicker 2017-03-03T19:37:40Z zellerin joined #sbcl 2017-03-03T19:41:55Z oleo quit (Quit: Leaving) 2017-03-03T19:45:47Z dra quit (Ping timeout: 260 seconds) 2017-03-03T19:48:56Z angavrilov quit (Remote host closed the connection) 2017-03-03T19:58:26Z dra_ joined #sbcl 2017-03-03T20:03:09Z oleo joined #sbcl 2017-03-03T20:10:17Z dra_ quit (Ping timeout: 260 seconds) 2017-03-03T20:26:21Z dra_ joined #sbcl 2017-03-03T20:29:14Z fiddlerwoaroof: What's the maximum dynamic space size? 2017-03-03T20:29:58Z fiddlerwoaroof: I consistently get errors trying to use a 40G size 2017-03-03T20:30:17Z fiddlerwoaroof: for the default 2017-03-03T20:30:55Z dra_ is now known as dra 2017-03-03T20:31:05Z fiddlerwoaroof: on FreeBSD, if it makes a difference 2017-03-03T20:31:10Z kslt1 joined #sbcl 2017-03-03T20:31:57Z fiddlerwoaroof: Hmm, I'm specifying it wrong, never mind 2017-03-03T20:44:43Z dra quit (Quit: Leaving) 2017-03-03T21:14:05Z zellerin quit (Ping timeout: 240 seconds) 2017-03-03T21:19:22Z wsenn left #sbcl 2017-03-03T21:19:56Z igajsin quit (Ping timeout: 260 seconds) 2017-03-03T21:34:56Z phoe is now known as phof 2017-03-03T21:35:30Z phof is now known as phoe 2017-03-03T22:01:30Z SolairePhantom quit (Quit: Leaving) 2017-03-03T22:01:35Z rumbler3_ joined #sbcl 2017-03-03T22:02:38Z rumbler3_ quit (Remote host closed the connection) 2017-03-03T22:21:27Z kslt1 quit (Ping timeout: 258 seconds) 2017-03-03T22:32:29Z gingerale quit (Remote host closed the connection) 2017-03-03T23:29:45Z akkad quit (Excess Flood) 2017-03-03T23:37:14Z akkad joined #sbcl 2017-03-04T00:03:09Z rumbler31 joined #sbcl 2017-03-04T00:07:13Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-04T00:14:43Z attila_lendvai quit (Read error: No route to host) 2017-03-04T00:14:47Z attila_lendvai1 joined #sbcl 2017-03-04T00:14:47Z attila_lendvai1 is now known as attila_lendvai 2017-03-04T00:14:47Z attila_lendvai quit (Changing host) 2017-03-04T00:14:47Z attila_lendvai joined #sbcl 2017-03-04T00:20:31Z akkad quit (Excess Flood) 2017-03-04T00:25:14Z akkad joined #sbcl 2017-03-04T00:26:56Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-04T01:27:08Z cromachina joined #sbcl 2017-03-04T02:42:48Z stassats quit (Ping timeout: 240 seconds) 2017-03-04T03:06:30Z rumbler31 joined #sbcl 2017-03-04T03:23:13Z sjl quit (Ping timeout: 246 seconds) 2017-03-04T03:39:47Z rumbler31 quit (Remote host closed the connection) 2017-03-04T04:41:52Z akkad quit (Excess Flood) 2017-03-04T04:51:14Z akkad joined #sbcl 2017-03-04T05:14:23Z akkad quit (Excess Flood) 2017-03-04T05:18:44Z akkad joined #sbcl 2017-03-04T05:40:19Z rumbler31 joined #sbcl 2017-03-04T05:46:34Z rumbler31 quit (Ping timeout: 264 seconds) 2017-03-04T06:03:05Z akkad quit (Excess Flood) 2017-03-04T06:11:44Z akkad joined #sbcl 2017-03-04T06:34:09Z jack_rabbit joined #sbcl 2017-03-04T06:42:24Z rumbler31 joined #sbcl 2017-03-04T06:47:37Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-04T07:09:52Z gingerale joined #sbcl 2017-03-04T07:50:17Z DGASAU quit (Read error: Connection reset by peer) 2017-03-04T07:51:51Z DGASAU joined #sbcl 2017-03-04T07:59:26Z angavrilov joined #sbcl 2017-03-04T08:11:16Z igajsin joined #sbcl 2017-03-04T08:36:31Z DGASAU quit (Read error: Connection reset by peer) 2017-03-04T09:35:31Z attila_lendvai joined #sbcl 2017-03-04T09:35:31Z attila_lendvai quit (Changing host) 2017-03-04T09:35:31Z attila_lendvai joined #sbcl 2017-03-04T09:44:55Z Bike quit (Quit: leaving) 2017-03-04T10:45:53Z rumbler31 joined #sbcl 2017-03-04T10:50:10Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-04T11:00:52Z liead is now known as adlai 2017-03-04T11:02:44Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-04T11:04:43Z attila_lendvai joined #sbcl 2017-03-04T11:18:41Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-04T11:36:19Z sjl joined #sbcl 2017-03-04T12:09:52Z scymtym quit (Remote host closed the connection) 2017-03-04T12:11:43Z stassats joined #sbcl 2017-03-04T12:25:20Z stassats: huh, why can't i make a special form with a &key, ds-bind complains in a roundabout way 2017-03-04T12:25:46Z stassats: (NAMED-DS-BIND (:SPECIAL-FORM . THE*) ((VALUE-TYPE &KEY CONTEXT SILENT-CONFLICT) FORM) (CDR #:EXPR)) => The value THE* is not of type LIST 2017-03-04T12:28:31Z stassats: ok, the difference is that macros use (:MACRO FOO . DEFMACRO), so the CDDR doesn't work 2017-03-04T12:46:45Z rumbler31 joined #sbcl 2017-03-04T12:47:36Z attila_lendvai joined #sbcl 2017-03-04T12:49:42Z scymtym joined #sbcl 2017-03-04T12:50:55Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-04T16:48:09Z rumbler31 joined #sbcl 2017-03-04T16:52:25Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-04T17:23:08Z zellerin joined #sbcl 2017-03-04T17:23:49Z DGASAU joined #sbcl 2017-03-04T17:27:44Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-04T17:49:11Z Bike joined #sbcl 2017-03-04T17:49:51Z scymtym quit (Ping timeout: 258 seconds) 2017-03-04T17:57:04Z zellerin quit (Ping timeout: 260 seconds) 2017-03-04T18:03:28Z scymtym joined #sbcl 2017-03-04T18:20:11Z zellerin joined #sbcl 2017-03-04T18:48:55Z rumbler31 joined #sbcl 2017-03-04T18:52:00Z stassats: and an ir1-translator with &key crashes in xc 2017-03-04T18:52:41Z stassats: i could have already solved the issue i was after, but no, the infrastructure is borked 2017-03-04T18:52:42Z stassats: sigh 2017-03-04T18:53:10Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-04T19:02:10Z zellerin quit (Ping timeout: 264 seconds) 2017-03-04T19:45:59Z zellerin joined #sbcl 2017-03-04T19:49:39Z rumbler31 joined #sbcl 2017-03-04T19:54:04Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-04T20:10:45Z stassats: a value protected by WHEN is actually nil 2017-03-04T20:11:17Z stassats: or rather, it's (values), but WHEN doesn't protect the nil 2017-03-04T20:23:38Z myrkraverk quit (Remote host closed the connection) 2017-03-04T20:36:05Z stassats: it's a miscompilation indeed, sigh 2017-03-04T20:36:49Z stassats: http://paste.lisp.org/display/340560 2017-03-04T20:39:28Z stassats: having a variable is necessary 2017-03-04T20:39:35Z stassats: probably the type gets derived wrong 2017-03-04T20:39:56Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-04T20:42:29Z stassats: or(defun bar (x) (let ((x (the (values &optional integer) (foo x)))) (when x 10))) 2017-03-04T20:42:39Z stassats: (bar nil) => 10 2017-03-04T20:43:56Z stassats: (let ((x (the (values &optional integer) (eval '(values))))) (when x 10)) => 10 2017-03-04T20:45:32Z zellerin quit (Ping timeout: 260 seconds) 2017-03-04T20:50:25Z rumbler31 joined #sbcl 2017-03-04T20:52:18Z hydan joined #sbcl 2017-03-04T20:54:37Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-04T20:58:51Z stassats: (let (sb-c::*constraint-propagate*) (eval '(let ((x (the (values &optional integer) (eval '(values))))) (when x 10)))) => NIL 2017-03-04T20:59:01Z stassats: that's not what i wanted to deal with, contraintprop 2017-03-04T21:02:39Z myrkraverk joined #sbcl 2017-03-04T21:05:52Z rumbler31 joined #sbcl 2017-03-04T21:06:52Z stassats` joined #sbcl 2017-03-04T21:13:19Z hydan quit (*.net *.split) 2017-03-04T21:13:20Z christoph_debian quit (*.net *.split) 2017-03-04T21:13:21Z kini quit (*.net *.split) 2017-03-04T21:14:23Z stassats quit (*.net *.split) 2017-03-04T21:14:23Z angavrilov quit (*.net *.split) 2017-03-04T21:17:45Z christoph_debian joined #sbcl 2017-03-04T21:18:38Z angavrilov joined #sbcl 2017-03-04T21:20:46Z stassats`: it's ultimately not in constraintprop 2017-03-04T21:21:22Z angavrilov quit (Remote host closed the connection) 2017-03-04T21:37:19Z stassats`: the whole checkgen is really silly about optional value types 2017-03-04T21:39:10Z stassats`: it thinks (values (the (values &optional integer) (eval '(values nil)))) is ok 2017-03-04T21:44:15Z hydan joined #sbcl 2017-03-04T21:44:25Z Bike: i'm glad it's not just me having dumb problems with values types 2017-03-04T21:51:54Z stassats`: just had to replace VALUES-TYPE-OUT with VALUES-TYPE-IN 2017-03-04T21:52:21Z stassats`: but that's for the miscompilation problem, not (values (the (values &optional integer) (eval '(values nil)))) 2017-03-04T21:52:52Z stassats`: the latter is more difficult to solve, as there's no way to know how many values m-v-bind actually received 2017-03-04T21:55:48Z Bike: the uses m-v-b? 2017-03-04T21:56:12Z stassats`: you mean THE? yes 2017-03-04T21:57:53Z stassats`: wanted to fix one bug, ending up fixing four bugs for the price of several more hours 2017-03-04T22:02:04Z igajsin quit (Ping timeout: 260 seconds) 2017-03-04T22:17:23Z stassats`: oh crap, now the thing that prompted this VALUES bug will break building with older SBCLs 2017-03-04T22:17:38Z stassats`: i guess i'll manually NIL it 2017-03-04T23:05:17Z gingerale quit (Read error: Connection reset by peer) 2017-03-05T01:16:51Z logrus joined #sbcl 2017-03-05T01:25:35Z hydan quit (Ping timeout: 240 seconds) 2017-03-05T01:29:12Z slyrus joined #sbcl 2017-03-05T01:42:33Z sjl quit (Read error: Connection reset by peer) 2017-03-05T01:46:25Z jamtho joined #sbcl 2017-03-05T01:47:30Z jamtho_ joined #sbcl 2017-03-05T01:49:06Z jamtho__ joined #sbcl 2017-03-05T01:50:48Z jamtho quit (Ping timeout: 240 seconds) 2017-03-05T01:52:08Z jamtho_ quit (Ping timeout: 240 seconds) 2017-03-05T01:59:10Z jamtho__ quit (Ping timeout: 264 seconds) 2017-03-05T02:16:15Z jack_rabbit joined #sbcl 2017-03-05T02:25:03Z akkad quit (Excess Flood) 2017-03-05T02:28:14Z akkad joined #sbcl 2017-03-05T02:38:45Z pjstirling joined #sbcl 2017-03-05T02:39:08Z pjstirling: hi, is there a way to control how alien values print? 2017-03-05T02:39:30Z stassats`: what for? 2017-03-05T02:39:59Z pjstirling: I'm debugging and it would make my life easier :) 2017-03-05T02:40:11Z stassats`: print-object doesn't work? 2017-03-05T02:40:33Z pjstirling: would it? 2017-03-05T02:40:39Z stassats`: did you try it? 2017-03-05T02:42:31Z pjstirling: not yet, I thought print-object was just for classes, not things like (alien (* thing)) 2017-03-05T02:43:18Z stassats`: every value in common lisp has a class 2017-03-05T02:43:37Z stassats`: (including classes) 2017-03-05T02:49:34Z stassats` quit (Ping timeout: 264 seconds) 2017-03-05T03:14:54Z slyrus quit (Ping timeout: 268 seconds) 2017-03-05T03:18:07Z slyrus joined #sbcl 2017-03-05T04:00:58Z slyrus quit (Ping timeout: 264 seconds) 2017-03-05T04:52:25Z kini joined #sbcl 2017-03-05T05:56:24Z logrus quit (Quit: Leaving) 2017-03-05T06:20:30Z reinuseslisp joined #sbcl 2017-03-05T06:29:52Z vtomole joined #sbcl 2017-03-05T06:58:30Z slyrus joined #sbcl 2017-03-05T07:10:40Z beatdown joined #sbcl 2017-03-05T07:13:50Z igajsin joined #sbcl 2017-03-05T07:36:20Z vtomole quit (Ping timeout: 260 seconds) 2017-03-05T07:36:29Z reinuseslisp quit (Quit: Leaving) 2017-03-05T08:04:34Z slyrus quit (Ping timeout: 264 seconds) 2017-03-05T08:47:13Z igajsin quit (Ping timeout: 246 seconds) 2017-03-05T08:49:00Z angavrilov joined #sbcl 2017-03-05T09:02:18Z igajsin joined #sbcl 2017-03-05T09:35:36Z gk_1wm_su joined #sbcl 2017-03-05T09:35:38Z gk_1wm_su left #sbcl 2017-03-05T09:37:30Z gk_1wm_su joined #sbcl 2017-03-05T09:39:36Z Bike quit (Quit: leaving) 2017-03-05T09:41:13Z gk_1wm_su quit (K-Lined) 2017-03-05T10:00:22Z shka_ joined #sbcl 2017-03-05T10:08:38Z igajsin quit (Ping timeout: 256 seconds) 2017-03-05T10:12:06Z akkad quit (Excess Flood) 2017-03-05T10:17:16Z akkad joined #sbcl 2017-03-05T10:22:57Z gingerale joined #sbcl 2017-03-05T10:24:16Z sjl joined #sbcl 2017-03-05T10:27:20Z igajsin joined #sbcl 2017-03-05T10:40:11Z oleo quit (Quit: Leaving) 2017-03-05T11:23:34Z sjl quit (Read error: Connection reset by peer) 2017-03-05T12:06:42Z sjl joined #sbcl 2017-03-05T13:17:26Z oleo joined #sbcl 2017-03-05T13:17:27Z oleo quit (Changing host) 2017-03-05T13:17:27Z oleo joined #sbcl 2017-03-05T13:34:51Z sjl quit (Read error: Connection reset by peer) 2017-03-05T14:07:39Z sjl joined #sbcl 2017-03-05T14:41:01Z oleo quit (Quit: Verlassend) 2017-03-05T14:41:29Z attila_lendvai joined #sbcl 2017-03-05T15:43:55Z slyrus joined #sbcl 2017-03-05T15:59:49Z sjl quit (Ping timeout: 246 seconds) 2017-03-05T16:53:08Z slyrus quit (Ping timeout: 240 seconds) 2017-03-05T17:05:50Z oleo joined #sbcl 2017-03-05T17:13:26Z stassats joined #sbcl 2017-03-05T17:17:39Z Bike joined #sbcl 2017-03-05T19:02:41Z sjl joined #sbcl 2017-03-05T19:54:48Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-05T20:49:09Z sjl quit (Read error: Connection reset by peer) 2017-03-05T20:57:12Z rumbler31 quit (Remote host closed the connection) 2017-03-05T20:57:37Z rumbler31 joined #sbcl 2017-03-05T20:59:51Z rumbler31 quit (Client Quit) 2017-03-05T21:10:01Z angavrilov quit (Remote host closed the connection) 2017-03-05T21:44:40Z igajsin quit (Ping timeout: 260 seconds) 2017-03-05T21:48:29Z pjstirling quit (Quit: Ex-Chat) 2017-03-05T22:30:57Z reinuseslisp joined #sbcl 2017-03-05T22:37:09Z shka_ quit (Ping timeout: 258 seconds) 2017-03-05T22:41:08Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-05T22:50:32Z gingerale quit (Remote host closed the connection) 2017-03-05T23:40:39Z reinuseslisp quit (Quit: Leaving) 2017-03-06T00:56:22Z stassats: so, (write-sequence *array* stream) is 4000 times slower than (sb-posix:write (sb-sys:fd-stream-fd stream) (sb-sys:vector-sap *array*) (length *array*)) 2017-03-06T00:56:45Z stassats: on an unbuffered stream 2017-03-06T00:57:52Z stassats: fully buffered, it's only 46 times slower 2017-03-06T01:18:43Z nyef`: Fun and games, huh? But there's a critical bit, which is that such games only work right if the sequence element type is compatible with the stream element type... And even then, something-something zero-copy something. 2017-03-06T01:19:02Z stassats: actuaully, we already have fast output, it's just not being triggered 2017-03-06T01:19:02Z nyef`: Still, there's a good possible win there. 2017-03-06T01:19:39Z stassats: and this is actually a bivalent stream 2017-03-06T01:21:40Z nyef`: ... There's an "ambivalent stream" joke around here somewhere, I know it! 2017-03-06T01:21:46Z stassats: and i have a trivial fix 2017-03-06T01:23:27Z stassats: that doesn't solve the fact that utf8 routines are quite slow 2017-03-06T01:24:01Z stassats: even file-string-length is inexcusably slow, because it calls a function for each character 2017-03-06T01:24:09Z nyef`: Mmm. And that goes back to the "external format handling needs a general rework", doesn't it? 2017-03-06T01:24:22Z stassats: yes, yes 2017-03-06T01:24:23Z stassats: badly 2017-03-06T01:24:46Z nyef`: No, they don't need to be reworked badly, they badly need to be reworked well. d-: 2017-03-06T01:24:57Z stassats: (and file-string-length is bad API, it doesn't have :start/:end) 2017-03-06T01:25:24Z edgar-rft quit (Quit: edgar-rft) 2017-03-06T01:26:08Z stassats: and our string-to-octets is bad API, you can't encode into an existing buffer 2017-03-06T01:26:25Z stassats: that's why i resorted to bivalent streams, but they turned out to be slow sometimes 2017-03-06T01:29:34Z stassats: ok, i wonder if read-sequence is suffering too and if the fix is still good 2017-03-06T01:31:48Z stassats: it was, and the fix got both, gotta ammend the commit message 2017-03-06T01:32:23Z slyrus joined #sbcl 2017-03-06T01:37:58Z stassats: write to /dev/null is awfully smart and doesn't do anything 2017-03-06T01:39:19Z stassats: sb-posix:write is also awfully dumb as it accepts INT length 2017-03-06T01:41:30Z stassats: and we grovel for ssize-t but don't export it 2017-03-06T01:42:51Z stassats: oh, that's just sb-bsd-sockets, it also clobbers size-t 2017-03-06T01:43:29Z rumbler31 joined #sbcl 2017-03-06T01:46:01Z stassats: curiously enough, write(2) only writes 2147479552 at at ime 2017-03-06T01:49:42Z stassats: and on macos writing 2^31 results in EINVAL 2017-03-06T01:51:57Z stassats: which is strange 2017-03-06T01:54:47Z stassats: same thing from C, ok 2017-03-06T01:57:47Z stassats: that's something i've come to expect from darwin 2017-03-06T02:06:53Z stassats: ok, write-sequence on a 16MB array is now about 2.5 slower than write(2) 2017-03-06T02:10:09Z stassats: i guess i can just use write(2) directly, which i already do for my variable-width binary streams, but it'd be nicer for write-sequence to be fast for everyone 2017-03-06T02:17:32Z stassats: ok, don't want to deal with making buffer-output faster, there's serve events involved 2017-03-06T02:18:06Z stassats: or, i can just make it faster when there's no serve events 2017-03-06T02:19:57Z stassats: but there are also timeouts, ugh, won't then 2017-03-06T02:28:50Z stassats: turns out, write-sequence on an unbuffered stream is actually buffered, so it's no longer about performance only 2017-03-06T02:37:26Z nicdev` is now known as nicdev 2017-03-06T02:53:37Z stassats quit (Ping timeout: 246 seconds) 2017-03-06T02:54:58Z slyrus quit (Ping timeout: 264 seconds) 2017-03-06T03:40:22Z rumbler31 quit (Remote host closed the connection) 2017-03-06T04:40:55Z rumbler31 joined #sbcl 2017-03-06T04:45:37Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-06T04:56:13Z shka_ joined #sbcl 2017-03-06T05:48:05Z shka_ quit (Ping timeout: 240 seconds) 2017-03-06T05:49:32Z igajsin joined #sbcl 2017-03-06T05:56:31Z slyrus joined #sbcl 2017-03-06T06:12:58Z akkad quit (Excess Flood) 2017-03-06T06:14:16Z slyrus quit (Ping timeout: 268 seconds) 2017-03-06T06:19:03Z akkad joined #sbcl 2017-03-06T06:22:54Z schjetne quit (Ping timeout: 268 seconds) 2017-03-06T06:32:10Z whiteline quit (Ping timeout: 264 seconds) 2017-03-06T07:17:41Z whiteline joined #sbcl 2017-03-06T08:27:24Z schjetne joined #sbcl 2017-03-06T09:04:22Z angavrilov joined #sbcl 2017-03-06T09:07:33Z eschatologist quit (Ping timeout: 268 seconds) 2017-03-06T09:11:34Z schjetne quit (Ping timeout: 258 seconds) 2017-03-06T09:13:08Z Bike quit (Quit: leaving) 2017-03-06T09:30:06Z eschatologist joined #sbcl 2017-03-06T09:32:38Z igajsin quit (Remote host closed the connection) 2017-03-06T09:35:22Z igajsin joined #sbcl 2017-03-06T10:04:12Z gingerale joined #sbcl 2017-03-06T10:12:08Z attila_lendvai joined #sbcl 2017-03-06T10:12:08Z attila_lendvai quit (Changing host) 2017-03-06T10:12:08Z attila_lendvai joined #sbcl 2017-03-06T10:13:06Z schjetne joined #sbcl 2017-03-06T10:19:57Z schjetne quit (Remote host closed the connection) 2017-03-06T10:50:06Z sjl joined #sbcl 2017-03-06T10:56:28Z stassats joined #sbcl 2017-03-06T11:08:05Z m00natic joined #sbcl 2017-03-06T11:33:34Z DGASAU quit (Read error: Connection reset by peer) 2017-03-06T11:36:35Z DGASAU joined #sbcl 2017-03-06T11:39:06Z DGASAU quit (Read error: Connection reset by peer) 2017-03-06T11:44:20Z DGASAU joined #sbcl 2017-03-06T11:50:18Z milanj joined #sbcl 2017-03-06T11:51:04Z DGASAU quit (Read error: Connection reset by peer) 2017-03-06T11:53:23Z DGASAU joined #sbcl 2017-03-06T12:33:28Z sjl quit (Ping timeout: 260 seconds) 2017-03-06T12:58:00Z rumbler31 joined #sbcl 2017-03-06T13:04:27Z igajsin quit (Read error: Connection reset by peer) 2017-03-06T13:08:17Z sjl joined #sbcl 2017-03-06T13:09:57Z igajsin joined #sbcl 2017-03-06T13:10:31Z DGASAU quit (Read error: Connection reset by peer) 2017-03-06T13:31:43Z rumbler31 quit (Remote host closed the connection) 2017-03-06T13:46:19Z edgar-rft joined #sbcl 2017-03-06T14:30:54Z cromachina quit (Read error: Connection reset by peer) 2017-03-06T14:45:29Z sjl quit (Ping timeout: 260 seconds) 2017-03-06T15:06:23Z |3b|: hmm, (SB-SYS:SAP-REF-SAP (CFFI:NULL-POINTER) 0) -> error "bad thing to be a type specifier: "EXCEPTION_ACCESS_VIOLATION"" (64bit windows if that wasn't obvious from the error) 2017-03-06T15:06:24Z schjetne joined #sbcl 2017-03-06T15:07:00Z stassats: is cffi really necessary? 2017-03-06T15:07:25Z |3b|: no, but i was too lazy to track down the sbcl equivalent 2017-03-06T15:07:33Z |3b|: (SB-SYS:SAP-REF-SAP (sb-sys:int-sap 0) 0) 2017-03-06T15:07:34Z stassats: (sb-sys:int-sap 0) 2017-03-06T15:10:04Z |3b|: linux gets a normal segfault, so assuming that is expected behavior problem is just the error reporting 2017-03-06T15:10:28Z schjetne quit (Ping timeout: 240 seconds) 2017-03-06T15:12:27Z |3b|: though if it can tell it came from user sap-ref and similar might be nice to get rid of the "CORRUPTION WARNING..." stuff on linux 2017-03-06T15:12:37Z schjetne joined #sbcl 2017-03-06T15:13:19Z stassats: linux doesn't have exceptions 2017-03-06T15:13:47Z stassats: |3b|: sap-ref can corrupt things just as well 2017-03-06T15:14:08Z |3b|: yeah, valid argument :) 2017-03-06T15:14:38Z |3b|: though seems like the times it corrupts something would be the ones that don't segfault :) 2017-03-06T15:15:44Z stassats: entering a:b on windows doesn't end well 2017-03-06T15:16:26Z stassats: that's in the mingw terminal, cmd.exe is ok 2017-03-06T15:16:52Z stassats: or powershell 2017-03-06T15:19:09Z stassats: the backtrace from SAP-REF-SAP is bonkers 2017-03-06T15:21:20Z |3b|: hmm, seems to have problems printing backtrace from file-not-found error in repl too 2017-03-06T15:21:51Z sjl joined #sbcl 2017-03-06T15:21:55Z |3b|: error printing initial error stuff, infinite loop after ba 2017-03-06T15:22:12Z |3b|: mingw terminal as well 2017-03-06T15:22:58Z stassats: i need to automate setting up the windows environment, i'm using a 90-days vm provided by MS 2017-03-06T15:23:36Z |3b|: doesn't seem to be specific to mingw terminal though, same in cmd 2017-03-06T15:24:10Z stassats: the mingw terminal is just broken 2017-03-06T15:25:23Z |3b|: but with correct filename, a:b READs fine from a file or from a string 2017-03-06T15:28:15Z stassats: argh, and emacs is silly out of the box on windows 2017-03-06T15:30:34Z stassats: mostly due to having three different ~ 2017-03-06T15:30:38Z |3b|: a:b might not be mingw specific, gets stuck in *inferior-lisp* too 2017-03-06T15:31:09Z |3b| starts emacs from ming term and gets msys2's ~ 2017-03-06T15:32:48Z stassats: i had it all set up, but now i'm using a time restricted vm so i don't want to redo it all every 90 days 2017-03-06T15:32:54Z stassats: also don't want to pay for windows or pirate it 2017-03-06T15:35:56Z |3b|: reader-find-package seems to get stuck in ERROR, apparently something to do with the stream argument 2017-03-06T15:36:56Z stassats: streams and terminals were broken, i may have made it more broken for the mingw terminal 2017-03-06T15:37:04Z |3b|: and looks like cmd uses a different stream (*terminal-io* vs sb-sys:*stdin*) 2017-03-06T15:39:24Z |3b|: possibly due to (sb-kernel::stream-error-position-info sb-sys:*stdin* nil) 2017-03-06T15:40:13Z |3b|: (in sb-kernel::%report-reader-error, called by simple-reader-error's report with no :position argument) 2017-03-06T15:43:15Z stassats: ok, got a nominal fix for the segfault 2017-03-06T15:43:26Z stassats: should do better, it doesn't report the address 2017-03-06T15:47:54Z |3b|: looks like problem with a:b is that it doesn't think sb-sys:*stdin* is interactive 2017-03-06T15:49:37Z sjl quit (Ping timeout: 268 seconds) 2017-03-06T15:50:32Z sjl joined #sbcl 2017-03-06T15:51:04Z |3b|: sb-sys:*stdin* seems to be a pipe according to getfiletype 2017-03-06T15:51:34Z |3b|: which should probably be treated same as interactive by stream-error-position-info 2017-03-06T15:54:41Z |3b| isn't sure whether windows-isatty, unix-isatty, fd-stream-misc-routine, or stream-error-position-info should be responsible for adding that check though 2017-03-06T15:54:52Z |3b|: (well, probably not unix-isatty) 2017-03-06T15:57:23Z stassats: there's some tricky way to detect if it's attached to a console 2017-03-06T15:58:32Z stassats: C-z doesn't work in mingw 2017-03-06T15:58:44Z stassats: just said "stopped" and that's it 2017-03-06T15:59:51Z |3b|: yeah, and C-d isn't much help either (i frequently hit C-d, remember it doesn't work then hit C-c, which kills sbcl and sends the C-d to the calling shell closing my terminal too) 2017-03-06T16:00:13Z |3b|: have to remember to give it some other input after C-d before hitting C-c 2017-03-06T16:01:12Z stassats: pushed the segfault fix, but without proper testing yet 2017-03-06T16:01:55Z stassats: (it's easier for me to push and then to build, because my windows setup is totally broken) 2017-03-06T16:02:18Z stassats: and it's better now, it actually mentions the address 2017-03-06T16:02:44Z |3b| starts a build 2017-03-06T16:04:40Z stassats: |3b|: i recently struggled a lot and fixed C-c, but only in cmd.exe or powershell 2017-03-06T16:05:29Z |3b|: looks like sbcl knows how to check for console handle, and it seems to match *stdin* 2017-03-06T16:05:32Z stassats: failed contribs, sb-posix and sb-simple-streams, weird 2017-03-06T16:06:33Z stassats: oh, is that due to ssize-t 2017-03-06T16:08:02Z stassats: how hard would it be to get rid of mingw completely? 2017-03-06T16:08:21Z stassats: probably not having x86-64-assem.S first, we already have tramps from lisp 2017-03-06T16:08:34Z |3b|: maybe add something like (member (sb-impl::fd-stream-fd sb-sys:*stdin*) (multiple-value-list (sb-win32::get-std-handles))) to windows-isatty ? 2017-03-06T16:08:44Z stassats: |3b|: you may want to restart the build 2017-03-06T16:08:50Z stassats: due to ssize-t 2017-03-06T16:09:14Z |3b| may want to restart it due to sb-concurrency tests getting stuck :p 2017-03-06T16:09:39Z stassats: that still a problem? i thought i dealt with it 2017-03-06T16:10:23Z |3b| hasn't seen it before that i can remember, could have just been the usual GC lock and back luck 2017-03-06T16:10:33Z |3b|: SB-CONCURRENCY-TEST::MAILBOX-TRIVIA.2 SB-CONCURRENCY-TEST::MAILBOX-TRIVIA.3 2017-03-06T16:10:41Z |3b|: was last thing it printed 2017-03-06T16:10:50Z |3b|: if that provides any useful info 2017-03-06T16:11:03Z stassats: the problem i dealt with is that sleep on windows is not very fine grained, and sb-concurrency-test used a lot of (sleep 0.00001) 2017-03-06T16:13:11Z |3b|: yeah, got past it this time 2017-03-06T16:15:13Z |3b|: and adding (member (fd-stream-fd *stdin*) (multiple-value-list (get-std-handles))) to windows-isatty seems to fix a:b for *inferior-lisp* (waiting for build to test on mingw) 2017-03-06T16:15:43Z stassats: what about pipes? 2017-03-06T16:16:21Z |3b| forgot about pipes 2017-03-06T16:17:08Z |3b| notes that it helps if i run the version i'm trying to test :/ 2017-03-06T16:18:20Z |3b|: and stuck at mailbox-trivia.3 again 2017-03-06T16:24:10Z |3b|: also helps if i save the file i'm trying to build :( 2017-03-06T16:24:22Z flip214: that's too easy. 2017-03-06T16:25:37Z schjetne quit (Ping timeout: 246 seconds) 2017-03-06T16:25:55Z |3b|: results from sap-ref thing: http://paste.lisp.org/+7AX6 2017-03-06T16:26:22Z stassats: |3b|: hm, no details? 2017-03-06T16:26:28Z stassats: ah 2017-03-06T16:26:31Z stassats: no 2017-03-06T16:26:36Z stassats: well, let me test here 2017-03-06T16:27:02Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-06T16:27:03Z stassats: well, that's i get for not testing 2017-03-06T16:27:51Z stassats: pushed more (without testing) 2017-03-06T16:28:19Z stassats: ok, i'm not anxious to deal with the mingw terminal, but backtraces seem to be borked everywhere, i'll look into that 2017-03-06T16:29:58Z |3b|: a:b problem applies to *inferior-lisp* in addition to mingw terminal 2017-03-06T16:30:12Z stassats: *inferior-lisp* is a pipe 2017-03-06T16:30:14Z |3b|: (which also might not be that high priority, since slime works otherwise) 2017-03-06T16:30:17Z |3b|: right 2017-03-06T16:30:37Z stassats: who the hell calls call-hooks 2017-03-06T16:30:38Z |3b|: but also a console 2017-03-06T16:31:22Z stassats: (SB-INT:CALL-HOOKS NIL 4750920 :ON-ERROR 25165960) 2017-03-06T16:31:28Z stassats: those 25165960 4750920 seem suspect 2017-03-06T16:31:31Z stassats: and NIL 2017-03-06T16:32:52Z sjl quit (Ping timeout: 260 seconds) 2017-03-06T16:33:10Z stassats: *initial-hooks*, probably 2017-03-06T16:35:09Z |3b| got a build failure, removing my changes and trying again 2017-03-06T16:36:58Z |3b| thinks dcdddd3 added some tab indentation if you care 2017-03-06T16:37:26Z |3b|: (more problems of lacking a good config on the temporary vm presumably) 2017-03-06T16:38:05Z stassats: strange, i committed from linux 2017-03-06T16:38:45Z |3b|: and stuck at mailbox-trivia.3 again 2017-03-06T16:39:05Z stassats: emacs didn't handle tabs when pasting 2017-03-06T16:39:12Z |3b| doesn't remember it being that bad in the past 2017-03-06T16:40:31Z stassats: ok, memory faults are properly fixed this time 2017-03-06T16:42:04Z stassats: setting *init-hooks* to nil, slading and backtraces are not crooked 2017-03-06T16:42:43Z stassats: though look a bit strange, there's a call-with-recursive-system-lock at the bottom 2017-03-06T16:42:44Z |3b|: yeah, "Unhandled memory fault at #x0." now 2017-03-06T16:42:46Z stassats: then a foreign function 2017-03-06T16:42:56Z stassats: only then there's restart-lisp 2017-03-06T16:44:57Z edgar-rft quit (Quit: edgar-rft) 2017-03-06T16:46:22Z attila_lendvai joined #sbcl 2017-03-06T16:46:22Z attila_lendvai quit (Changing host) 2017-03-06T16:46:22Z attila_lendvai joined #sbcl 2017-03-06T16:47:56Z |3b|: linux doesn't seem to have a problem with read errors from a pipe 2017-03-06T16:48:26Z |3b| isn't sure what is different about linux and windows there 2017-03-06T16:48:44Z nyef`: Linux fd APIs are typically saner than Windows HANDLE APIs. 2017-03-06T16:48:52Z stassats: well, windows console is not really a real file 2017-03-06T16:49:07Z stassats: so it has to do everything differently 2017-03-06T16:50:05Z stassats: ok, the foreign function in the backtrace is call_into_lisp, which is plausible, but why is call-with-recursive-system-lock before it 2017-03-06T16:51:58Z stassats: diving into ldb, there the backtrace looks sane and two frames shorter 2017-03-06T16:57:06Z |3b|: apparently you can seek a console/pipe on windows :/ 2017-03-06T16:57:23Z |3b|: or rather you can't but it does anyway or something 2017-03-06T17:00:26Z |3b|: so probably either unix-lseek or sb-win32:lseeki64 should check file type before continuing 2017-03-06T17:01:11Z |3b|: linux gets ESPIPE from pipe/socket/fifo/terminal 2017-03-06T17:03:31Z |3b|: i guess probably checking for file-type-disk rather than checking for not the others 2017-03-06T17:04:00Z |3b|: 'remote' is the only one that sounds like it might be seekable, and docs say it is unused 2017-03-06T17:08:44Z stassats: x86-call-context seems to be hosed on win32 2017-03-06T17:09:06Z sjl joined #sbcl 2017-03-06T17:09:45Z |3b|: checking for file-type-disk in sb-win32:lseeki64 also seems to fix a:b 2017-03-06T17:10:14Z stassats: why is it seeking something? 2017-03-06T17:10:37Z |3b|: to find file position 2017-03-06T17:11:13Z stassats: i see 2017-03-06T17:11:29Z stassats: |3b|: what windows version are you using? 2017-03-06T17:11:32Z |3b|: 8.1 2017-03-06T17:11:54Z stassats: did you ever see those endless backtraces before? 2017-03-06T17:11:59Z stassats: i don't remember them 2017-03-06T17:12:52Z |3b|: possibly 2017-03-06T17:13:47Z stassats: and i don't get how it came to be 2017-03-06T17:15:30Z attila_lendvai: stassats: I think you mean this one: https://bugs.launchpad.net/sbcl/+bug/1520694 2017-03-06T17:15:52Z stassats: attila_lendvai: no, this is different 2017-03-06T17:16:01Z stassats: the stack seems to be corrupted here 2017-03-06T17:16:27Z stassats: conveniently corrupted, though, it looks like real stack 2017-03-06T17:16:27Z attila_lendvai: oh, ok, I misunderstood "endless backtraces" 2017-03-06T17:18:47Z Bike joined #sbcl 2017-03-06T17:19:25Z shka_ joined #sbcl 2017-03-06T17:21:26Z stassats: somehow it goes to the function that was called before 2017-03-06T17:21:30Z stassats: and then loops to it 2017-03-06T17:22:25Z |3b|: http://paste.lisp.org/display/338524 is closest i can remember similar to the endless backtrace (from a test failure when running tests) 2017-03-06T17:22:29Z stassats: though even weirder, the function that was called, but it's shown after call_into_lisp 2017-03-06T17:22:33Z stassats: which couldn't happen 2017-03-06T17:22:49Z stassats: so it writes its return address somewhere wrong 2017-03-06T17:23:28Z stassats: |3b|: that's the same thing 2017-03-06T17:23:34Z stassats: so, i guess i didn't break it after all 2017-03-06T17:23:42Z stassats: well, with my windows changes, that is 2017-03-06T17:25:21Z stassats: the hook installs windows-console-control-handler 2017-03-06T17:25:24Z nyef`: ... if you broke it, it wasn't recently? (-: 2017-03-06T17:25:24Z |3b|: i think that paste was from 1.3.14.194 2017-03-06T17:25:55Z stassats: i think that's well before my windows changes 2017-03-06T17:26:16Z stassats: but i can't count 2017-03-06T17:30:21Z schjetne joined #sbcl 2017-03-06T17:30:44Z stassats: the arguments seem broken, but the function is actually called 2017-03-06T17:30:57Z stassats: a time some point 2017-03-06T17:31:26Z stassats: s/a time/at/ 2017-03-06T17:40:20Z stassats: it doesn't matter what's called in init hooks 2017-03-06T17:41:25Z stassats: but there's no init hooks on linux, maybe they're hosed too 2017-03-06T17:42:11Z stassats: no, too bad, that would've been easier to debug 2017-03-06T17:43:51Z |3b|: hmm, maybe it isn't pipes specifically on windows, since echo a:b | sbcl errors properly in mingw or cmd 2017-03-06T17:44:33Z stassats: calling print-backtrace from an init hook already shows a loop, but with proper argument values 2017-03-06T17:49:02Z stassats: a redefined call-hooks doesn't demonstrate the problem 2017-03-06T17:51:23Z stassats: redefining with safety 2 space 1 speed 2 does 2017-03-06T17:51:47Z |3b| can't find any difference between what works and doesn't with pipes 2017-03-06T17:52:01Z |3b|: ah, probably not checking the right things 2017-03-06T17:53:55Z stassats: ok, it's speed 1 vs speed 2 2017-03-06T17:53:57Z stassats: huuuh 2017-03-06T17:54:31Z m00natic quit (Remote host closed the connection) 2017-03-06T17:55:16Z stassats: ok, this is probably due to stack allocated restarts 2017-03-06T17:56:57Z edgar-rft joined #sbcl 2017-03-06T17:59:09Z stassats: notinline make-restart doesn't help 2017-03-06T18:30:13Z stassats: i'm out of ideas 2017-03-06T18:32:29Z rumbler3_ joined #sbcl 2017-03-06T18:37:00Z rumbler3_ quit (Ping timeout: 260 seconds) 2017-03-06T18:59:24Z sjl quit (Ping timeout: 260 seconds) 2017-03-06T19:45:18Z DGASAU joined #sbcl 2017-03-06T20:08:27Z slyrus joined #sbcl 2017-03-06T20:40:55Z angavrilov quit (Remote host closed the connection) 2017-03-06T20:47:09Z igajsin quit (Remote host closed the connection) 2017-03-06T21:11:01Z specbot quit (Remote host closed the connection) 2017-03-06T21:11:01Z minion quit (Remote host closed the connection) 2017-03-06T21:11:20Z neuronsong joined #sbcl 2017-03-06T21:19:21Z minion joined #sbcl 2017-03-06T21:19:23Z specbot joined #sbcl 2017-03-06T21:20:48Z minion quit (Remote host closed the connection) 2017-03-06T21:20:49Z specbot quit (Remote host closed the connection) 2017-03-06T21:22:17Z specbot joined #sbcl 2017-03-06T21:22:17Z minion joined #sbcl 2017-03-06T21:44:13Z sjl joined #sbcl 2017-03-06T21:47:24Z shka_ quit (Ping timeout: 260 seconds) 2017-03-06T21:51:46Z milanj quit (Quit: Leaving) 2017-03-06T22:33:06Z rumbler3_ joined #sbcl 2017-03-06T22:37:08Z rumbler3_ quit (Ping timeout: 240 seconds) 2017-03-06T22:49:07Z prxq joined #sbcl 2017-03-06T23:02:59Z gingerale quit (Remote host closed the connection) 2017-03-06T23:16:28Z slyrus quit (Ping timeout: 240 seconds) 2017-03-06T23:46:39Z leo_song left #sbcl 2017-03-07T00:00:03Z prxq quit (Remote host closed the connection) 2017-03-07T00:00:05Z sjl quit (Ping timeout: 240 seconds) 2017-03-07T00:00:19Z stassats: what if we run every foreign function in a separate thread? 2017-03-07T00:01:03Z nyef`: ... You mean, we have lisp threads, and foreign-proxy threads, but no lisp thread runs foreign code other than exception handling? 2017-03-07T00:01:09Z stassats: right 2017-03-07T00:01:23Z nyef`: I've seen that arrangement before, possibly in a Smalltalk system. 2017-03-07T00:01:31Z |3b|: what about foreign things with thread-local data? 2017-03-07T00:01:35Z stassats: the lisp thread waits on some condition variable thingy, that way you can always safely interrupt 2017-03-07T00:02:14Z |3b| is thinking in particular of opengl contexts, but seems like i've seen others 2017-03-07T00:02:32Z |3b|: also the "run gui code on main thread" in osx 2017-03-07T00:02:43Z stassats: i'm thinking about a completely signalless sbcl, for sbcl.so 2017-03-07T00:03:40Z stassats: right now even safepoints use interrupts for foreign code (except on windows, but that's utterly and exhaustively broken) 2017-03-07T00:03:47Z stassats: and safepoints themselves depend on MMU 2017-03-07T00:04:34Z stassats: (safepoints are broken everywhere, to be fair) 2017-03-07T00:05:34Z nyef`: Foreign threads with thread-local data is fine, it's a 1:1 correspondance to lisp threads. 2017-03-07T00:05:55Z stassats: some systems can accept sbcl.so with signals, but not without a fully relocatable core, so that should be a priority 2017-03-07T00:06:32Z nyef`: Hell, a MOSTLY relocatable core should be a good start. 2017-03-07T00:06:52Z stassats: but these days, safety and all, is sbcl.so even a good idea? process isolation and all that 2017-03-07T00:06:58Z nyef`: Fully relocatable would require a touch more work, since stuff in static and read-only space is occasionally referenced without fixups. 2017-03-07T00:07:53Z stassats: i'm dancing around a new emacs-less IDE, i'm actually thinking about running sbcl in a container, so that all the macros can be expanded and examined, downloaded from the evil internet 2017-03-07T00:09:09Z stassats: and there's always shmem 2017-03-07T00:14:33Z stassats: and that's why i was poking around write-sequence yesterday, measuring it up for socket ipc 2017-03-07T00:16:06Z stassats: now that write-sequence is using unix-write when possible, what does read-sequence do? 2017-03-07T00:20:58Z stassats: ansi-stream has in-buffer and fd-stream has ibuf 2017-03-07T00:24:59Z stassats: yeah, the both the i/o system and external-formats need to be redesigned 2017-03-07T00:29:30Z cromachina joined #sbcl 2017-03-07T00:59:27Z jamtho joined #sbcl 2017-03-07T01:15:02Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-07T01:24:59Z nyef`: emacs-less IDE, huh? Somewhere in my project idea list is a version of SBCL where the debugger doesn't run in the same address space as the code being debugged, possibly combined with running an emacs-type editor under a CLIM API. 2017-03-07T01:25:30Z stassats: yeah, no clim for me 2017-03-07T01:26:06Z nyef`: No CLIM for me either until I get NQ-CLIM working properly. 2017-03-07T01:26:33Z stassats: not least because there's no clim implementation 2017-03-07T01:26:57Z stassats: and then that the api itself is lacking 2017-03-07T01:27:58Z nyef`: Yeah, the CLIM II spec contains some really bizarre things. 2017-03-07T01:28:24Z stassats: so for the UI i'm settling for native solutions 2017-03-07T01:28:28Z stassats: like cocoa on macos 2017-03-07T01:29:28Z stassats: with all the processing done in lisp and communicated via tcp 2017-03-07T01:33:03Z rumbler31 joined #sbcl 2017-03-07T01:34:47Z stassats: maybe i can hook into ir1 for code analysis, because i'm tired of slime not understanding code 2017-03-07T01:35:51Z stassats: or at least being able to do real code walking, but that means expanding macros, which is potentially unsafe, although compiling code is just as unsafe 2017-03-07T01:37:39Z rumbler31 quit (Ping timeout: 260 seconds) 2017-03-07T01:40:07Z stassats: though even sbcl sometimes does not understand code, for source locations 2017-03-07T01:46:24Z jamtho quit (Ping timeout: 260 seconds) 2017-03-07T02:01:40Z stassats quit (Ping timeout: 256 seconds) 2017-03-07T05:12:34Z shka_ joined #sbcl 2017-03-07T05:55:50Z slyrus joined #sbcl 2017-03-07T05:58:15Z jack_rabbit joined #sbcl 2017-03-07T06:33:57Z slyrus quit (Ping timeout: 240 seconds) 2017-03-07T06:39:52Z Colleen quit (Ping timeout: 260 seconds) 2017-03-07T06:57:36Z shka_ quit (Ping timeout: 260 seconds) 2017-03-07T07:47:08Z angavrilov joined #sbcl 2017-03-07T07:49:47Z gingerale joined #sbcl 2017-03-07T09:25:57Z Bike quit (Quit: leaving) 2017-03-07T09:25:58Z DeadTrickster quit (Ping timeout: 268 seconds) 2017-03-07T09:43:27Z Lord_Nightmare quit (Ping timeout: 240 seconds) 2017-03-07T09:45:56Z Lord_Nightmare joined #sbcl 2017-03-07T09:49:02Z DeadTrickster joined #sbcl 2017-03-07T09:54:00Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-07T10:04:07Z jamtho joined #sbcl 2017-03-07T10:07:17Z DeadTrickster joined #sbcl 2017-03-07T10:12:40Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-07T10:23:03Z DeadTrickster joined #sbcl 2017-03-07T10:32:30Z jamtho quit (Ping timeout: 240 seconds) 2017-03-07T10:32:44Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-07T10:33:19Z Colleen joined #sbcl 2017-03-07T10:38:06Z DeadTrickster joined #sbcl 2017-03-07T10:39:20Z jdz quit (Ping timeout: 252 seconds) 2017-03-07T10:44:06Z jdz joined #sbcl 2017-03-07T10:44:31Z sjl joined #sbcl 2017-03-07T10:44:35Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-07T10:47:46Z schjetne quit (Ping timeout: 264 seconds) 2017-03-07T11:16:49Z attila_lendvai joined #sbcl 2017-03-07T11:16:49Z attila_lendvai quit (Changing host) 2017-03-07T11:16:49Z attila_lendvai joined #sbcl 2017-03-07T11:47:08Z m00natic joined #sbcl 2017-03-07T12:00:54Z schjetne joined #sbcl 2017-03-07T12:46:23Z dtornabene joined #sbcl 2017-03-07T13:01:59Z stassats joined #sbcl 2017-03-07T13:06:30Z attila_lendvai quit (Quit: Leaving.) 2017-03-07T13:06:34Z attila_lendvai1 joined #sbcl 2017-03-07T13:06:34Z attila_lendvai1 is now known as attila_lendvai 2017-03-07T13:06:34Z attila_lendvai quit (Changing host) 2017-03-07T13:06:34Z attila_lendvai joined #sbcl 2017-03-07T13:09:57Z hydan joined #sbcl 2017-03-07T13:13:20Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-07T13:54:23Z dtornabene is now known as cognotronicist 2017-03-07T14:18:48Z cromachina quit (Read error: Connection reset by peer) 2017-03-07T14:19:52Z sjl quit (Ping timeout: 246 seconds) 2017-03-07T14:21:51Z sjl joined #sbcl 2017-03-07T14:44:47Z scymtym: stassats: https://github.com/scymtym/sbcl/tree/wip-walk-forms-new-marco-stuff separates code walking from ir1 conversion (also has composable code walking hooks), but lacks the safe macro expansion. one of the examples does a little bit of accurate syntax highlighting. maybe one of the interpreters could be used to do macro expansion without affecting the global environment 2017-03-07T14:45:58Z scymtym: i would like to eventually (not in the near future, probably) merge that code if performance concerns don't make that infeasible 2017-03-07T14:55:16Z eschatologist quit (Ping timeout: 268 seconds) 2017-03-07T14:55:26Z eschatologist joined #sbcl 2017-03-07T16:00:48Z schjetne quit (Ping timeout: 260 seconds) 2017-03-07T16:08:02Z slyrus joined #sbcl 2017-03-07T16:26:44Z gargaml joined #sbcl 2017-03-07T17:06:28Z slyrus quit (Ping timeout: 246 seconds) 2017-03-07T17:19:06Z sjl quit (Ping timeout: 256 seconds) 2017-03-07T17:30:08Z schjetne joined #sbcl 2017-03-07T17:57:01Z m00natic quit (Remote host closed the connection) 2017-03-07T18:00:18Z shka_ joined #sbcl 2017-03-07T18:13:20Z cognotronicist quit (Quit: Leaving) 2017-03-07T18:16:51Z Bike joined #sbcl 2017-03-07T18:18:18Z sjl joined #sbcl 2017-03-07T18:27:45Z stassats: scymtym: but even the interpreter can call foreign code 2017-03-07T18:28:16Z attila_lendvai joined #sbcl 2017-03-07T18:28:16Z attila_lendvai quit (Changing host) 2017-03-07T18:28:16Z attila_lendvai joined #sbcl 2017-03-07T18:33:30Z scymtym: stassats: sure, baby steps. and some things probably just can't work 2017-03-07T18:34:40Z stassats: that's why i wanted to do in this newfangled docker thing 2017-03-07T19:00:49Z milanj joined #sbcl 2017-03-07T19:09:18Z akkad quit (Excess Flood) 2017-03-07T19:15:44Z akkad joined #sbcl 2017-03-07T19:20:47Z slyrus joined #sbcl 2017-03-07T19:30:07Z akkad quit (Excess Flood) 2017-03-07T19:32:44Z akkad joined #sbcl 2017-03-07T20:03:24Z reinuseslisp joined #sbcl 2017-03-07T20:31:16Z angavrilov quit (Remote host closed the connection) 2017-03-07T21:14:12Z stassats: if sbcl on windows-subsystem-for-linux, what do we need the native port for anyway 2017-03-07T21:14:26Z stassats: works 2017-03-07T21:15:06Z |3b|: interacting with windows :/ 2017-03-07T21:16:04Z Shinmera: Supporting versions other than Windows 10. 2017-03-07T21:17:13Z stassats: the build dies somewhere in cold-init 2017-03-07T21:17:32Z stassats: or rather, when reloading target-2 fasls 2017-03-07T21:18:40Z stassats: slad, it seems 2017-03-07T21:19:17Z stassats: though slad works from the host sbcl 2017-03-07T21:24:53Z stassats: ok, it is save-lisp-and-die 2017-03-07T21:28:20Z stassats: so, yeah, it can't exactly work yet 2017-03-07T21:28:23Z stassats: but it's close 2017-03-07T21:30:01Z prxq joined #sbcl 2017-03-07T21:41:37Z shka_ quit (Quit: Konversation terminated!) 2017-03-07T21:50:15Z reinuseslisp quit (Quit: Leaving) 2017-03-07T22:47:32Z prxq quit (Read error: Connection reset by peer) 2017-03-07T23:03:49Z edgar-rft quit (Quit: edgar-rft) 2017-03-07T23:27:35Z slyrus quit (Ping timeout: 240 seconds) 2017-03-07T23:45:34Z milanj quit (Quit: Leaving) 2017-03-08T00:04:08Z cromachina joined #sbcl 2017-03-08T00:18:55Z sjl quit (Ping timeout: 268 seconds) 2017-03-08T00:22:13Z gingerale quit (Remote host closed the connection) 2017-03-08T01:58:56Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-08T02:59:28Z logrus joined #sbcl 2017-03-08T03:14:40Z adlai quit (Ping timeout: 256 seconds) 2017-03-08T03:28:01Z adlai joined #sbcl 2017-03-08T03:42:49Z oleo quit (Read error: Connection reset by peer) 2017-03-08T04:06:18Z eschatologist quit (Quit: ZNC 1.6.3+deb2 - http://znc.in) 2017-03-08T04:06:56Z eschatologist joined #sbcl 2017-03-08T04:32:08Z stassats quit (Ping timeout: 260 seconds) 2017-03-08T04:36:19Z logrus quit (Quit: Textual IRC Client: www.textualapp.com) 2017-03-08T05:17:02Z oleo joined #sbcl 2017-03-08T05:27:47Z nyef`: ... a decade later, give or take, and I /finally/ get the joke behind the TI LispM function "PROCESS-SCHEDULER-FOR-CHAPARRAL". /-: 2017-03-08T05:28:54Z nyef`: (Actually, CHAPARRAL shows up in a couple of other places, but still... Takes me a decade to pick up on a joke like that?!?) 2017-03-08T06:10:00Z shka_ joined #sbcl 2017-03-08T06:25:23Z oleo quit (Quit: Verlassend) 2017-03-08T06:47:05Z shka_ quit (Ping timeout: 240 seconds) 2017-03-08T07:15:55Z Bike quit (Quit: slep) 2017-03-08T08:18:42Z angavrilov joined #sbcl 2017-03-08T08:42:23Z akkad quit (Excess Flood) 2017-03-08T08:50:43Z akkad joined #sbcl 2017-03-08T09:32:35Z gingerale joined #sbcl 2017-03-08T09:35:43Z DeadTrickster joined #sbcl 2017-03-08T09:41:24Z igajsin joined #sbcl 2017-03-08T10:11:13Z sjl joined #sbcl 2017-03-08T10:23:21Z tormar joined #sbcl 2017-03-08T10:24:03Z tormar: Cross-posting my question from #lisp: Need help getting slimv to work (SLIME for Vim) on Windows. For some reason the server will not start. To debug I start SBCL in the directory of my slimv bundle and evaluate (load "swank-loader.lisp") (swank-loader:init). This signals an error: Can't create directory c:\Users\tormar\.slime\fasl\2.18 2017-03-08T10:24:27Z tormar: c:\Users\tormar\.slime\fasl\ exist, and I've even tried to manually create folder 2.18, but keep getting this error. Tried to start SBCL as Administrator, no luck. 2017-03-08T10:30:04Z sigjuice_ quit (Ping timeout: 260 seconds) 2017-03-08T10:31:22Z sigjuice joined #sbcl 2017-03-08T10:32:02Z sjl__ joined #sbcl 2017-03-08T10:34:22Z sjl quit (Ping timeout: 246 seconds) 2017-03-08T10:34:52Z attila_lendvai joined #sbcl 2017-03-08T10:34:52Z attila_lendvai quit (Changing host) 2017-03-08T10:34:52Z attila_lendvai joined #sbcl 2017-03-08T11:00:55Z salva quit (Read error: Connection reset by peer) 2017-03-08T11:01:33Z salva joined #sbcl 2017-03-08T11:01:35Z salva quit (Read error: Connection reset by peer) 2017-03-08T11:02:23Z salva joined #sbcl 2017-03-08T11:23:15Z sjl__ is now known as sjl 2017-03-08T11:36:13Z sjl quit (Read error: Connection reset by peer) 2017-03-08T11:48:03Z sjl joined #sbcl 2017-03-08T11:48:56Z schjetne quit (Ping timeout: 260 seconds) 2017-03-08T11:55:26Z sjl quit (Ping timeout: 256 seconds) 2017-03-08T12:07:58Z stassats joined #sbcl 2017-03-08T12:39:26Z edgar-rft joined #sbcl 2017-03-08T12:41:53Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-08T12:45:18Z sjl joined #sbcl 2017-03-08T13:08:39Z schjetne joined #sbcl 2017-03-08T13:09:51Z whiteline quit (Remote host closed the connection) 2017-03-08T13:17:31Z m00natic joined #sbcl 2017-03-08T13:21:49Z tormar left #sbcl 2017-03-08T13:22:00Z tormar joined #sbcl 2017-03-08T13:26:02Z tormar quit (Quit: Leaving) 2017-03-08T13:34:58Z sjl__ joined #sbcl 2017-03-08T13:37:08Z sjl quit (Ping timeout: 240 seconds) 2017-03-08T13:39:44Z attila_lendvai joined #sbcl 2017-03-08T13:44:18Z sjl__ quit (Read error: Connection reset by peer) 2017-03-08T13:44:30Z tormar joined #sbcl 2017-03-08T13:46:00Z sjl__ joined #sbcl 2017-03-08T13:46:30Z whiteline joined #sbcl 2017-03-08T13:57:04Z sjl__ is now known as sjl 2017-03-08T14:18:16Z schjetne quit (Ping timeout: 260 seconds) 2017-03-08T14:19:56Z cromachina quit (Read error: Connection reset by peer) 2017-03-08T14:21:06Z schjetne joined #sbcl 2017-03-08T14:25:10Z jamtho joined #sbcl 2017-03-08T14:33:56Z edgar-rft quit (Quit: edgar-rft) 2017-03-08T14:52:33Z m00natic quit (Remote host closed the connection) 2017-03-08T14:55:59Z sjl quit (Ping timeout: 240 seconds) 2017-03-08T14:58:29Z jamtho quit (Ping timeout: 240 seconds) 2017-03-08T15:00:39Z gargaml quit (Quit: WeeChat 1.7) 2017-03-08T15:04:05Z tormar quit (Ping timeout: 240 seconds) 2017-03-08T15:17:29Z gargaml joined #sbcl 2017-03-08T15:18:38Z gargaml quit (Client Quit) 2017-03-08T15:18:46Z gargaml joined #sbcl 2017-03-08T15:22:50Z jamtho joined #sbcl 2017-03-08T15:34:22Z jamtho quit (Ping timeout: 260 seconds) 2017-03-08T15:39:28Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-08T15:40:05Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-08T16:04:27Z schjetne quit (Ping timeout: 240 seconds) 2017-03-08T16:08:09Z attila_lendvai joined #sbcl 2017-03-08T16:20:41Z karswell joined #sbcl 2017-03-08T16:22:50Z oleo joined #sbcl 2017-03-08T16:23:58Z karswell` joined #sbcl 2017-03-08T16:24:26Z sjl joined #sbcl 2017-03-08T16:24:47Z karswell quit (Read error: Connection reset by peer) 2017-03-08T16:59:17Z stassats: trying to make sense of the windows backtrace problems, the difference between a good state and bad state is 208 bytes of RSP 2017-03-08T16:59:32Z stassats: so is something not cleaning up the stack properly? 2017-03-08T17:29:03Z stassats: rounded down to (labels ((f () (let (*) (reinit) (funcall toplevel)))) (declare (notinline f)) (unwind-protect (f))) 2017-03-08T17:29:45Z stassats: though what reinit does is important as well 2017-03-08T17:30:19Z stassats: with (print 10) after (reinit) it's not triggered 2017-03-08T17:32:56Z stassats: the fact that this is happening during start up complicates debugging 2017-03-08T17:35:09Z stassats: or how toplevel is called 2017-03-08T17:39:48Z stassats: linux or macos do not have call-into-lisp in their backtraces 2017-03-08T17:41:31Z stassats: that's no the main thread, new threads do have it 2017-03-08T17:42:56Z slyrus joined #sbcl 2017-03-08T17:43:02Z |3b|: does the build directly use save-lisp-and-die to create the core? 2017-03-08T17:43:05Z nyef`: Something-something call_into_lisp_first_time, stack switching, exception-handling-time stack-bounds checks. 2017-03-08T17:43:32Z stassats: yes, i'm looking at call_into_lisp_first_time, but it's used on windows too 2017-03-08T17:43:43Z stassats: |3b|: make-target-2 does 2017-03-08T17:43:53Z nyef`: The windows port does/did it differently because windows does/did more validation. 2017-03-08T17:44:06Z stassats: |3b|: make-host-2 uses genesis to create a cold core 2017-03-08T17:44:23Z |3b|: the final one that the user sees i mean 2017-03-08T17:44:39Z stassats: make-target-2 is the final build step 2017-03-08T17:45:45Z milanj joined #sbcl 2017-03-08T17:45:58Z stassats: so i if i remove call_into_lisp_first_time on linux, add an *init-hook*, maybe i'll be able to recreate this 2017-03-08T17:47:17Z stassats: that didn't go over well 2017-03-08T17:47:58Z stassats: ok, windows has something about fomit_frame_pointer 2017-03-08T17:48:19Z stassats: no explanation provided 2017-03-08T17:49:09Z stassats: was part of the windows ports 2017-03-08T17:50:56Z stassats: if i comment that fomit_frame_pointer feature, the backtrace is normal 2017-03-08T17:53:58Z gargaml quit (Ping timeout: 246 seconds) 2017-03-08T17:56:07Z Colleen quit (Read error: Connection reset by peer) 2017-03-08T17:56:10Z stassats: ok, there should be no fomit frame pointer actually going on, so this feature somehow mucks the stack up 2017-03-08T17:57:31Z stassats: x86-64-assem.S is intel syntax? 2017-03-08T17:59:09Z Colleen joined #sbcl 2017-03-08T18:01:41Z oleo quit (Remote host closed the connection) 2017-03-08T18:03:47Z stassats: it looks like the ise fomit_frame_pointer thingy ignores call_into_lisp_first_time, since RAX in that case is not RBP but THREAD 2017-03-08T18:09:36Z Colleen quit (Remote host closed the connection) 2017-03-08T18:10:16Z Colleen joined #sbcl 2017-03-08T18:10:29Z stassats: yeah, that seems to be the problem 2017-03-08T18:13:55Z slyrus quit (Ping timeout: 246 seconds) 2017-03-08T18:24:53Z stassats: but i don't quite get what it's trying to achieve 2017-03-08T18:32:16Z oleo joined #sbcl 2017-03-08T18:36:02Z nyef`: ... On x86oids, I thought that we never wanted to omit the frame pointer? 2017-03-08T18:36:23Z nyef`: Specifically because we don't do table-based backtrace for foreign code. 2017-03-08T18:36:26Z stassats: library code might think otherwise 2017-03-08T18:37:05Z nyef`: Sure, we can't help that, and it'd screw up the trace, but the bits that we CAN control we tend to like to have frame pointers on. 2017-03-08T18:38:07Z akkad quit (Excess Flood) 2017-03-08T18:38:55Z stassats: what i figure it does is saving a fake rbp 2017-03-08T18:38:59Z stassats: then restoring it 2017-03-08T18:39:47Z stassats: where does the fake rbp come from? somewhere from a callback wrapper 2017-03-08T18:40:08Z stassats: i don't want to understand that, just want to have call_into_lisp_first_time not suck 2017-03-08T18:41:28Z stassats: and i don't see the concrete problem with it now 2017-03-08T18:41:51Z stassats: ok, i see it again 2017-03-08T18:43:14Z akkad joined #sbcl 2017-03-08T18:55:40Z stassats: this actually looks like a stack alignment issue 2017-03-08T18:58:16Z oleo quit (Ping timeout: 260 seconds) 2017-03-08T19:06:38Z stassats: x86-64-assem comments have some vestiges of old return conventions 2017-03-08T19:07:04Z stassats: /* If the function returned multiple values, it will return to this point. Lose them */ 2017-03-08T19:08:24Z oleo joined #sbcl 2017-03-08T19:11:53Z stassats: ok, it is an alignment issue indeed 2017-03-08T19:12:11Z stassats: that's why it was so tricky to trigger 2017-03-08T19:12:55Z edgar-rft joined #sbcl 2017-03-08T19:19:32Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-08T19:20:34Z slyrus joined #sbcl 2017-03-08T19:21:39Z stassats: that fixed, bring on the next win32 challenge 2017-03-08T19:22:00Z minion quit (Remote host closed the connection) 2017-03-08T19:22:00Z specbot quit (Remote host closed the connection) 2017-03-08T19:23:48Z stassats: but does this feature actually work? it passes the argument to a c function in %rcx, not %rdi 2017-03-08T19:24:26Z stassats: i don't want to test this, but i'm sure it's wrong, so i'll just change it 2017-03-08T19:24:30Z pkhuong: win64 is different 2017-03-08T19:24:46Z pkhuong: win64 goes rcx, rdx, r8, r9. 2017-03-08T19:25:26Z stassats: ok, that means that all the -LISP_FEAATURE_GCC_TLS will be broken on win32, if it ever decides to not use them 2017-03-08T19:27:00Z stassats: but alloc_tramp passes the value in %rdi as well 2017-03-08T19:27:57Z stassats: and doesn't use the shadow space 2017-03-08T19:28:32Z stassats: as broken as the win32 port is, i don't think allocating has ever shown problems 2017-03-08T19:29:39Z minion joined #sbcl 2017-03-08T19:30:14Z pkhuong: win32 or win64? 2017-03-08T19:30:19Z specbot joined #sbcl 2017-03-08T19:30:30Z stassats: win32-x86-64 2017-03-08T19:32:28Z stassats: it's marked AMD64_SYSV_ABI 2017-03-08T19:36:08Z shka_ joined #sbcl 2017-03-08T19:38:29Z stassats: but why was the stack alignment important for lisp code 2017-03-08T19:39:42Z stassats: we misalign the stack on purpose, so why does it matter here 2017-03-08T19:40:52Z |3b|: does it do any SIMD things on DX values that needs alignment? 2017-03-08T19:41:15Z |3b|: (assuming the wouldn't be aligned on their own) 2017-03-08T19:41:19Z fe[nl]ix: stassats: why on purpose ? 2017-03-08T19:41:37Z stassats: fe[nl]ix: size calculation 2017-03-08T19:42:05Z stassats: |3b|: it works fine, the stack pointer is just crooked and backtrace gets confused 2017-03-08T19:46:39Z fe[nl]ix: nyef`: any news on whether you can review the ELF code ? 2017-03-08T19:48:32Z fe[nl]ix: also, I'm in Cambridge until end of next week 2017-03-08T19:52:17Z pkhuong: fe[nl]ix: do you think you can swing by NYC? (: 2017-03-08T19:55:38Z tormar joined #sbcl 2017-03-08T20:15:48Z nyef`: fe[nl]ix: On the first commit, "Import libelf...", it looks like LIBELF_LIB, LIBELF_OBS, and LIBELF_CLEAN may be referenced without being defined if LISP_FEATURE_SB_ELF_CORE isn't defined. 2017-03-08T20:18:48Z nyef`: fe[nl]ix: The third commit, "Move stack smashing code", why does it forward-declare the new function instead of just defining it before using it? 2017-03-08T20:21:20Z stassats: i don't know if it's just the backtrace having problems with foreign frames and misaligned frames or there is a bigger problem 2017-03-08T20:22:13Z nyef`: stassats: It's SBCL. And Windows, but still SBCL. Of *course* there's a bigger problem. /-: 2017-03-08T20:22:30Z stassats: there may be a problem not on windows 2017-03-08T20:22:36Z stassats: not just 2017-03-08T20:22:50Z nyef`: See? Bigger problem already! d-: 2017-03-08T20:22:53Z stassats: since only windows had misaligned rsp comint of call_into_lisp 2017-03-08T20:23:15Z stassats: but we have misaligned rsp all the time 2017-03-08T20:26:26Z nyef`: fe[nl]ix: I am in the area at least through the end of this week, but I'm probably not here next week. 2017-03-08T20:28:38Z stassats: x86-call-context calls (control-stack-pointer-valid-p ocfp) with align-t 2017-03-08T20:29:22Z stassats: and so what? it should just not succeed, if anything 2017-03-08T20:31:12Z sjl quit (Read error: Connection reset by peer) 2017-03-08T20:33:50Z stassats: it only checks for single-word alignment 2017-03-08T20:41:01Z nyef`: fe[nl]ix: I'm seeing a few things in the fourth commit, "Add support for..." that don't quite seem right. I don't really have the patience today to work through the rest of this, but I'll try to keep poking at it (and make notes) over the next few days. 2017-03-08T20:46:42Z fe[nl]ix: pkhuong: I could come during the weekend 2017-03-08T20:46:44Z fe[nl]ix: let me think 2017-03-08T20:47:24Z fe[nl]ix: nyef`: in a makefile referencing an undefined variable in a dependency list makes it a no-op 2017-03-08T20:47:29Z fe[nl]ix: by design 2017-03-08T20:48:00Z quadresce joined #sbcl 2017-03-08T20:48:03Z fe[nl]ix: nyef`: good point, i'll remove the forward declaration 2017-03-08T20:48:17Z nyef`: fe[nl]ix: But it allows the external environment to define the variable, which is more what I'm concerned about. 2017-03-08T20:48:32Z fe[nl]ix: pkhuong: so you're not in Montreal any more ? 2017-03-08T20:49:10Z fe[nl]ix: nyef`: yes but you can have a make invocation override anything in a Makefile 2017-03-08T20:49:29Z fe[nl]ix: we just rely on nobody doing that 2017-03-08T20:49:42Z fe[nl]ix: imagine if the external enviroment redefined LIBS 2017-03-08T20:50:25Z pkhuong: fe[nl]ix: moved to NYC for work after my phd 2017-03-08T20:50:32Z nyef`: Typically, a Makefile would override LIBS anyway. 2017-03-08T20:51:36Z fe[nl]ix: nyef`: I could add a few underscores before the variable name, but I'm not sure what more we can do 2017-03-08T20:51:41Z nyef`: I'm not saying that you have to change it, of course, just "it's something to consider". 2017-03-08T20:52:44Z fe[nl]ix: ok 2017-03-08T20:55:50Z quadresce: hey folks 2017-03-08T20:56:46Z stassats: actually, with the aligned stack there's some bogus thingy in OCFP, that's why it stops 2017-03-08T20:57:54Z pkhuong: fe[nl]ix: but this coming weekend I'll be in new mexico (: 2017-03-08T20:58:15Z fe[nl]ix: oh, bummer 2017-03-08T20:58:36Z fe[nl]ix: I should be back in a few months 2017-03-08T21:14:59Z tormar quit (Quit: Leaving) 2017-03-08T21:20:58Z igajsin quit (Ping timeout: 256 seconds) 2017-03-08T21:31:16Z stassats: i think the problem is that incorrectly detects the lisp stack transition 2017-03-08T21:37:32Z schjetne joined #sbcl 2017-03-08T21:49:45Z stassats: some strange stack configuration 2017-03-08T21:55:27Z shka_ quit (Ping timeout: 240 seconds) 2017-03-08T21:56:58Z stassats: call_into_lisp starts with rbp and rsp already within the lisp stack 2017-03-08T22:06:14Z angavrilov quit (Remote host closed the connection) 2017-03-08T22:11:14Z stassats: win32 the stack away from the thread struct (wasted space, btw) 2017-03-08T22:12:37Z stassats: and call_into_lisp_first_time moves into that stack, but the parent frame is already inside that stack 2017-03-08T22:12:39Z stassats: i don't get it 2017-03-08T22:14:52Z stassats: and it's clear now why i'm seeing functions that were called previously, because it overlaps with the old stack 2017-03-08T22:15:19Z stassats: and my alignment fix is no good, it just moves away the target and makes the stack look like it's invalid 2017-03-08T22:18:00Z stassats: nyef`: do you know what's going on? 2017-03-08T22:19:18Z gingerale quit (Remote host closed the connection) 2017-03-08T22:23:11Z nyef`: For Win32, I distinctly recall using call_into_lisp rather than call_into_lisp_first_time. 2017-03-08T22:23:35Z stassats: that's not important 2017-03-08T22:24:55Z stassats: so, there's some windows stack, hiding in the GS segment, and we init our stack from it, but the initial thread is already in the middle, so we stumble on the old stack 2017-03-08T22:25:23Z stassats: probably in because of that empty third stack slot (but that's not important either) 2017-03-08T22:26:11Z nyef`: Well, no, it sortof *is* important. If we don't need/want to switch stacks, we shouldn't be reinitializing the stack, surely? 2017-03-08T22:26:36Z stassats: maybe we do want to switch 2017-03-08T22:27:04Z stassats: it never returns, might as well restart, save for some garbage 2017-03-08T22:27:26Z stassats: but i guess those saved frames are inconsequential 2017-03-08T22:27:35Z stassats: so, for my sanity i guess i'll try getting rid of call_into_lisp_first_time 2017-03-08T22:28:13Z nyef`: On 32-bit, IIRC, there's the SEH frame chain to worry about. Not sure about 64-bit. 2017-03-08T22:28:56Z stassats: but i don't see any indication that win32 ever not called call_into_lisp_first_time 2017-03-08T22:31:02Z nyef`: Hrm. Ah, I see. I didn't change which function was called, I disabled the stack switch in call_into_lisp_first_time. 2017-03-08T22:33:30Z stassats: that didn't make it into x86-64 2017-03-08T22:38:02Z stassats: ok, the stack doesn't loop and there's an additional foreign function in the backtrace 2017-03-08T22:38:31Z jamtho joined #sbcl 2017-03-08T22:38:48Z stassats: no dladdr, i take it? 2017-03-08T22:40:34Z stassats: create_initial_thread 2017-03-08T22:40:39Z nyef`: Different linkage mechanism. GetSymbolAddressA or something like that? 2017-03-08T22:41:03Z stassats: that seems like the right solution 2017-03-08T22:41:17Z stassats: nyef`: it's SymFromAddr but it needs to link with something additional 2017-03-08T22:49:31Z akkad quit (Excess Flood) 2017-03-08T22:51:29Z nyef`: ... Ah, right, the mapping from address to symbol, not the other way around. 2017-03-08T22:52:04Z nyef`: It's been far too long since I've done much winapi stuff, and I've never really done any 64-bit winapi stuff, so I'm somewhat out of my depth there. 2017-03-08T22:52:12Z stassats: good thing i didn't find my previous solution satisfying 2017-03-08T22:52:53Z stassats: and looks like i'm approaching my monthly dose of sbcl-windows work 2017-03-08T22:55:44Z akkad joined #sbcl 2017-03-08T23:06:28Z edgar-rft quit (Quit: edgar-rft) 2017-03-08T23:13:29Z jamtho quit (Ping timeout: 240 seconds) 2017-03-08T23:18:02Z milanj quit (Quit: Leaving) 2017-03-08T23:43:14Z attila_lendvai joined #sbcl 2017-03-08T23:43:14Z attila_lendvai quit (Changing host) 2017-03-08T23:43:14Z attila_lendvai joined #sbcl 2017-03-08T23:43:23Z sjl joined #sbcl 2017-03-08T23:56:36Z Bicyclidine joined #sbcl 2017-03-09T00:02:35Z cromachina joined #sbcl 2017-03-09T00:11:11Z slyrus quit (Ping timeout: 260 seconds) 2017-03-09T00:18:14Z slyrus joined #sbcl 2017-03-09T00:42:41Z jamtho joined #sbcl 2017-03-09T00:51:31Z neuronsong quit (Ping timeout: 246 seconds) 2017-03-09T01:00:29Z jamtho quit (Ping timeout: 240 seconds) 2017-03-09T01:08:12Z sjl quit (Read error: Connection reset by peer) 2017-03-09T01:09:43Z slyrus quit (Ping timeout: 246 seconds) 2017-03-09T01:17:35Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-09T01:17:53Z slyrus joined #sbcl 2017-03-09T01:18:19Z jamtho joined #sbcl 2017-03-09T01:34:50Z karswell` quit (Ping timeout: 256 seconds) 2017-03-09T01:35:37Z slyrus quit (Ping timeout: 246 seconds) 2017-03-09T02:02:49Z Suzuran joined #sbcl 2017-03-09T02:03:03Z Suzuran: nyef`: ping, ping~ 2017-03-09T02:03:57Z stassats quit (Ping timeout: 240 seconds) 2017-03-09T02:07:06Z nyef`: Suzuran: ECHO REPLY 2017-03-09T02:07:16Z Suzuran: Hi there! Been awhile. 2017-03-09T02:07:20Z Suzuran: How's life treating you? 2017-03-09T02:07:47Z nyef`: Well enough. Yourself? 2017-03-09T02:08:03Z Suzuran: Not so much. My kidneys went byebye and I'm on dialysis now. 2017-03-09T02:08:21Z nyef`: Oof. That's gotta suck. 2017-03-09T02:08:31Z jamtho quit (Ping timeout: 246 seconds) 2017-03-09T02:09:18Z Suzuran: Anyway, I got something interesting to show off 2017-03-09T02:09:29Z Suzuran: I figured you could appreciate it 2017-03-09T02:09:30Z Suzuran: https://dl.dropboxusercontent.com/u/12368207/Screenshots/1o.jpeg 2017-03-09T02:09:45Z Suzuran: It's the brute-forcest of brute-force flying-pig bullshit, but it IS. 2017-03-09T02:10:28Z nyef`: Sweet! 2017-03-09T02:10:29Z Suzuran: Took about a year and a half from the word "Go", should see a release in the next several weeks. 2017-03-09T02:10:49Z Suzuran: The problem is that someone has to come up with a distribution tape. 2017-03-09T02:10:53Z Suzuran: I don't have one. 2017-03-09T02:14:53Z nyef`: Hrm. I think that any tape images for the original system that I may-or-may-not have would be corrupt. 2017-03-09T02:16:49Z Suzuran: That sucks. 2017-03-09T02:17:25Z Suzuran: I have parts of it - I have the standalone utilities and firmware partition from bitsavers (they have copies of the optional unix for it, and one of the unix tapes had the installation bits) 2017-03-09T02:17:47Z Suzuran: but no actually released microcode, load band, or filesystem. 2017-03-09T02:19:50Z Suzuran: I got pretty far on skulldugged bits, then some helpful soul provided bitsavers with firmware rom dumps 2017-03-09T02:20:08Z Suzuran: and some nice pictures of the boards 2017-03-09T02:21:33Z nyef`: Is this the LMI version of the Explorer, or is this something similar-but-different? 2017-03-09T02:22:16Z Suzuran: Apparently they were compatible at one point early in their development, but it's very different in final form. 2017-03-09T02:22:41Z Suzuran: They're both nubus machines, but the processors and peripherals differ 2017-03-09T02:23:13Z Suzuran: The software is more like a super CADR rather than the Symbolics-like environment that the Explorer wound up with 2017-03-09T02:23:22Z nyef`: Not surprised. AUIU, LMI had a CADR license, TI licensed the design from LMI and made it the Explorer, and then licensed the Explorer design back to LMI. 2017-03-09T02:23:52Z Suzuran: The firmware has lots of TI fingerprints on it, and there's some similarity in bus structure bits, but that's about it. 2017-03-09T02:24:37Z nyef`: Hrm. Do you have the microcode assembler itself? 2017-03-09T02:24:40Z Suzuran: It has CL and TCP and Chaosnet over ethernet and such, but still looks CADR-ish 2017-03-09T02:24:45Z Suzuran: Yes 2017-03-09T02:24:55Z nyef`: Ooh. Is that on the bitsavers tape images? 2017-03-09T02:24:59Z Suzuran: Nope. 2017-03-09T02:25:17Z Suzuran: It should be on a release tape though 2017-03-09T02:25:27Z nyef`: Ah, okay. 2017-03-09T02:25:44Z Suzuran: You're familiar with the end of the ITSes right? 2017-03-09T02:26:03Z nyef`: If I was, my memory is a bit rusty. 2017-03-09T02:26:28Z Suzuran: Is it enough to remember what happened before the last two KSes were turned off? 2017-03-09T02:27:09Z nyef`: Something about full backups? 2017-03-09T02:27:31Z Suzuran: Someone at LMI behaved similarly 2017-03-09T02:27:36Z Suzuran: at the end 2017-03-09T02:27:39Z nyef`: Ahh. 2017-03-09T02:28:04Z Suzuran: Unfortunately it had the same result 2017-03-09T02:28:36Z nyef`: Huge amount of data, semi-corrupt tapes after all this time, and nothing can actually read the damned things anymore? 2017-03-09T02:29:01Z Suzuran: Close, but the other direction. 2017-03-09T02:29:21Z nyef`: Oh, destruction, not preservation? 2017-03-09T02:29:48Z Suzuran: Huge amount of data, lasted all this time, but since there's no clear license... 2017-03-09T02:30:08Z Suzuran: Well 2017-03-09T02:30:12Z nyef`: Ah! 2017-03-09T02:30:18Z Suzuran: Not so much "no clear license" as "it's blatantly stolen" 2017-03-09T02:31:03Z Suzuran: Actually distributed tapes could be explained away, but THIS stuff... 2017-03-09T02:31:33Z nyef`: Lovely. So the only real angles available are figure out who owns the rights, buy the rights, and then "find" copies to use, or wait until any copyright claim has expired... 2017-03-09T02:33:32Z Suzuran: Nobody really knows who wound up with the copyright. Chances are good whoever it is doesn't know themselves. They wouldn't notice or care about distribution of a real release tape running around. It would be really hard to work out who to sue over it anyway. 2017-03-09T02:35:20Z Suzuran: As I understand it, they were playing a game of musical liabilities at the end of things and nobody really knows who was holding what when the music stopped. 2017-03-09T02:36:18Z Suzuran: The only contestants have since gone bankrupt or been acquired 2017-03-09T02:37:18Z Suzuran: In a way it's worse than the situation with TI - At least with TI we knew who we'd be hearing from if someone got upset. 2017-03-09T02:49:17Z nyef`: Mmm. 2017-03-09T02:50:04Z nyef`: I need to head out to get food. I'll be back in an hour and a half to two hours, and probably have another hour or two of runtime after that before I crash. 2017-03-09T02:50:30Z Suzuran: Yeah, today was a dialysis day so I am on borrowed time. I just showed up on a lark. 2017-03-09T02:50:53Z nyef`: Well, thank you for showing up, it was good to "talk" to you again. 2017-03-09T02:51:17Z Suzuran: Yeah, I'll let you know as things progress 2017-03-09T02:51:41Z Suzuran: I'll probably hang around, at least until comcast does their thing next 2017-03-09T02:54:28Z nyef`: Okay, heading out now. I'll hope to see you when I get back, or otherwise soon. Don't wait up for me, though. (-: 2017-03-09T03:13:59Z hydan quit (Ping timeout: 240 seconds) 2017-03-09T03:22:18Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-09T05:44:55Z igajsin joined #sbcl 2017-03-09T05:54:58Z fiddlerwoaroof quit (Ping timeout: 264 seconds) 2017-03-09T06:24:03Z slyrus joined #sbcl 2017-03-09T06:37:00Z fiddlerwoaroof joined #sbcl 2017-03-09T06:37:02Z myrkraverk quit (Remote host closed the connection) 2017-03-09T06:37:05Z slyrus quit (Ping timeout: 240 seconds) 2017-03-09T06:51:54Z shka_ joined #sbcl 2017-03-09T06:54:36Z angavrilov joined #sbcl 2017-03-09T06:59:50Z myrkraverk joined #sbcl 2017-03-09T07:28:38Z Bicyclidine quit (Ping timeout: 268 seconds) 2017-03-09T07:42:12Z schjetne quit (Ping timeout: 260 seconds) 2017-03-09T07:44:57Z shka_ quit (Ping timeout: 240 seconds) 2017-03-09T07:52:20Z jack_rabbit joined #sbcl 2017-03-09T08:05:30Z gargaml joined #sbcl 2017-03-09T09:06:19Z schjetne joined #sbcl 2017-03-09T09:09:56Z mateuszb quit (Ping timeout: 260 seconds) 2017-03-09T09:13:04Z edgar-rft joined #sbcl 2017-03-09T09:22:04Z jamtho joined #sbcl 2017-03-09T09:26:22Z jamtho quit (Ping timeout: 246 seconds) 2017-03-09T09:29:48Z mateuszb joined #sbcl 2017-03-09T10:16:54Z sjl joined #sbcl 2017-03-09T11:00:37Z m00natic joined #sbcl 2017-03-09T11:12:52Z schjetne quit (Ping timeout: 258 seconds) 2017-03-09T12:04:03Z schjetne joined #sbcl 2017-03-09T12:10:37Z attila_lendvai joined #sbcl 2017-03-09T12:10:49Z attila_lendvai quit (Changing host) 2017-03-09T12:10:49Z attila_lendvai joined #sbcl 2017-03-09T12:26:32Z congnotronicist joined #sbcl 2017-03-09T12:33:22Z schjetne quit (Ping timeout: 264 seconds) 2017-03-09T12:51:15Z schjetne joined #sbcl 2017-03-09T13:09:21Z schjetne quit (Ping timeout: 260 seconds) 2017-03-09T13:31:50Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-09T14:20:34Z sjl quit (Ping timeout: 268 seconds) 2017-03-09T14:23:45Z schjetne joined #sbcl 2017-03-09T14:33:59Z cromachina quit (Read error: Connection reset by peer) 2017-03-09T14:38:21Z stassats joined #sbcl 2017-03-09T14:47:56Z schjetne quit (Ping timeout: 260 seconds) 2017-03-09T15:32:56Z sjl joined #sbcl 2017-03-09T15:42:33Z Bicyclidine joined #sbcl 2017-03-09T15:47:07Z gingerale joined #sbcl 2017-03-09T16:00:50Z slyrus joined #sbcl 2017-03-09T16:08:26Z jamtho joined #sbcl 2017-03-09T16:23:34Z jamtho quit (Ping timeout: 246 seconds) 2017-03-09T16:42:03Z quadresce joined #sbcl 2017-03-09T17:01:44Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-09T17:20:48Z sjl quit (Ping timeout: 240 seconds) 2017-03-09T17:48:56Z gargaml quit (Ping timeout: 256 seconds) 2017-03-09T17:50:35Z schjetne joined #sbcl 2017-03-09T18:03:20Z m00natic quit (Remote host closed the connection) 2017-03-09T18:03:26Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-09T18:04:53Z quadresce joined #sbcl 2017-03-09T18:12:29Z shka_ joined #sbcl 2017-03-09T18:14:05Z Bicyclidine quit (Ping timeout: 240 seconds) 2017-03-09T18:38:57Z schjetne quit (Ping timeout: 240 seconds) 2017-03-09T18:40:13Z zellerin joined #sbcl 2017-03-09T18:40:46Z slyrus quit (Ping timeout: 246 seconds) 2017-03-09T18:44:42Z congnotronicist quit (Quit: Leaving) 2017-03-09T18:57:21Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-09T19:00:38Z quadresce joined #sbcl 2017-03-09T19:11:36Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-09T19:11:53Z quadresce joined #sbcl 2017-03-09T19:14:30Z quadresce quit (Client Quit) 2017-03-09T19:15:03Z quadresce joined #sbcl 2017-03-09T19:30:44Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-09T19:34:02Z quadresce joined #sbcl 2017-03-09T20:00:51Z slyrus joined #sbcl 2017-03-09T20:02:07Z sjl joined #sbcl 2017-03-09T20:23:25Z stassats: an out of process debugger... but what about using a debugger from another thread? 2017-03-09T20:24:06Z stassats: that will avoid the problems with non-reentrant foreign functions 2017-03-09T20:24:12Z stassats: but not with unwinding... 2017-03-09T20:38:00Z schjetne joined #sbcl 2017-03-09T20:47:53Z Guest96445 joined #sbcl 2017-03-09T20:53:52Z Guest96445 quit (Remote host closed the connection) 2017-03-09T20:53:59Z angavrilov quit (Remote host closed the connection) 2017-03-09T20:57:31Z nyef`: Can you use ptrace() on a separate thread within the same process? 2017-03-09T20:58:01Z stassats: i don't think you can 2017-03-09T20:58:19Z igajsin quit (Ping timeout: 246 seconds) 2017-03-09T20:58:22Z nyef`: I can get you a proof-of-concept Linux x86-64 ptrace() implementation, but it's for a cross-process setup, not a cross-thread-in-same-process setup. 2017-03-09T20:59:25Z Shinmera quit (Quit: fuck everything) 2017-03-09T21:00:18Z stassats: with all this theoretical stuff about interrupts being unsafe, i can't say i've ever really been impacted by unsafe interrupts 2017-03-09T21:01:32Z stassats: and say we have really safe interrupts, what's the use beyond debugging? 2017-03-09T21:02:57Z attila_lendvai joined #sbcl 2017-03-09T21:02:57Z attila_lendvai quit (Changing host) 2017-03-09T21:02:57Z attila_lendvai joined #sbcl 2017-03-09T21:04:13Z nyef`: SIGCHLD ? 2017-03-09T21:04:42Z stassats: delegate a thread for handling sigchld? 2017-03-09T21:04:47Z nyef`: I realize that you're plausibly not all that interested in SIGWINCH, but it's there and has a purpose. 2017-03-09T21:04:50Z nyef`: And so on. 2017-03-09T21:16:30Z stassats: but whichever interrupt strategy unwinding foreign code is still a problem 2017-03-09T21:19:45Z foom: The only sane things to do in a signal handler have *always* been to either set a flag, or send a byte down a pipe. 2017-03-09T21:20:58Z stassats: that doesn't help interrupting foreign code 2017-03-09T21:23:18Z foom: hmm? 2017-03-09T21:23:46Z foom: Not sure what you mean by "interrupting foreign code"? Like, stopping it and doing something else? 2017-03-09T21:24:21Z foom: That's a crazy thing to do, regardless. :) 2017-03-09T21:25:07Z stassats: so the solution when foreign code misbehaves is to restart? now that's a quick way to become crazy 2017-03-09T21:25:51Z foom: "misbehaves" like what? 2017-03-09T21:26:26Z stassats: anything 2017-03-09T21:26:37Z foom: Yes. The only safe response is to die. 2017-03-09T21:26:58Z zellerin quit (Ping timeout: 240 seconds) 2017-03-09T21:27:03Z stassats: well, i'd been writing C if i had to restart and recompile each time 2017-03-09T21:28:44Z foom: I mean, having an unsafe debugging mode is fine. but...that's only because the person using it can understand that trying to continue after segfault or whatever is "best of luck, hope it works" 2017-03-09T21:28:53Z nyef`: Clearly, the answer is to write a C-in-Lisp environment, and use that to compile the entire Linux userland. d-: 2017-03-09T21:29:05Z stassats: or not call foreign code, ever 2017-03-09T21:29:15Z foom: Nothing should ever *depend* on that sort of functionality. 2017-03-09T21:29:22Z nyef`: Let's not get crazy, now. d-: 2017-03-09T21:29:54Z foom: I mean, it's not like asynchronous unwinds from most lisp code is safe either. 2017-03-09T21:31:03Z nyef`: Right, asynch unwind from *anything* is unsafe, unless the unwind semantics have been defined for *every* level of abstraction involved. 2017-03-09T21:31:28Z nyef`: In practice, this basically means using assembly language to write your program. 2017-03-09T21:31:59Z stassats: it's as safe as concurrent access, it may leave your data crooked but shouldn't crash and burn 2017-03-09T21:32:14Z foom: Not at all 2017-03-09T21:32:37Z nyef`: Concurrent access can compose. Asynch unwind doesn't. 2017-03-09T21:32:37Z foom: Take a mutex. Concurrent access is fine, because the other thread will wait for the mutex. 2017-03-09T21:32:58Z stassats: our mutexes are without-interrupts protected 2017-03-09T21:33:03Z foom: async unwind from the thing holding the mutex will typically either leave the mutex locked, or will unlock it in an inconsistent state. 2017-03-09T21:33:24Z foom: C code will do the former, SBCL mutexes typically the latter. 2017-03-09T21:33:29Z foom: *Both* are bad. 2017-03-09T21:33:37Z stassats: at least system mutexes 2017-03-09T21:34:21Z foom: Yes, the implementation of sbcl tries to make sure that it doesn't blow up the repl when you're doing crazy stuff while debugging. 2017-03-09T21:34:26Z stassats: but of course locked code is safe, i'm talking about concurrent access without locks involved 2017-03-09T21:35:06Z foom: I really don't debate that it's useful to be able to do crazy stuff in a debugging environment. 2017-03-09T21:35:21Z foom: Just so long as that's the ONLY time when that sort of stuff happens. 2017-03-09T21:35:42Z stassats: the point is to make it non crazy 2017-03-09T21:35:55Z foom: That's impossible. 2017-03-09T21:36:47Z stassats: sbcl is currently not even approaching minimum non-craziness 2017-03-09T21:38:13Z foom: Just consider how you would write code which can safely allow an unwind between any two instructions. It's simply not feasible. 2017-03-09T21:38:35Z stassats: for example, having a better system will allow on interrupt if, say, there's a foreign call right now to choose from "interrupt on exit" "interrupt anyway" "continue" 2017-03-09T21:40:23Z foom: Running code within a signal handler invoked during a lisp function is not really better than running it within a signal handler invoked during a foreign function. 2017-03-09T21:40:50Z stassats: you can always run code outside of the signal handler 2017-03-09T21:42:16Z stassats: or even from separate process 2017-03-09T21:42:48Z stassats: since you don't really need to run code, but to observe state and possible unwind 2017-03-09T21:48:26Z stassats: though presently i can decide whether to unwind, the problem with foreign code is that it's not reentrant so debugging from it is really limited 2017-03-09T21:50:39Z nyef`: The thing about "unwind from arbitrary instruction boundaries" isn't so much a matter of "make it not crash", it's "it's basically impossible to specify that such-and-such cleanup needs to occur from *this* point in returning function A to *this* point in its caller B" and similar cases. 2017-03-09T21:51:03Z stassats: but even exiting can result in corrupt data, on disk 2017-03-09T21:52:21Z stassats: my real biggest concern is reentrancy of C code 2017-03-09T21:52:41Z nyef`: If you *can* specify such cleanup actions, then you can do asynch unwind safely. But that's not typically possible outside of assembly language. 2017-03-09T21:52:46Z stassats: only out of process debugging can help there 2017-03-09T21:53:03Z nyef`: Yeah, basically. 2017-03-09T21:53:18Z nyef`: So, you want that prototype x86-64 linux ptrace thing? (-: 2017-03-09T21:53:41Z stassats: is it already public? 2017-03-09T21:54:59Z nyef`: Umm... Not that I see immediately. 2017-03-09T21:55:21Z nyef`: And I don't really have time to package it up right now, I'm just getting ready to head out the door. 2017-03-09T21:55:39Z stassats: yeah, if it's not already public then don't bother 2017-03-09T21:55:50Z nyef`: But I can find the bits and put something together either late tonight or tomorrow. 2017-03-09T21:56:05Z stassats: not like i plan on really working on this 2017-03-09T21:56:46Z gingerale quit (Remote host closed the connection) 2017-03-09T21:57:17Z nyef`: I should probably package it up and push it to my webspace anyway. 2017-03-09T22:03:22Z shka_ quit (Ping timeout: 256 seconds) 2017-03-09T22:13:29Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-09T22:29:17Z oleo quit (Remote host closed the connection) 2017-03-09T22:53:02Z quadresce joined #sbcl 2017-03-09T22:53:20Z quadresce quit (Client Quit) 2017-03-09T22:54:21Z oleo joined #sbcl 2017-03-09T23:11:16Z Bicyclidine joined #sbcl 2017-03-09T23:23:34Z slyrus quit (Ping timeout: 246 seconds) 2017-03-09T23:54:59Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-10T00:24:38Z neuronsong joined #sbcl 2017-03-10T00:29:34Z cromachina joined #sbcl 2017-03-10T00:35:11Z Bicyclidine quit (Ping timeout: 258 seconds) 2017-03-10T00:48:58Z schjetne quit (Ping timeout: 264 seconds) 2017-03-10T01:35:19Z jamtho joined #sbcl 2017-03-10T01:44:15Z karswell joined #sbcl 2017-03-10T01:44:46Z Bicyclidine joined #sbcl 2017-03-10T02:23:18Z jamtho quit (Ping timeout: 268 seconds) 2017-03-10T02:51:36Z Bicyclidine quit (Ping timeout: 260 seconds) 2017-03-10T03:54:13Z Bicyclidine joined #sbcl 2017-03-10T04:02:04Z stassats quit (Ping timeout: 260 seconds) 2017-03-10T04:30:30Z akkad quit (Excess Flood) 2017-03-10T04:32:49Z akkad joined #sbcl 2017-03-10T04:33:49Z quadresce joined #sbcl 2017-03-10T04:43:23Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-10T04:53:57Z akkad quit (Excess Flood) 2017-03-10T04:58:07Z quadresce joined #sbcl 2017-03-10T05:03:19Z akkad joined #sbcl 2017-03-10T05:09:10Z jack_rabbit joined #sbcl 2017-03-10T05:14:04Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-10T05:18:42Z igajsin joined #sbcl 2017-03-10T06:07:24Z shka_ joined #sbcl 2017-03-10T06:28:16Z shka_ quit (Ping timeout: 256 seconds) 2017-03-10T06:28:29Z oleo quit (Quit: Leaving) 2017-03-10T06:44:25Z akkad quit (Excess Flood) 2017-03-10T06:52:49Z akkad joined #sbcl 2017-03-10T07:00:48Z Bicyclidine quit (Ping timeout: 260 seconds) 2017-03-10T07:20:26Z karswell quit (Remote host closed the connection) 2017-03-10T07:20:54Z karswell joined #sbcl 2017-03-10T07:28:36Z angavrilov joined #sbcl 2017-03-10T07:57:58Z clintm joined #sbcl 2017-03-10T08:10:20Z clintm quit (Ping timeout: 260 seconds) 2017-03-10T08:40:30Z gargaml joined #sbcl 2017-03-10T08:41:12Z gingerale joined #sbcl 2017-03-10T08:48:11Z clintm joined #sbcl 2017-03-10T09:08:35Z salv0 joined #sbcl 2017-03-10T10:07:14Z jamtho joined #sbcl 2017-03-10T10:20:01Z engblom joined #sbcl 2017-03-10T10:21:29Z engblom: The aarch64 version downloadable on your site seem to not work any more (at least not in current voidlinux). I want to compile from source, but I have difficulties to find any implementation able to bootstrap sbcl 2017-03-10T10:21:50Z engblom: So far I have tried both abcl and ecl. Neither will compile sbcl 2017-03-10T10:22:26Z flip214: engblom: for arm (32bit) I was using an NFS export and a amd64 machine for cross-compiling. 2017-03-10T10:22:29Z flip214: that did work fine. 2017-03-10T10:24:07Z engblom: So, if I use sshfs/nfs/smb/whatever, how do I crosscompile it? 2017-03-10T10:25:53Z flip214: make-config.sh on the target machine 2017-03-10T10:26:27Z flip214: engblom: look into make.sh. 2017-03-10T10:26:36Z flip214: that has an extensive comment blob starting around line 20 2017-03-10T10:26:40Z flip214: explaining what to do. 2017-03-10T10:26:56Z engblom: Okay! Thanks 2017-03-10T10:27:06Z flip214: I'd suggest some NFS/smb, that's much easier than rsync etc. 2017-03-10T10:27:44Z flip214: engblom: how about using the arm64 package from https://packages.debian.org/sid/sbcl? 2017-03-10T10:28:01Z flip214: what's the difference to aarch64? 2017-03-10T10:28:16Z engblom: It is the same. 2017-03-10T10:28:32Z flip214: well, so how about just using that binary? 2017-03-10T10:28:41Z engblom: That could be an option too 2017-03-10T10:28:47Z flip214: even if just for bootstrapping, if you want/need a newer git head version 2017-03-10T10:29:05Z flip214: cross-compiling is faster, in my experience. 2017-03-10T10:50:32Z engblom: The debian version got the same problem as the one from sbcl.org. Both complain about mmap: 2017-03-10T10:50:38Z engblom: mmap: Invalid argument 2017-03-10T10:50:54Z engblom: They both drop off at ldb 2017-03-10T11:10:10Z engblom: Hmm, the same is also happening when crosscompiling 2017-03-10T11:11:22Z engblom: http://lpaste.net/4211952716403965952 2017-03-10T11:11:56Z jamtho_ joined #sbcl 2017-03-10T11:14:15Z jamtho quit (Ping timeout: 268 seconds) 2017-03-10T11:19:29Z m00natic joined #sbcl 2017-03-10T11:25:04Z DGASAU quit (Read error: Connection reset by peer) 2017-03-10T11:25:53Z DGASAU joined #sbcl 2017-03-10T11:39:47Z flip214: engblom: which kernel version are you running? 2017-03-10T11:40:14Z flip214: try to reduce the dynamic space size in the make-target-2.sh script 2017-03-10T11:59:21Z attila_lendvai joined #sbcl 2017-03-10T11:59:21Z attila_lendvai quit (Changing host) 2017-03-10T11:59:21Z attila_lendvai joined #sbcl 2017-03-10T12:08:12Z stassats joined #sbcl 2017-03-10T12:08:51Z stassats: engblom: if the binary succumbs to mmap then your build will as well 2017-03-10T12:11:22Z stassats: but i've never seen that failure 2017-03-10T12:11:27Z stassats: anywhere 2017-03-10T12:11:40Z stassats: engblom: what's your ulimit? 2017-03-10T12:11:49Z stassats: is overcommit allowed? 2017-03-10T12:12:13Z flip214: newer kernel versions changed the way the data segment size is calculated w.r.t. ulimit 2017-03-10T12:12:22Z flip214: that's why I asked about the kernel version 2017-03-10T12:12:43Z flip214: and I've seen similar problems a few years ago when using (trying to use) sbcl in a vserver 2017-03-10T12:12:57Z stassats: not from os_map 2017-03-10T12:15:30Z stassats: huh, it uses MAP_FIXED 2017-03-10T12:16:02Z stassats: which is a) wrong b) shouldn't fail because it steps on some other mapping 2017-03-10T12:22:02Z stassats: it fails to map at the right place on darwin without map_fixed 2017-03-10T12:22:05Z stassats: huh-huh 2017-03-10T12:23:03Z stassats: and linux, i'm missing something, clearly 2017-03-10T12:25:11Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-10T12:27:43Z flip214: stassats: without MAP_FIXED the kernel will choose the address... why should it be the "right" expected location without that flag? 2017-03-10T12:28:00Z stassats: MAP_FIXED is almost never right 2017-03-10T12:29:38Z stassats: the only right usecase if you already control the mapping 2017-03-10T12:29:43Z stassats: maybe that's the case here 2017-03-10T12:32:59Z stassats: yeah, i think we map the whole dynamic space first 2017-03-10T12:34:03Z stassats: ok then, why would something fail to remap a memory region? 2017-03-10T12:34:46Z stassats: EINVAL We don't like addr, length, or offset (e.g., they are too large, or not aligned on a page boundary). 2017-03-10T12:34:56Z stassats: alignment is the only plausible one 2017-03-10T12:35:12Z flip214: because ulimit, because overcommit, because not enough RAM+swap, .. ? 2017-03-10T12:35:19Z stassats: flip214: no 2017-03-10T12:35:25Z stassats: it's already there, nothing of that matters 2017-03-10T12:35:28Z flip214: stassats: well, 0x10000000 should be nicely aligned, too 2017-03-10T12:36:14Z stassats: engblom: can you modify os_map in linux-os.c to lose("unexpected mmap(%d, %d) failure\n", addr, len); 2017-03-10T12:36:27Z stassats: flip214: is it 0x10000000? 2017-03-10T12:37:01Z dougk_ quit (Ping timeout: 246 seconds) 2017-03-10T12:38:03Z stassats: the most plausible explanation, it's using to small a page size 2017-03-10T12:41:29Z stassats: my device has 4096 and (setf *backend-page-bytes* 4096) 2017-03-10T12:41:45Z stassats: engblom: scratch the patch, what's your getconf PAGESIZE 2017-03-10T12:44:48Z stassats: in any case, it can be 64K, i'll have to change it to that 2017-03-10T12:45:15Z flip214: on my 32bit RPi it starts at 0x4000000 2017-03-10T12:45:29Z stassats: flip214: that's even less relevant 2017-03-10T12:46:02Z flip214: why? the above is about ARM, too. 2017-03-10T12:46:25Z stassats: arm64 2017-03-10T12:49:25Z dougk_ joined #sbcl 2017-03-10T12:56:37Z jamtho_ quit (Ping timeout: 268 seconds) 2017-03-10T13:01:07Z engblom: flip214: Kernel: 4.10.1 2017-03-10T13:01:19Z stassats: engblom: quick, getconf PAGESIZE 2017-03-10T13:01:26Z engblom: ulimit is unlimited 2017-03-10T13:03:34Z engblom: 65536 2017-03-10T13:03:41Z stassats: yes! 2017-03-10T13:03:53Z stassats: that's incompatible, i'm readying a patch 2017-03-10T13:04:17Z stassats: and i can give you a binary with the new value 2017-03-10T13:05:35Z engblom: Okay! Nice 2017-03-10T13:07:44Z flip214: engblom: ah, that one will already have the ulimit accounting changes in... _what_ is unlimited in "ulimit"? would you please paste "ulimit -aH" and "ulimit -aS"? 2017-03-10T13:09:27Z engblom: flip214: http://lpaste.net/353389 2017-03-10T13:09:55Z stassats: engblom: http://common-lisp.net/~sboukarev/sbcl-arm64-binary.tar.bz2 2017-03-10T13:10:03Z stassats: no backdoors, i promise 2017-03-10T13:11:26Z engblom: stassats: It is working! 2017-03-10T13:11:49Z stassats: that was easy 2017-03-10T13:13:02Z flip214: engblom: thanks a lot, that's a clean ulimit ;) 2017-03-10T13:16:53Z engblom: When compiling, can I use run-sbcl.sh or do I need to install the binary first? 2017-03-10T13:16:59Z stassats: engblom: you can 2017-03-10T13:17:09Z engblom: Okay, thanks! 2017-03-10T13:29:47Z milanj joined #sbcl 2017-03-10T13:31:30Z engblom is building the git-version of sbcl. 2017-03-10T13:43:26Z stassats: engblom: the binary is the git HEAD, the version number is just bogus 2017-03-10T13:45:23Z oleo joined #sbcl 2017-03-10T13:47:14Z gargaml quit (Quit: WeeChat 1.7) 2017-03-10T13:47:46Z engblom: Ah, okay. I saw the version number and got tricked by it. 2017-03-10T13:49:09Z stassats: i'll make sure to upload a new binary come release time, since this is an important fix 2017-03-10T13:53:09Z cromachina quit (Read error: Connection reset by peer) 2017-03-10T14:10:52Z oleo quit (Quit: Verlassend) 2017-03-10T14:18:40Z eudoxia joined #sbcl 2017-03-10T14:53:05Z paroneayea quit (Read error: Connection reset by peer) 2017-03-10T14:53:33Z paroneayea joined #sbcl 2017-03-10T15:03:01Z DGASAU quit (Read error: Connection reset by peer) 2017-03-10T15:07:43Z DGASAU joined #sbcl 2017-03-10T15:17:59Z eudoxia quit (Quit: Leaving) 2017-03-10T15:22:52Z schjetne joined #sbcl 2017-03-10T15:28:19Z shka_ joined #sbcl 2017-03-10T15:30:58Z foom quit (Ping timeout: 246 seconds) 2017-03-10T15:32:34Z DGASAU quit (Read error: Connection reset by peer) 2017-03-10T15:42:05Z Bicyclidine joined #sbcl 2017-03-10T15:43:08Z foom joined #sbcl 2017-03-10T15:56:39Z DGASAU joined #sbcl 2017-03-10T16:10:04Z DGASAU quit (Read error: Connection reset by peer) 2017-03-10T16:20:18Z DGASAU joined #sbcl 2017-03-10T16:40:52Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-10T16:46:09Z nyef`: So, I thought of an angle for the ephemeral gen-0 evacuation thing from last night, but it requires a precise collector. 2017-03-10T16:56:47Z milanj quit (Quit: This computer has gone to sleep) 2017-03-10T17:08:50Z schjetne quit (Ping timeout: 268 seconds) 2017-03-10T17:11:16Z schjetne joined #sbcl 2017-03-10T17:16:11Z quadresce joined #sbcl 2017-03-10T17:18:27Z sjl quit (Ping timeout: 240 seconds) 2017-03-10T17:31:17Z quadresce` joined #sbcl 2017-03-10T17:31:44Z quadresce quit (Ping timeout: 260 seconds) 2017-03-10T17:36:52Z quadresce` quit (Quit: This computer has gone to sleep) 2017-03-10T18:01:28Z milanj joined #sbcl 2017-03-10T18:01:42Z m00natic quit (Remote host closed the connection) 2017-03-10T18:08:02Z schjetne quit (Ping timeout: 268 seconds) 2017-03-10T18:28:27Z schjetne joined #sbcl 2017-03-10T18:33:27Z Bicyclidine quit (Remote host closed the connection) 2017-03-10T19:03:41Z nyef`: stassats: Turns out that that ptrace stuff that I mentioned was already in my webspace, I just didn't notice until I went to package it up again: http://www.lisphacker.com/temp/unfinished-ptrace-stuff.tgz 2017-03-10T19:04:05Z stassats: k 2017-03-10T19:46:08Z schjetne quit (Ping timeout: 260 seconds) 2017-03-10T20:27:11Z Bicyclidine joined #sbcl 2017-03-10T20:34:00Z dougk_: whenever we set a package name / nickname, would it be reasonable to compute the "best" nickname for each package, where "best" is the shortest in characters? and then use the best nickname when printing symbols? 2017-03-10T20:36:44Z |3b|: and if you do, do you include local nicknames? 2017-03-10T20:37:08Z nyef`: Isn't the requirement to use the official package-name, not any of the nicknames? 2017-03-10T20:38:40Z dougk_: that's what i'm asking 2017-03-10T20:39:27Z |3b|: clhs 22.1.3.3.1 2017-03-10T20:39:28Z specbot: Package Prefixes for Symbols: http://www.lispworks.com/reference/HyperSpec/Body/22_acca.htm 2017-03-10T20:39:42Z |3b| isn't sure if that says or not 2017-03-10T20:39:43Z quadresce` joined #sbcl 2017-03-10T20:40:08Z |3b|: top just says 'a package prefic', but summary says home package 2017-03-10T20:40:08Z dougk_: "package name" is not the same as package-name. 2017-03-10T20:40:22Z stassats: i'd be for an option to do that anyway 2017-03-10T20:40:37Z dougk_: "package prefix" is a "package name", and a "package" "name" is an "identifier" of a "package", not PACKAGE-NAME per se 2017-03-10T20:40:38Z |3b|: yeah, sounds like useful option 2017-03-10T20:41:02Z |3b|: dougk_: yeah, the top doesn't seem to require it, but summary sounds like it does 2017-03-10T20:41:04Z stassats: because some packages have some zany names 2017-03-10T20:41:16Z |3b|: "foo:bar is printed when symbol bar is external in its home package foo and is not accessible in the current package. " 2017-03-10T20:42:02Z dougk_: but 'foo' could be a designator for a package named 'really-foo'. again it doesn't talk about how you derived foo: 2017-03-10T20:42:32Z nyef`: Name, sense 3 says package-name. 2017-03-10T20:43:17Z nyef`: clhs 11.1.1.1 2017-03-10T20:43:17Z specbot: Package Names and Nicknames: http://www.lispworks.com/reference/HyperSpec/Body/11_aaa.htm 2017-03-10T20:44:56Z dougk_: well, nonstandard or not, i think I'll add *print-package-nickname* which allows use of the shortest nickname if any for any given package 2017-03-10T20:46:35Z dougk_: also people here seem to want LOOP to not print package names on any loop keyword. So (loop repeat 10 for x ... collect (frob)) regardless of current package should have no qualifiers on 'repeat','for','collect' (unless *print-readably*, even if *print-escape* = t) 2017-03-10T20:46:40Z nyef`: Remember to include it in WITH-STANDARD-IO-SYNTAX or whatever it is and friends. (-: 2017-03-10T20:48:56Z stassats: though options like these tend to be used only by a handful of people who know about them 2017-03-10T21:02:02Z angavrilov quit (Remote host closed the connection) 2017-03-10T21:03:46Z shka_ quit (Ping timeout: 256 seconds) 2017-03-10T21:03:50Z quadresce` quit (Quit: This computer has gone to sleep) 2017-03-10T21:23:11Z attila_lendvai joined #sbcl 2017-03-10T21:23:11Z attila_lendvai quit (Changing host) 2017-03-10T21:23:11Z attila_lendvai joined #sbcl 2017-03-10T21:41:36Z neuronsong quit (Read error: Connection reset by peer) 2017-03-10T21:45:38Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-10T21:50:27Z |3b|: is there any reason why coercing a vector to specialized simple-array requires dimensions, but unspecialized doesn't? 2017-03-10T21:50:55Z stassats: you mean rank? 2017-03-10T21:51:00Z |3b|: yeah 2017-03-10T21:52:01Z stassats: only sequences are actually coercible 2017-03-10T21:52:28Z |3b|: (coerce #() '(simple-array (unsigned-byte 8))) -> error, (coerce #() '(simple-array t)) -> #() 2017-03-10T21:52:53Z stassats: neither is standard 2017-03-10T21:52:59Z prxq joined #sbcl 2017-03-10T21:53:31Z |3b|: so just luck either works at all? 2017-03-10T21:53:34Z stassats: and the latter is because no coercion takes place 2017-03-10T21:53:43Z |3b|: ah 2017-03-10T21:55:52Z stassats: it should have been specified for any arrays 2017-03-10T21:56:12Z stassats: together with COPY-ARRAY 2017-03-10T21:56:40Z aeth joined #sbcl 2017-03-10T21:56:47Z stassats: that's what i'd do if i were making a new standard 2017-03-10T21:57:02Z stassats: and call it a success and not change anything else 2017-03-10T21:57:08Z aeth: change what? 2017-03-10T21:57:25Z stassats: coerce COERCE into being more sensible 2017-03-10T21:57:45Z aeth: There are many more flaws in the language than that. 2017-03-10T21:58:18Z stassats: says who? 2017-03-10T21:58:41Z aeth: Look at where popular libraries exist. pathnames, unicode, threads, etc. 2017-03-10T22:00:07Z stassats: disregarding pathnames and threads as being too os specific and unicode as changing too frequently 2017-03-10T22:00:14Z aeth: The environments where Lisp run are a lot less diverse in 2017 than in 1987. More assumptions can be made, including perhaps an assumption of utf-8. 2017-03-10T22:00:34Z aeth: It's not that the standard made wrong decisions there, it's that the winners were unclear in the early 1990s. 2017-03-10T22:03:03Z aeth: Another thing where the standard is outdated is that the minimum size assumptions for many things, including fixnums ((signed-byte 16)) and array sizes (1024), are far too small... 2017-03-10T22:03:29Z aeth: ...except for short-float, which is too precise by two bits (13 vs 11) to use IEEE or ARM half-precision floating point, which would be the only useful modern meaning of "short-float" 2017-03-10T22:03:31Z stassats: would 1025 do 2017-03-10T22:03:32Z stassats: ? 2017-03-10T22:03:56Z aeth: Strings are character arrays. There are probably some programs with docstrings longer than 1024 and thus not complying with the standard. 2017-03-10T22:04:21Z stassats: or are you leaning towards 1026? 2017-03-10T22:05:13Z aeth: Idk 4611686018427387901 would make a nice array-total-size-limit ;) 2017-03-10T22:05:29Z stassats: i've never heard anyone complaining about array-total-size-limit 2017-03-10T22:05:44Z aeth: It's just one of dozens of my stock answers to flaws in the standard. 2017-03-10T22:05:55Z prxq: aeth: I don't think many would protest if a CL would use half-precision floats for short-floats. 2017-03-10T22:06:21Z aeth: prxq: it would however violate the standard 2017-03-10T22:06:58Z stassats: aeth: you're also misreading the standard 2017-03-10T22:07:00Z stassats: it wouldn't 2017-03-10T22:07:02Z |3b| has objected to the array-total-size-limit on 32-bit clisp 2017-03-10T22:07:12Z prxq: even if it would, why would anyone care in this particular instance? 2017-03-10T22:07:18Z |3b|: possibly even on 32bit sbcl before teh fixnums got widened 2017-03-10T22:07:34Z stassats: 32-bit didn't get its fixnums widened, though 2017-03-10T22:07:39Z |3b|: but 2^24 is not nearly enough 2017-03-10T22:07:47Z |3b| thought it got 1 bit more? 2017-03-10T22:07:57Z stassats: nope 2017-03-10T22:08:21Z |3b|: is it 30bit? 2017-03-10T22:08:22Z aeth: stassats: I am misreading it somewhat. It says "Figure 12-12. Recommended Minimum Floating-Point Precision and Exponent Size", so short-float is not bindingly flawed, just flawed 2017-03-10T22:09:12Z aeth: It actually would be nice to be more precise in the specification of floating point, and handle edge cases that implementations non-portably handle. 2017-03-10T22:09:20Z stassats: aeth: nothing flowed about _recommended_ sizes 2017-03-10T22:09:37Z stassats: |3b|: yes 2017-03-10T22:10:50Z |3b|: yeah, might have hit that at some point 2017-03-10T22:11:13Z |3b|: 2^30 length bitvectors fit in ram pretty easily even on 32bit :) 2017-03-10T22:11:29Z aeth: stassats: merely recommended sizes are actually worse than too-low minimum sizes because then you cannot know *anything* about a given piece of portable CL code, and just have to rely on the popular implementations like SBCL being sane. 2017-03-10T22:12:03Z stassats: clhs *features* 2017-03-10T22:12:03Z specbot: http://www.lispworks.com/reference/HyperSpec/Body/v_featur.htm 2017-03-10T22:12:07Z aeth: Imo, recommended float sizes should be binding if the type is distinct, e.g. if long-float exists rather than is just double-float. 2017-03-10T22:12:14Z stassats: #+ieee-floating-point 2017-03-10T22:12:17Z stassats: there, solved it for you 2017-03-10T22:12:42Z aeth: stassats: wonderful, but insufficient because it doesn't handle +inf, -inf, nan, etc. 2017-03-10T22:12:58Z aeth: stassats: and it doesn't say which IEEE floats map to which CL floats. 2017-03-10T22:13:22Z stassats: well, duh, don't use CL for floats then 2017-03-10T22:13:28Z stassats: since it's that horrible 2017-03-10T22:16:41Z aeth: Generally the assumption can be made that where they exist and where that feature is set single-float = IEEE single precision and double-float = IEEE double precision, on modern 64-bit CLs. 2017-03-10T22:18:16Z aeth: But feature :ieee-single-precision-float-is-single-float and :ieee-double-precision-float-is-double-float etc. or something could help there. 2017-03-10T22:18:25Z stassats: that'd be silly 2017-03-10T22:19:14Z aeth: It would mostly be useful for short-float and long-float, especially long-float, where it could be IEEE quad float (128 bit) or x86 extended precision (80 bit) or something else. 2017-03-10T22:19:25Z aeth: If, of course, there's even a separate long-float. 2017-03-10T22:19:40Z aeth: The names, of course, are silly. And it could even be done in one 2017-03-10T22:20:08Z stassats: how stupid would the implementers be to support ieee-floats and misname them? 2017-03-10T22:21:16Z aeth: It's technically possible afaik. 2017-03-10T22:22:47Z stassats: yes, unreason is not technically prohibited 2017-03-10T22:22:59Z aeth: Disambiguation would be mostly useful for short-float and long-float, though, because they don't (roughly) map to IEEE names and could be used for other purposes. 2017-03-10T22:23:07Z aeth: Where an implementation uses them. 2017-03-10T22:25:25Z aeth: It'd probably actually just be easier to introduce a separate quad-float or quadruple-float if IEEE quadruple-precision floating point is ever common in the future because a lot of existing long-float code is probably unportably assuming some implementation's long-float, which is probably not quad-float. 2017-03-10T22:26:06Z stassats: ok, you've convinced me 2017-03-10T22:26:12Z stassats: for my revision i'd remove short and long floats 2017-03-10T22:27:49Z aeth: One possible revision would be to make short-float and long-float implementation-specific and introduce half-float and quad-float 2017-03-10T22:27:57Z stassats: and FLOAT is already not exhaustively partitioned 2017-03-10T22:28:22Z aeth: Oh, one mess that can come from this is people who assume pi is double-float because they only code in SBCL. 2017-03-10T22:28:41Z aeth: Then they wonder why their number is getting coerced to a different float and then breaking code somewhere else that requires the double-float type 2017-03-10T22:28:52Z rumbler31 joined #sbcl 2017-03-10T22:28:54Z aeth: because pi is long-float! 2017-03-10T22:30:00Z stassats: clearly, that's why Swift and Rust are overtaking CL in popularity 2017-03-10T22:31:13Z aeth: Afaik, those languages are mainly popular because of people who hate GC. 2017-03-10T22:31:54Z aeth: Every shiny new language has GC, and C and C++ are neither shiny nor new. Ergo, those languages. 2017-03-10T22:32:34Z aeth: Clearly, CL needs an alternative way of operating without the use of GC added to its standard. ;) 2017-03-10T22:32:47Z stassats: there's no GC in the standard 2017-03-10T22:33:42Z stassats: so, you can't allocate, depend on single-float being single floats and have larger than 1023 arrays 2017-03-10T22:33:44Z aeth: How would you practically program in CL without GC? Make sure every foo is (declare (dynamic-extent foo)) and hope the compiler warns when that can't be done? 2017-03-10T22:34:24Z stassats: aeth: run it on a turing machine 2017-03-10T22:34:37Z aeth: infinite memory? 2017-03-10T22:53:22Z quadresce joined #sbcl 2017-03-10T22:58:45Z aeth: I wonder what would be the best way to portably add fast vector and matrix operations into CL. I was talking in #lisp about this yesterday. 2017-03-10T23:01:32Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-10T23:11:40Z rumbler31 quit (Remote host closed the connection) 2017-03-10T23:21:03Z prxq quit (Remote host closed the connection) 2017-03-10T23:30:02Z gingerale quit (Remote host closed the connection) 2017-03-10T23:36:34Z aeth: oooh, I just found a big flaw in the CL spec. (format t "~36r" foo) is allowed (going from 0 to 9 then from A to Z) but (format t "~62r" foo) isn't (going from 0 to 9 then from A to Z then from a to z) 2017-03-10T23:37:11Z stassats: what? 2017-03-10T23:37:15Z aeth: alternatively, *print-base* is limited to 36. alternatively, #36r1 is the highest allowed radix prefix 2017-03-10T23:38:35Z aeth: Or, alternatively, because the numerical representation is case-insensitive (e.g. #x1f = #x1F) the CL built-ins cannot be used to encode common-on-the-Internet case-insensitive reprsentations of higher bases than 32. 2017-03-10T23:39:30Z aeth: It's actually not a simple problem to solve because #x1f = #x1F is an assumption that a lot of people would make. 2017-03-10T23:39:45Z stassats: i can't tell if you're being serious 2017-03-10T23:40:42Z aeth: You don't think bases higher than 36 are worth representing? 2017-03-10T23:41:18Z aeth: I guess it would be too problematic, though, because there isn't one standard way of doing it. 2017-03-10T23:41:20Z Quadrescence quit (Ping timeout: 260 seconds) 2017-03-10T23:43:36Z aeth: It would need some hack like case-insensitivity from 2 to 36 and case-sensitivity from 37 to 62 2017-03-10T23:58:13Z |3b| quit (Remote host closed the connection) 2017-03-11T00:01:44Z edgar-rft is now known as again 2017-03-11T00:03:03Z again is now known as edgar-rft 2017-03-11T00:27:59Z |3b| joined #sbcl 2017-03-11T01:04:08Z milanj quit (Quit: Leaving) 2017-03-11T01:26:04Z cromachina joined #sbcl 2017-03-11T01:32:21Z rumbler31 joined #sbcl 2017-03-11T01:48:48Z stassats quit (Ping timeout: 256 seconds) 2017-03-11T02:17:44Z rumbler31 quit (Remote host closed the connection) 2017-03-11T02:38:18Z rumbler31 joined #sbcl 2017-03-11T02:40:15Z karswell quit (Remote host closed the connection) 2017-03-11T02:40:27Z karswell joined #sbcl 2017-03-11T02:47:48Z slyrus joined #sbcl 2017-03-11T03:01:46Z edgar-rft quit (Quit: edgar-rft) 2017-03-11T03:01:57Z slyrus quit (Ping timeout: 240 seconds) 2017-03-11T03:05:18Z adlai quit (Ping timeout: 256 seconds) 2017-03-11T03:10:21Z adlai joined #sbcl 2017-03-11T03:12:02Z adlai quit (Read error: Connection reset by peer) 2017-03-11T03:12:26Z liead joined #sbcl 2017-03-11T03:36:39Z rumbler31 quit (Remote host closed the connection) 2017-03-11T03:37:16Z jack_rabbit joined #sbcl 2017-03-11T06:08:28Z Bicyclidine quit (Ping timeout: 240 seconds) 2017-03-11T07:18:24Z edgar-rft joined #sbcl 2017-03-11T07:28:04Z Shinmera joined #sbcl 2017-03-11T07:43:33Z schjetne joined #sbcl 2017-03-11T07:52:44Z schjetne_ joined #sbcl 2017-03-11T07:53:06Z schjetne quit (Read error: Connection reset by peer) 2017-03-11T07:56:05Z schjetne joined #sbcl 2017-03-11T07:56:19Z schjetne_ quit (Read error: Connection reset by peer) 2017-03-11T08:02:14Z schjetne quit (Ping timeout: 256 seconds) 2017-03-11T08:13:00Z schjetne joined #sbcl 2017-03-11T08:17:32Z schjetne quit (Read error: Connection reset by peer) 2017-03-11T08:17:47Z angavrilov joined #sbcl 2017-03-11T08:17:59Z schjetne joined #sbcl 2017-03-11T08:37:27Z rumbler31 joined #sbcl 2017-03-11T08:41:51Z rumbler31 quit (Ping timeout: 268 seconds) 2017-03-11T08:52:45Z schjetne quit (Read error: Connection reset by peer) 2017-03-11T08:52:46Z schjetne_ joined #sbcl 2017-03-11T09:15:27Z edgar-rft quit (Quit: edgar-rft) 2017-03-11T09:25:55Z gingerale joined #sbcl 2017-03-11T09:38:50Z IRCFrEAK joined #sbcl 2017-03-11T09:39:20Z IRCFrEAK quit (K-Lined) 2017-03-11T09:42:14Z schjetne_ quit (Read error: Connection reset by peer) 2017-03-11T09:42:53Z schjetne joined #sbcl 2017-03-11T09:48:37Z gingerale quit (Read error: Connection reset by peer) 2017-03-11T09:48:44Z schjetne_ joined #sbcl 2017-03-11T09:48:49Z schjetne quit (Read error: Connection reset by peer) 2017-03-11T09:52:16Z |3b| quit (Remote host closed the connection) 2017-03-11T09:52:30Z shka_ joined #sbcl 2017-03-11T10:00:06Z schjetne_ quit (Ping timeout: 256 seconds) 2017-03-11T10:49:32Z myrkraverk quit (Remote host closed the connection) 2017-03-11T11:21:50Z myrkraverk joined #sbcl 2017-03-11T12:28:16Z sunnny joined #sbcl 2017-03-11T12:37:57Z rumbler31 joined #sbcl 2017-03-11T12:38:44Z edgar-rft joined #sbcl 2017-03-11T12:42:10Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-11T12:45:13Z salv0 quit (Remote host closed the connection) 2017-03-11T12:53:56Z sunnny left #sbcl 2017-03-11T13:26:52Z Bicyclidine joined #sbcl 2017-03-11T13:41:12Z Bicyclidine quit (Ping timeout: 258 seconds) 2017-03-11T13:47:54Z DGASAU quit (Ping timeout: 256 seconds) 2017-03-11T13:49:24Z DGASAU joined #sbcl 2017-03-11T14:37:56Z m00natic joined #sbcl 2017-03-11T14:45:36Z m00natic quit (Remote host closed the connection) 2017-03-11T14:49:14Z beatdown quit (Quit: Error 666: Connection terminated by the underworld.) 2017-03-11T14:50:34Z beatdown joined #sbcl 2017-03-11T14:57:12Z |3b| joined #sbcl 2017-03-11T15:00:12Z gingerale joined #sbcl 2017-03-11T15:02:18Z beatdown quit (Quit: Error 666: Connection terminated by the underworld.) 2017-03-11T15:38:03Z rumbler3_ joined #sbcl 2017-03-11T15:42:05Z rumbler3_ quit (Ping timeout: 240 seconds) 2017-03-11T16:19:25Z DGASAU quit (Read error: Connection reset by peer) 2017-03-11T16:19:51Z DGASAU joined #sbcl 2017-03-11T16:53:12Z DGASAU quit (Ping timeout: 256 seconds) 2017-03-11T17:02:37Z DGASAU joined #sbcl 2017-03-11T18:19:04Z eschatologist quit (Ping timeout: 268 seconds) 2017-03-11T18:20:20Z igajsin quit (Quit: Leaving.) 2017-03-11T18:45:49Z sjl joined #sbcl 2017-03-11T19:15:12Z schjetne joined #sbcl 2017-03-11T19:20:27Z DGASAU quit (Ping timeout: 258 seconds) 2017-03-11T19:21:22Z schjetne quit (Ping timeout: 264 seconds) 2017-03-11T19:26:36Z schjetne joined #sbcl 2017-03-11T19:31:19Z angavrilov quit (Remote host closed the connection) 2017-03-11T19:55:55Z sjl quit (Ping timeout: 240 seconds) 2017-03-11T20:06:52Z DGASAU joined #sbcl 2017-03-11T20:11:29Z |3b| quit (Remote host closed the connection) 2017-03-11T20:13:25Z |3b| joined #sbcl 2017-03-11T21:04:39Z sjl joined #sbcl 2017-03-11T21:07:55Z shka_ quit (Ping timeout: 240 seconds) 2017-03-11T21:46:48Z whiteline quit (Remote host closed the connection) 2017-03-11T21:47:39Z whiteline joined #sbcl 2017-03-11T21:54:59Z |3b| quit (Remote host closed the connection) 2017-03-11T21:59:00Z |3b| joined #sbcl 2017-03-11T22:14:22Z |3b| quit (Remote host closed the connection) 2017-03-11T22:36:52Z |3b| joined #sbcl 2017-03-11T22:50:09Z gingerale quit (Remote host closed the connection) 2017-03-11T23:04:36Z rumbler31 joined #sbcl 2017-03-11T23:16:40Z rumbler31 quit (Remote host closed the connection) 2017-03-11T23:22:24Z stassats joined #sbcl 2017-03-12T00:18:20Z stassats quit (Ping timeout: 260 seconds) 2017-03-12T00:29:18Z karswell quit (Remote host closed the connection) 2017-03-12T00:29:28Z karswell joined #sbcl 2017-03-12T00:34:47Z jamtho_ joined #sbcl 2017-03-12T00:44:05Z karswell quit (Ping timeout: 240 seconds) 2017-03-12T00:48:29Z attila_lendvai joined #sbcl 2017-03-12T00:48:29Z attila_lendvai quit (Changing host) 2017-03-12T00:48:29Z attila_lendvai joined #sbcl 2017-03-12T01:04:23Z jamtho_ quit (Ping timeout: 260 seconds) 2017-03-12T01:12:22Z jamtho_ joined #sbcl 2017-03-12T01:32:42Z Bike joined #sbcl 2017-03-12T01:38:11Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-12T01:53:33Z sjl quit (Read error: Connection reset by peer) 2017-03-12T02:13:25Z jamtho_ quit (Ping timeout: 240 seconds) 2017-03-12T03:32:37Z irsol_ joined #sbcl 2017-03-12T03:34:33Z PuercoPope joined #sbcl 2017-03-12T03:35:16Z les` joined #sbcl 2017-03-12T03:36:39Z les quit (*.net *.split) 2017-03-12T03:36:39Z irsol quit (*.net *.split) 2017-03-12T03:36:40Z drmeister quit (*.net *.split) 2017-03-12T03:36:40Z PuercoPop quit (*.net *.split) 2017-03-12T03:36:42Z irsol_ is now known as irsol 2017-03-12T03:36:52Z PuercoPope is now known as PuercoPop 2017-03-12T03:49:19Z drmeister joined #sbcl 2017-03-12T04:17:27Z rumbler31 joined #sbcl 2017-03-12T04:21:34Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-12T04:44:34Z karswell joined #sbcl 2017-03-12T04:57:52Z rszeno joined #sbcl 2017-03-12T05:10:43Z rszeno quit (Quit: Leaving.) 2017-03-12T05:56:40Z liead is now known as adlai 2017-03-12T06:04:08Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-12T06:14:57Z DGASAU quit (Read error: Connection reset by peer) 2017-03-12T06:18:20Z jack_rabbit joined #sbcl 2017-03-12T06:41:31Z aeth: I'm confused. Is it possible to not be able to declare dynamic-extent and still not stack allocate? 2017-03-12T06:41:57Z aeth: I'm not getting any consing, and I am getting the compiler warning that dynamic extent doesn't work. 2017-03-12T06:45:11Z aeth: Is it because '(simple-array single-float (2)) is small enough for this to not matter in 64-bit? But that doesn't make sense, either because arrays would take up more space than that afaik. 2017-03-12T06:45:25Z aeth: s/still not stack allocate/still not heap allocate/ 2017-03-12T06:47:01Z Bike: it might be too small to be caught by room 2017-03-12T06:48:20Z aeth: ah, it *is* getting caught by room... incrementing the count of simple-vector objects 2017-03-12T06:48:58Z aeth: It's not getting caught by the other things 2017-03-12T06:49:02Z Bike: well, i mean, time or whatever only catches certain size increments 2017-03-12T06:50:29Z DGASAU joined #sbcl 2017-03-12T06:51:12Z schjetne quit (Remote host closed the connection) 2017-03-12T07:54:17Z DGASAU quit (Read error: Connection reset by peer) 2017-03-12T07:58:14Z DGASAU joined #sbcl 2017-03-12T07:58:16Z Bike quit (Quit: u) 2017-03-12T08:18:03Z rumbler31 joined #sbcl 2017-03-12T08:19:20Z shka_ joined #sbcl 2017-03-12T08:20:46Z angavrilov joined #sbcl 2017-03-12T08:22:08Z DGASAU quit (Read error: Connection reset by peer) 2017-03-12T08:22:22Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-12T08:23:22Z DGASAU joined #sbcl 2017-03-12T09:03:52Z gingerale joined #sbcl 2017-03-12T09:39:02Z attila_lendvai joined #sbcl 2017-03-12T09:49:34Z karswell quit (Remote host closed the connection) 2017-03-12T09:50:16Z karswell joined #sbcl 2017-03-12T09:53:11Z jamtho_ joined #sbcl 2017-03-12T10:27:10Z kini quit (Ping timeout: 240 seconds) 2017-03-12T10:28:29Z schjetne joined #sbcl 2017-03-12T10:32:12Z kini joined #sbcl 2017-03-12T10:32:51Z jamtho_ quit (Quit: Leaving) 2017-03-12T10:48:40Z kini quit (Ping timeout: 240 seconds) 2017-03-12T11:12:32Z sjl joined #sbcl 2017-03-12T11:17:25Z attila_lendvai quit (Quit: Leaving.) 2017-03-12T11:25:35Z attila_lendvai joined #sbcl 2017-03-12T11:25:49Z attila_lendvai quit (Changing host) 2017-03-12T11:25:49Z attila_lendvai joined #sbcl 2017-03-12T12:08:29Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-12T12:29:46Z sjl quit (Read error: Connection reset by peer) 2017-03-12T14:38:02Z slyrus joined #sbcl 2017-03-12T14:55:45Z jackdaniel quit (Remote host closed the connection) 2017-03-12T14:57:03Z jackdaniel joined #sbcl 2017-03-12T15:03:25Z slyrus quit (Ping timeout: 240 seconds) 2017-03-12T15:10:32Z angavrilov quit (Ping timeout: 256 seconds) 2017-03-12T15:20:38Z angavrilov joined #sbcl 2017-03-12T15:34:36Z angavrilov_ joined #sbcl 2017-03-12T15:38:05Z angavrilov quit (Ping timeout: 240 seconds) 2017-03-12T16:18:51Z stassats joined #sbcl 2017-03-12T16:19:10Z Bike joined #sbcl 2017-03-12T16:56:10Z nyef` quit (Ping timeout: 264 seconds) 2017-03-12T17:40:25Z kini joined #sbcl 2017-03-12T19:00:40Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-12T19:25:22Z Intensity joined #sbcl 2017-03-12T19:31:11Z Intensity quit (Quit: Quit) 2017-03-12T19:31:21Z Intensity joined #sbcl 2017-03-12T19:46:44Z Intensity quit (Changing host) 2017-03-12T19:46:44Z Intensity joined #sbcl 2017-03-12T19:58:07Z attila_lendvai joined #sbcl 2017-03-12T20:37:24Z prxq joined #sbcl 2017-03-12T20:41:43Z angavrilov_ quit (Remote host closed the connection) 2017-03-12T21:55:25Z gingerale quit (Remote host closed the connection) 2017-03-12T22:23:20Z jack_rabbit joined #sbcl 2017-03-12T22:58:26Z prxq quit (Remote host closed the connection) 2017-03-12T23:02:41Z specbot quit (Disconnected by services) 2017-03-12T23:03:27Z specbot joined #sbcl 2017-03-12T23:05:44Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-12T23:12:28Z shka_ quit (Ping timeout: 240 seconds) 2017-03-12T23:14:33Z rumbler31 joined #sbcl 2017-03-13T00:52:22Z stassats quit (Ping timeout: 260 seconds) 2017-03-13T00:55:34Z cromachina quit (Read error: Connection timed out) 2017-03-13T00:56:33Z cromachina joined #sbcl 2017-03-13T01:15:57Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-13T01:52:43Z rumbler31 quit (Remote host closed the connection) 2017-03-13T02:10:55Z rumbler31 joined #sbcl 2017-03-13T03:05:27Z adlai quit (Ping timeout: 240 seconds) 2017-03-13T03:11:11Z adlai joined #sbcl 2017-03-13T03:21:37Z cromachina quit (Read error: Connection reset by peer) 2017-03-13T03:22:11Z cromachina joined #sbcl 2017-03-13T03:30:51Z karswell` joined #sbcl 2017-03-13T03:31:38Z karswell quit (Read error: Connection reset by peer) 2017-03-13T03:31:51Z rumbler31 quit (Remote host closed the connection) 2017-03-13T04:51:41Z gingerale joined #sbcl 2017-03-13T05:19:18Z eschatologist joined #sbcl 2017-03-13T05:27:42Z csziacobus joined #sbcl 2017-03-13T05:31:04Z csziacobus: think i finally managed to kill the almost identical build machinery in warm.lisp the right way 2017-03-13T05:34:10Z csziacobus: now pcl in compiler 2017-03-13T05:47:30Z igajsin joined #sbcl 2017-03-13T06:47:27Z shka_ joined #sbcl 2017-03-13T07:05:58Z csziacobus quit (Ping timeout: 246 seconds) 2017-03-13T07:21:16Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-13T07:23:27Z shka_ quit (Ping timeout: 240 seconds) 2017-03-13T07:33:26Z jack_rabbit joined #sbcl 2017-03-13T08:01:59Z Bike quit (Quit: gone) 2017-03-13T08:32:37Z rumbler31 joined #sbcl 2017-03-13T08:36:58Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-13T09:04:57Z angavrilov joined #sbcl 2017-03-13T09:15:36Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-13T09:55:01Z sjl joined #sbcl 2017-03-13T11:02:20Z csziacobus joined #sbcl 2017-03-13T11:31:48Z m00natic joined #sbcl 2017-03-13T11:39:19Z milanj joined #sbcl 2017-03-13T12:10:50Z attila_lendvai joined #sbcl 2017-03-13T12:10:50Z attila_lendvai quit (Changing host) 2017-03-13T12:10:50Z attila_lendvai joined #sbcl 2017-03-13T12:31:43Z stassats joined #sbcl 2017-03-13T12:32:06Z stassats: csziacobus: no pcl in the compiler, please 2017-03-13T12:33:10Z rumbler31 joined #sbcl 2017-03-13T12:37:25Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-13T13:24:50Z Xof: I think pcl in cold-core would be good 2017-03-13T13:25:21Z stassats: pcl during make-host-2 is fine 2017-03-13T13:25:26Z stassats: just not in the compiler 2017-03-13T13:27:58Z whiteline quit (Ping timeout: 264 seconds) 2017-03-13T13:31:20Z sjl quit (Ping timeout: 260 seconds) 2017-03-13T13:33:15Z cromachina quit (Read error: Connection reset by peer) 2017-03-13T13:34:08Z csziacobus quit (Ping timeout: 240 seconds) 2017-03-13T14:28:06Z whiteline joined #sbcl 2017-03-13T15:22:10Z quadresce joined #sbcl 2017-03-13T15:49:42Z Bike joined #sbcl 2017-03-13T16:19:17Z slyrus joined #sbcl 2017-03-13T16:28:01Z sebboh``` is now known as sebboh 2017-03-13T16:28:21Z sebboh quit (Changing host) 2017-03-13T16:28:21Z sebboh joined #sbcl 2017-03-13T16:40:16Z sjl joined #sbcl 2017-03-13T16:44:02Z slyrus quit (Ping timeout: 258 seconds) 2017-03-13T16:44:24Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-13T16:58:25Z DeadTrickster joined #sbcl 2017-03-13T17:29:58Z akkad quit (Excess Flood) 2017-03-13T17:30:19Z akkad joined #sbcl 2017-03-13T17:32:47Z shka_ joined #sbcl 2017-03-13T17:56:27Z m00natic quit (Remote host closed the connection) 2017-03-13T17:58:20Z quadresce joined #sbcl 2017-03-13T17:59:50Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-13T18:00:22Z quadresce` joined #sbcl 2017-03-13T18:02:33Z quadresce quit (Ping timeout: 245 seconds) 2017-03-13T18:06:04Z gk-- joined #sbcl 2017-03-13T18:06:52Z gk-- quit (K-Lined) 2017-03-13T18:10:49Z attila_lendvai joined #sbcl 2017-03-13T18:10:49Z attila_lendvai quit (Changing host) 2017-03-13T18:10:49Z attila_lendvai joined #sbcl 2017-03-13T18:16:43Z drmeister: Do I need to build a special version of sbcl? I'm getting this error when I start up cl-jupyter with sbcl: 2017-03-13T18:16:43Z attila_lendvai quit (Read error: Connection reset by peer) 2017-03-13T18:16:46Z drmeister: Unhandled BORDEAUX-THREADS::BORDEAUX-MP-CONDITION: There is no thread support in this instance. 2017-03-13T18:17:45Z drmeister: This is SBCL 1.2.11, an implementation of ANSI Common Lisp. 2017-03-13T18:17:56Z drmeister: I may have installed it with homebrew (OS X ) 2017-03-13T18:23:32Z dougk_: the default build for macOS does not support threads. You have to build it yourself with the --fancy option 2017-03-13T18:35:23Z sjl quit (Ping timeout: 240 seconds) 2017-03-13T18:36:29Z akkad: ^^ and perhaps get a newer one. 1.2.11 is pretty old 2017-03-13T18:40:52Z slyrus joined #sbcl 2017-03-13T18:48:46Z attila_lendvai joined #sbcl 2017-03-13T18:48:46Z attila_lendvai quit (Changing host) 2017-03-13T18:48:46Z attila_lendvai joined #sbcl 2017-03-13T18:58:12Z slyrus quit (Ping timeout: 258 seconds) 2017-03-13T19:02:35Z slyrus joined #sbcl 2017-03-13T19:04:11Z milanj quit (Quit: This computer has gone to sleep) 2017-03-13T19:54:46Z quadresce` quit (Quit: This computer has gone to sleep) 2017-03-13T19:59:38Z quadresce` joined #sbcl 2017-03-13T20:03:19Z slyrus quit (Ping timeout: 246 seconds) 2017-03-13T20:03:48Z quadresce` quit (Ping timeout: 245 seconds) 2017-03-13T20:54:53Z angavrilov quit (Remote host closed the connection) 2017-03-13T20:56:32Z karswell` quit (Ping timeout: 260 seconds) 2017-03-13T21:06:20Z igajsin quit (Ping timeout: 260 seconds) 2017-03-13T21:06:44Z milanj joined #sbcl 2017-03-13T21:18:06Z gingerale quit (Remote host closed the connection) 2017-03-13T21:20:29Z slyrus joined #sbcl 2017-03-13T22:02:20Z sjl joined #sbcl 2017-03-13T22:07:26Z sjl quit (Read error: Connection reset by peer) 2017-03-13T22:08:41Z shka_ quit (Ping timeout: 260 seconds) 2017-03-13T22:19:37Z sjl joined #sbcl 2017-03-13T23:10:33Z sjl quit (Read error: Connection reset by peer) 2017-03-13T23:29:59Z cromachina joined #sbcl 2017-03-13T23:31:56Z scymtym quit (Ping timeout: 256 seconds) 2017-03-13T23:51:45Z scymtym joined #sbcl 2017-03-13T23:57:13Z karswell` joined #sbcl 2017-03-14T00:03:35Z karswell` is now known as karswell 2017-03-14T00:06:30Z scymtym quit (Ping timeout: 256 seconds) 2017-03-14T00:37:50Z milanj quit (Quit: Leaving) 2017-03-14T00:45:32Z slyrus quit (Ping timeout: 268 seconds) 2017-03-14T00:58:03Z dougk_: just noticed that coerce-to-interpreted-function doesn't know the sb-fasteval feature. that's a bug 2017-03-14T01:32:42Z jrm quit (Quit: ciao) 2017-03-14T01:33:14Z jrm joined #sbcl 2017-03-14T02:10:44Z attila_lendvai1 joined #sbcl 2017-03-14T02:10:44Z attila_lendvai quit (Read error: Connection reset by peer) 2017-03-14T02:10:57Z attila_lendvai1 quit (Changing host) 2017-03-14T02:10:57Z attila_lendvai1 joined #sbcl 2017-03-14T02:19:47Z attila_lendvai1 quit (Ping timeout: 252 seconds) 2017-03-14T02:22:11Z csziacobus joined #sbcl 2017-03-14T02:22:31Z csziacobus quit (Client Quit) 2017-03-14T02:44:57Z stassats quit (Ping timeout: 240 seconds) 2017-03-14T05:22:27Z Reinisch quit (Ping timeout: 240 seconds) 2017-03-14T05:30:03Z Suzuran42 joined #sbcl 2017-03-14T05:30:23Z Suzuran quit (Ping timeout: 240 seconds) 2017-03-14T05:30:23Z Suzuran42 is now known as Suzuran 2017-03-14T05:32:32Z akkad left #sbcl 2017-03-14T05:34:08Z igajsin joined #sbcl 2017-03-14T06:07:04Z shka_ joined #sbcl 2017-03-14T06:17:23Z shka_ quit (Ping timeout: 240 seconds) 2017-03-14T06:35:26Z angavrilov joined #sbcl 2017-03-14T07:22:07Z akkad joined #sbcl 2017-03-14T08:25:19Z Bike quit (Quit: leaving) 2017-03-14T09:27:28Z jdz quit (Ping timeout: 240 seconds) 2017-03-14T09:33:11Z jdz joined #sbcl 2017-03-14T09:47:19Z gingerale joined #sbcl 2017-03-14T09:51:51Z attila_lendvai joined #sbcl 2017-03-14T09:51:51Z attila_lendvai quit (Changing host) 2017-03-14T09:51:51Z attila_lendvai joined #sbcl 2017-03-14T11:26:14Z m00natic joined #sbcl 2017-03-14T11:39:47Z sjl joined #sbcl 2017-03-14T11:48:29Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-14T13:39:25Z cromachina quit (Read error: Connection reset by peer) 2017-03-14T14:00:52Z sjl quit (Ping timeout: 240 seconds) 2017-03-14T14:07:50Z scymtym joined #sbcl 2017-03-14T14:33:09Z sjl joined #sbcl 2017-03-14T14:33:35Z myrkraverk quit (Ping timeout: 240 seconds) 2017-03-14T14:33:41Z myrkraverk_ joined #sbcl 2017-03-14T14:34:53Z myrkraverk_ is now known as myrkraverk 2017-03-14T14:40:15Z Reinisch joined #sbcl 2017-03-14T15:03:13Z slyrus_ joined #sbcl 2017-03-14T15:05:26Z slyrus_ is now known as slyrus 2017-03-14T15:16:42Z Reinisch quit (Quit: Have a great day!) 2017-03-14T15:17:00Z milanj joined #sbcl 2017-03-14T15:32:40Z slyrus quit (Ping timeout: 246 seconds) 2017-03-14T15:38:30Z slyrus joined #sbcl 2017-03-14T15:40:22Z Reinisch joined #sbcl 2017-03-14T15:58:13Z slyrus quit (Ping timeout: 246 seconds) 2017-03-14T16:03:15Z slyrus joined #sbcl 2017-03-14T16:07:53Z stassats joined #sbcl 2017-03-14T16:09:16Z quadresce` joined #sbcl 2017-03-14T16:13:48Z quadresce` quit (Ping timeout: 245 seconds) 2017-03-14T16:58:48Z sjl quit (Ping timeout: 240 seconds) 2017-03-14T17:00:37Z shka_ joined #sbcl 2017-03-14T17:01:22Z milanj quit (Quit: This computer has gone to sleep) 2017-03-14T17:02:12Z Bike joined #sbcl 2017-03-14T17:19:21Z sjl joined #sbcl 2017-03-14T17:27:35Z m00natic quit (Remote host closed the connection) 2017-03-14T17:35:10Z slyrus quit (Ping timeout: 246 seconds) 2017-03-14T18:32:59Z sjl quit (Ping timeout: 268 seconds) 2017-03-14T18:37:38Z milanj joined #sbcl 2017-03-14T19:15:19Z sjl joined #sbcl 2017-03-14T20:48:46Z vydd joined #sbcl 2017-03-14T20:48:46Z vydd quit (Changing host) 2017-03-14T20:48:46Z vydd joined #sbcl 2017-03-14T20:56:39Z attila_lendvai joined #sbcl 2017-03-14T20:56:39Z attila_lendvai quit (Changing host) 2017-03-14T20:56:39Z attila_lendvai joined #sbcl 2017-03-14T21:09:17Z slyrus joined #sbcl 2017-03-14T21:16:31Z gingerale quit (Remote host closed the connection) 2017-03-14T21:44:21Z angavrilov quit (Remote host closed the connection) 2017-03-14T22:32:15Z shka_ quit (Ping timeout: 260 seconds) 2017-03-14T22:52:28Z aeth quit (Ping timeout: 240 seconds) 2017-03-14T22:54:38Z aeth joined #sbcl 2017-03-14T23:02:49Z cromachina joined #sbcl 2017-03-14T23:06:08Z jdz quit (Ping timeout: 240 seconds) 2017-03-14T23:06:15Z jdz joined #sbcl 2017-03-14T23:32:14Z attila_lendvai1 joined #sbcl 2017-03-14T23:32:14Z attila_lendvai1 quit (Client Quit) 2017-03-14T23:32:14Z attila_lendvai quit (Disconnected by services) 2017-03-14T23:32:16Z attila_lendvai2 joined #sbcl 2017-03-14T23:32:16Z attila_lendvai2 is now known as attila_lendvai 2017-03-14T23:32:16Z attila_lendvai quit (Changing host) 2017-03-14T23:32:16Z attila_lendvai joined #sbcl 2017-03-15T00:00:20Z vydd quit 2017-03-15T00:14:31Z milanj quit (Quit: Leaving) 2017-03-15T00:28:02Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-15T00:32:06Z cromachina quit (Read error: Connection reset by peer) 2017-03-15T00:33:20Z cromachina joined #sbcl 2017-03-15T00:39:52Z csziacobus joined #sbcl 2017-03-15T00:59:20Z sjl quit (Read error: Connection reset by peer) 2017-03-15T00:59:37Z cromachina quit (Read error: Connection reset by peer) 2017-03-15T01:00:35Z slyrus quit (Ping timeout: 240 seconds) 2017-03-15T01:14:24Z cromachina joined #sbcl 2017-03-15T01:36:15Z BitPuffin|osx joined #sbcl 2017-03-15T01:38:00Z cromachina quit (Read error: Connection reset by peer) 2017-03-15T01:44:44Z cromachina joined #sbcl 2017-03-15T01:48:00Z csziacobus: anyone else have sb-cltl2 fail building/testing on a --fancy build on os x? 2017-03-15T01:52:40Z cromachina quit (Read error: Connection reset by peer) 2017-03-15T02:28:35Z stassats quit (Ping timeout: 240 seconds) 2017-03-15T03:14:42Z quadresce joined #sbcl 2017-03-15T03:19:13Z quadresce quit (Ping timeout: 245 seconds) 2017-03-15T03:47:52Z myrkraverk quit (Ping timeout: 240 seconds) 2017-03-15T03:54:46Z Intensity quit (Ping timeout: 256 seconds) 2017-03-15T04:01:36Z myrkraverk joined #sbcl 2017-03-15T04:40:53Z cromachina joined #sbcl 2017-03-15T04:42:50Z cromachina quit (Client Quit) 2017-03-15T04:43:10Z cromachina joined #sbcl 2017-03-15T05:32:14Z cromachina quit (Ping timeout: 256 seconds) 2017-03-15T05:38:05Z Reinisch quit (Ping timeout: 240 seconds) 2017-03-15T05:40:53Z Reinisch joined #sbcl 2017-03-15T06:05:39Z clintm: csziacobus: I just built head with 1.3.15 on os x and it built just fine, and the tests passed with one unexpected success in wait-on-semaphore. fwiw. 2017-03-15T06:07:57Z karswell quit (Remote host closed the connection) 2017-03-15T06:08:28Z karswell joined #sbcl 2017-03-15T06:17:19Z quadresce joined #sbcl 2017-03-15T06:17:48Z quadresce quit (Client Quit) 2017-03-15T06:18:25Z quadresce joined #sbcl 2017-03-15T06:18:36Z quadresce quit (Client Quit) 2017-03-15T06:19:38Z quadresce joined #sbcl 2017-03-15T06:20:12Z quadresce quit (Client Quit) 2017-03-15T06:20:25Z quadresce joined #sbcl 2017-03-15T06:21:00Z quadresce quit (Client Quit) 2017-03-15T06:21:15Z quadresce joined #sbcl 2017-03-15T06:21:48Z quadresce quit (Client Quit) 2017-03-15T06:22:00Z quadresce joined #sbcl 2017-03-15T06:22:36Z quadresce quit (Client Quit) 2017-03-15T06:33:53Z shka_ joined #sbcl 2017-03-15T06:39:50Z gingerale joined #sbcl 2017-03-15T06:45:44Z angavrilov joined #sbcl 2017-03-15T06:49:45Z akkad left #sbcl 2017-03-15T07:16:41Z shka_ quit (Ping timeout: 258 seconds) 2017-03-15T07:17:33Z jack_rabbit joined #sbcl 2017-03-15T07:54:55Z Bike quit (Quit: sleep) 2017-03-15T08:04:01Z karswell quit (Read error: Connection reset by peer) 2017-03-15T08:04:34Z karswell joined #sbcl 2017-03-15T08:53:11Z engblom left #sbcl 2017-03-15T09:00:22Z whiteline quit (Ping timeout: 264 seconds) 2017-03-15T09:03:51Z sjl joined #sbcl 2017-03-15T09:32:25Z sjl quit (Ping timeout: 246 seconds) 2017-03-15T10:26:24Z whiteline joined #sbcl 2017-03-15T10:30:40Z salva quit (Ping timeout: 268 seconds) 2017-03-15T10:31:06Z salva joined #sbcl 2017-03-15T10:38:02Z sjl joined #sbcl 2017-03-15T10:42:13Z attila_lendvai joined #sbcl 2017-03-15T10:45:20Z specbot quit (Write error: Broken pipe) 2017-03-15T10:45:22Z specbot joined #sbcl 2017-03-15T11:16:22Z m00natic joined #sbcl 2017-03-15T11:57:20Z sjl quit (Ping timeout: 260 seconds) 2017-03-15T12:00:13Z salv0 joined #sbcl 2017-03-15T12:05:11Z cromachina joined #sbcl 2017-03-15T12:05:55Z sjl joined #sbcl 2017-03-15T12:58:33Z milanj joined #sbcl 2017-03-15T13:29:35Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-15T13:29:56Z csziacobus quit (Ping timeout: 256 seconds) 2017-03-15T13:35:40Z attila_lendvai joined #sbcl 2017-03-15T13:35:40Z attila_lendvai quit (Changing host) 2017-03-15T13:35:40Z attila_lendvai joined #sbcl 2017-03-15T13:37:59Z cromachina quit (Read error: Connection reset by peer) 2017-03-15T14:18:32Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-15T14:19:05Z gargaml joined #sbcl 2017-03-15T14:19:14Z quadresce joined #sbcl 2017-03-15T14:38:29Z vydd joined #sbcl 2017-03-15T14:55:04Z quadresce quit (Quit: This computer has gone to sleep) 2017-03-15T15:03:50Z attila_lendvai joined #sbcl 2017-03-15T15:03:50Z attila_lendvai quit (Changing host) 2017-03-15T15:03:50Z attila_lendvai joined #sbcl 2017-03-15T15:22:32Z Bike joined #sbcl 2017-03-15T15:25:08Z phoe quit (Ping timeout: 240 seconds) 2017-03-15T15:36:22Z pkhuong quit (Ping timeout: 264 seconds) 2017-03-15T15:37:11Z pkhuong joined #sbcl 2017-03-15T15:37:34Z pkhuong is now known as Guest8631 2017-03-15T15:38:00Z Cthulhux quit (Ping timeout: 256 seconds) 2017-03-15T15:39:20Z Shinmera quit (Remote host closed the connection) 2017-03-15T15:39:42Z Shinmera joined #sbcl 2017-03-15T15:40:21Z Cthulhux joined #sbcl 2017-03-15T15:40:50Z DeadTrickster quit (Ping timeout: 256 seconds) 2017-03-15T15:45:36Z DeadTrickster joined #sbcl 2017-03-15T15:46:37Z Cthulhux quit (Changing host) 2017-03-15T15:46:37Z Cthulhux joined #sbcl 2017-03-15T15:54:07Z slyrus joined #sbcl 2017-03-15T16:03:24Z sjl quit (Ping timeout: 260 seconds) 2017-03-15T16:08:46Z vydd quit 2017-03-15T16:15:24Z BitPuffin|osx quit (Ping timeout: 256 seconds) 2017-03-15T16:15:57Z slyrus quit (Remote host closed the connection) 2017-03-15T16:21:17Z quadresce joined #sbcl 2017-03-15T16:25:40Z quadresce quit (Ping timeout: 260 seconds) 2017-03-15T16:26:56Z sebboh` joined #sbcl 2017-03-15T16:27:28Z edgar-rft quit (Quit: edgar-rft) 2017-03-15T16:27:52Z clintm quit (Ping timeout: 256 seconds) 2017-03-15T16:28:26Z sebboh quit (Ping timeout: 256 seconds) 2017-03-15T16:30:49Z mateuszb quit (Read error: Connection reset by peer) 2017-03-15T16:31:18Z mateuszb joined #sbcl 2017-03-15T16:41:22Z sjl joined #sbcl 2017-03-15T16:42:11Z phoe joined #sbcl 2017-03-15T16:56:04Z gargaml quit (Read error: Connection reset by peer) 2017-03-15T17:14:23Z gargaml joined #sbcl 2017-03-15T17:14:23Z shka_ joined #sbcl 2017-03-15T17:21:41Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-15T17:31:44Z BitPuffin|osx joined #sbcl 2017-03-15T17:33:21Z gargaml quit (Quit: WeeChat 1.7) 2017-03-15T17:43:37Z m00natic quit (Remote host closed the connection) 2017-03-15T18:01:33Z Reinisch quit (Quit: Have a great day!) 2017-03-15T18:01:53Z Reinisch joined #sbcl 2017-03-15T18:14:52Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-15T18:37:49Z vydd joined #sbcl 2017-03-15T18:37:49Z vydd quit (Changing host) 2017-03-15T18:37:49Z vydd joined #sbcl 2017-03-15T19:19:32Z milanj quit (Quit: Leaving) 2017-03-15T19:26:25Z edgar-rft joined #sbcl 2017-03-15T19:29:27Z sjl quit (Ping timeout: 240 seconds) 2017-03-15T19:56:19Z karswell quit (Read error: Connection reset by peer) 2017-03-15T19:57:14Z karswell joined #sbcl 2017-03-15T20:15:28Z mateuszb quit (Ping timeout: 240 seconds) 2017-03-15T20:19:11Z mateuszb joined #sbcl 2017-03-15T21:19:56Z gingerale quit (Remote host closed the connection) 2017-03-15T21:30:27Z igajsin quit (Ping timeout: 240 seconds) 2017-03-15T21:31:01Z angavrilov quit (Remote host closed the connection) 2017-03-15T21:37:52Z shka_ quit (Ping timeout: 260 seconds) 2017-03-15T21:50:03Z vydd quit 2017-03-15T21:57:55Z dougk_ quit (Ping timeout: 246 seconds) 2017-03-15T22:06:15Z gargaml joined #sbcl 2017-03-15T22:10:14Z dougk_ joined #sbcl 2017-03-15T22:28:01Z attila_lendvai joined #sbcl 2017-03-15T22:47:16Z cromachina joined #sbcl 2017-03-15T22:54:22Z gargaml quit (Quit: WeeChat 1.7) 2017-03-15T23:31:44Z stassats joined #sbcl 2017-03-16T00:17:46Z karswell quit (Ping timeout: 264 seconds) 2017-03-16T00:20:28Z cromachina quit (Read error: Connection reset by peer) 2017-03-16T00:37:47Z whiteline quit (Remote host closed the connection) 2017-03-16T00:39:10Z whiteline joined #sbcl 2017-03-16T00:39:24Z cromachina joined #sbcl 2017-03-16T01:02:58Z salva quit (Ping timeout: 260 seconds) 2017-03-16T01:03:26Z salva joined #sbcl 2017-03-16T01:11:56Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-16T02:18:31Z dougk_: looks like the addition of share-vm broke something: unknown alien type: SB!VM::OS-CONTEXT-REGISTER-T 2017-03-16T02:19:04Z dougk_: somebody didn't run crossbuild-runner/build-all.sh, methinks 2017-03-16T02:20:12Z csziacobus joined #sbcl 2017-03-16T02:20:27Z csziacobus: why are files loaded from warm load not compiled but loaded? am i 2017-03-16T02:20:27Z csziacobus: understanding *compile-files-p* correctly? 2017-03-16T02:21:51Z dougk_: they are compiled, then loaded. why do you think they aren't ? 2017-03-16T02:24:42Z csziacobus: in warm.lisp, do-srcs dispatches on what compile-files-p is 2017-03-16T02:24:46Z csziacobus: and its nil 2017-03-16T02:24:51Z csziacobus: which calls compile-file-pathname 2017-03-16T02:25:01Z csziacobus: and doesn't actually compile the fullname 2017-03-16T02:25:22Z csziacobus: i guess i don't understand why *compile-files-p* exists in the first place 2017-03-16T02:26:59Z dougk_: two things: (1) you've said "files are not compiled but loaded". That's not true. The output clearly says stuff like compiling file ".../src/code/warm-error.lisp" 2017-03-16T02:27:17Z dougk_: and (2) you've said "it's nil". That's also not true. It's unbound, which is defaulted to T 2017-03-16T02:28:17Z csziacobus: about to LOAD warm.lisp (with *compile-files-p* = NIL)? 2017-03-16T02:28:27Z csziacobus: in make-target-2-load.lisp 2017-03-16T02:29:15Z dougk_: yes, in make-target-2-load it's NIL. But you asked why files are not compiled, and i'm claiming they _do_ get compiled just not in make-target-2-load 2017-03-16T02:29:24Z csziacobus: oh okay 2017-03-16T02:29:25Z csziacobus: i see now 2017-03-16T02:29:39Z csziacobus: i didnt see the second pass 2017-03-16T02:29:45Z csziacobus: thank you 2017-03-16T02:32:37Z csziacobus: or i guess i didnt see the first pass; didn't read make-target-2.sh carefully enough 2017-03-16T02:44:37Z Guest8631 is now known as pkhuong 2017-03-16T02:46:56Z whiteline quit (Read error: Connection reset by peer) 2017-03-16T02:47:28Z whiteline joined #sbcl 2017-03-16T03:01:01Z stassats quit (Ping timeout: 246 seconds) 2017-03-16T03:20:58Z csziacobus: oh, well now i'm pretty sure that the compiler output lies when it says that a host file is compiled into a /from-host/.../foo.fasl-tmp during make-host-1, since checking the actual directory shows only /from-host/.../foo.fasl 2017-03-16T03:45:15Z csziacobus: ok, nvm, it seems like it's just silently renaming the object file, and the point is to support build parallelism 2017-03-16T03:47:13Z nyef joined #sbcl 2017-03-16T04:49:24Z jack_rabbit joined #sbcl 2017-03-16T05:33:12Z igajsin joined #sbcl 2017-03-16T06:03:34Z csziacobus quit (Read error: Connection reset by peer) 2017-03-16T06:03:46Z csziacobus joined #sbcl 2017-03-16T06:09:43Z shka_ joined #sbcl 2017-03-16T06:52:50Z Intensity joined #sbcl 2017-03-16T06:55:34Z shka_ quit (Ping timeout: 264 seconds) 2017-03-16T07:14:21Z angavrilov joined #sbcl 2017-03-16T07:38:22Z Bike quit (Quit: sleep tho) 2017-03-16T07:41:25Z igajsin left #sbcl 2017-03-16T08:29:40Z gingerale joined #sbcl 2017-03-16T08:38:05Z jibanes quit (Ping timeout: 240 seconds) 2017-03-16T08:40:01Z jibanes joined #sbcl 2017-03-16T09:09:58Z whiteline quit (Ping timeout: 264 seconds) 2017-03-16T09:22:40Z gargaml joined #sbcl 2017-03-16T09:26:58Z angavrilov quit (Ping timeout: 260 seconds) 2017-03-16T09:28:39Z angavrilov joined #sbcl 2017-03-16T09:43:14Z whiteline joined #sbcl 2017-03-16T10:07:39Z angavrilov quit (Remote host closed the connection) 2017-03-16T10:08:02Z angavrilov joined #sbcl 2017-03-16T10:16:50Z angavrilov quit (Ping timeout: 240 seconds) 2017-03-16T10:17:09Z angavrilov joined #sbcl 2017-03-16T10:22:27Z sjl joined #sbcl 2017-03-16T10:48:11Z attila_lendvai joined #sbcl 2017-03-16T10:48:11Z attila_lendvai quit (Changing host) 2017-03-16T10:48:11Z attila_lendvai joined #sbcl 2017-03-16T11:09:46Z attila_lendvai quit (Quit: Leaving.) 2017-03-16T11:16:20Z angavrilov quit (Ping timeout: 240 seconds) 2017-03-16T11:18:19Z angavrilov joined #sbcl 2017-03-16T11:29:54Z akotlarski joined #sbcl 2017-03-16T11:41:03Z stassats joined #sbcl 2017-03-16T12:46:12Z clintm joined #sbcl 2017-03-16T12:50:50Z edgar-rft quit (Quit: edgar-rft) 2017-03-16T12:54:03Z scymtym quit (Read error: Connection reset by peer) 2017-03-16T12:54:57Z sjl quit (Ping timeout: 240 seconds) 2017-03-16T13:12:54Z scymtym joined #sbcl 2017-03-16T13:27:01Z csziacobus quit (Remote host closed the connection) 2017-03-16T13:27:52Z cromachina quit (Read error: Connection reset by peer) 2017-03-16T13:42:29Z sjl joined #sbcl 2017-03-16T14:09:12Z clintm quit (Remote host closed the connection) 2017-03-16T15:02:28Z rpg joined #sbcl 2017-03-16T15:03:43Z rszeno joined #sbcl 2017-03-16T15:10:51Z Bike joined #sbcl 2017-03-16T15:42:24Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-16T16:07:42Z rszeno quit (Quit: Leaving.) 2017-03-16T16:22:27Z BusFactor1 joined #sbcl 2017-03-16T16:43:16Z dougk_: csziacobus: the point of renaming is not parallelism. It's atomicity of the result. Suppose Lisp crashes (in the sense of immediate process death) while writing a fasl. The fasl is newer than the input. Is it valid? Timestamps would suggest yes 2017-03-16T16:54:33Z gargaml quit (Quit: WeeChat 1.7) 2017-03-16T16:58:05Z nyef: Ooh. No more invalid fasls due to compiler crashes from things like OOM, heap exhaustion, rogue kill -9, and the like? 2017-03-16T16:59:03Z dougk_: nyef: i think he's talking about the self-build scripts. the compiler does not in general do that behavior of using a temporary output file afaik 2017-03-16T16:59:39Z nyef: Aww. /-: 2017-03-16T16:59:40Z dougk_: "he or she or it" is ..., i mean. 2017-03-16T17:00:10Z nyef: ... Spivak pronoun time? 2017-03-16T17:05:46Z stassats: "$0's talking about" 2017-03-16T17:05:50Z stassats: perl-pronouns 2017-03-16T17:30:30Z gargaml joined #sbcl 2017-03-16T17:34:26Z shka_ joined #sbcl 2017-03-16T17:39:32Z shka_ quit (Ping timeout: 260 seconds) 2017-03-16T17:40:43Z shka_ joined #sbcl 2017-03-16T17:48:23Z shka_ quit (Read error: Connection reset by peer) 2017-03-16T17:48:46Z shka_ joined #sbcl 2017-03-16T17:51:43Z mcrist_ joined #sbcl 2017-03-16T17:54:42Z edgar-rft joined #sbcl 2017-03-16T17:54:42Z shka_ quit (Read error: Connection reset by peer) 2017-03-16T17:58:40Z shka_ joined #sbcl 2017-03-16T18:02:05Z akotlarski quit (Remote host closed the connection) 2017-03-16T18:23:01Z dougk_: i wonder if we have a good heuristic, at least for compiler-generated transforms, that the output of make-gensym-list symbols will not be used as "data", i.e. not '#:foo 2017-03-16T18:23:43Z dougk_: the reason this matters to me is that with #+immobile-symbols, jit-compiled functions are using up immobile space with a crap load of symbols 2017-03-16T18:29:17Z sebboh` is now known as sebboh 2017-03-16T18:29:26Z sebboh quit (Changing host) 2017-03-16T18:29:26Z sebboh joined #sbcl 2017-03-16T18:34:49Z foom: What's wrong with putting a crapload of symbols in immobile symbol-space? 2017-03-16T18:41:01Z dougk_: the space is very small relative to dynamic space, and the use-case for immobile symbols is codegen improvement when referencing constant symbols as data, not for all the symbols that the compiler creates 2017-03-16T18:42:05Z dougk_: the compiler can spew out 10 gensyms just for (lambda (x) x), and those all consume immobile space now 2017-03-16T18:46:30Z foom: Time to kill fixed-sizeness of spaces and let them grow as needed, I guess. :) 2017-03-16T18:46:36Z dougk_: i'm working on that ! 2017-03-16T18:47:20Z dougk_: every problem that ever plagued {man,woman}kind as well 2017-03-16T18:51:22Z gargaml quit (Ping timeout: 264 seconds) 2017-03-16T19:29:03Z salv0 quit (Remote host closed the connection) 2017-03-16T19:38:16Z attila_lendvai joined #sbcl 2017-03-16T20:18:19Z mcrist_ quit (Quit: Page closed) 2017-03-16T20:26:43Z eschatologist quit (Quit: ZNC 1.6.3+deb2 - http://znc.in) 2017-03-16T20:27:54Z eschatologist joined #sbcl 2017-03-16T20:43:02Z shka_ quit (Ping timeout: 260 seconds) 2017-03-16T20:45:02Z scymtym quit (Read error: Connection reset by peer) 2017-03-16T20:58:34Z neuronsong joined #sbcl 2017-03-16T21:06:18Z shka_ joined #sbcl 2017-03-16T21:17:04Z dougk_: nyef: I touched some lines of yours 68508f10fb9c235d1f29a55b05f5158e90b8e0b6, you probably don't care, but may as well look at my "fix signed-vs-unsigned" commit 2017-03-16T21:17:46Z vydd joined #sbcl 2017-03-16T21:17:46Z vydd quit (Changing host) 2017-03-16T21:17:47Z vydd joined #sbcl 2017-03-16T21:18:36Z nyef: dougk_: Checking now. 2017-03-16T21:20:55Z angavrilov quit (Remote host closed the connection) 2017-03-16T21:22:39Z nyef: Yeah, looks good. 2017-03-16T21:27:20Z stassats quit (Ping timeout: 268 seconds) 2017-03-16T21:35:35Z shka_ quit (Ping timeout: 240 seconds) 2017-03-16T21:40:20Z stassats joined #sbcl 2017-03-16T21:44:44Z stassats quit (Ping timeout: 258 seconds) 2017-03-16T21:57:30Z gargaml joined #sbcl 2017-03-16T21:59:05Z dougk_: we never came up with a remedy for https://bugs.launchpad.net/sbcl/+bug/1366263 did we? That means testing cheneygc builds is so terribly impractical. "Don't read files" i guess? 2017-03-16T22:00:04Z dougk_: ... actually i put in a mitigation, but it's not working 2017-03-16T22:00:43Z dougk_: https://sourceforge.net/p/sbcl/sbcl/ci/25237efaed44e510eec59eae0286fd8f4f909335 does not work around the problem and the crash is still in the same place 2017-03-16T22:04:26Z vydd quit (Remote host closed the connection) 2017-03-16T22:05:01Z vydd joined #sbcl 2017-03-16T22:07:44Z stassats joined #sbcl 2017-03-16T22:08:09Z nyef: The "right thing" to do is probably to allow WP violation from alien code on cheneygc. 2017-03-16T22:09:22Z vydd quit (Ping timeout: 260 seconds) 2017-03-16T22:11:01Z nyef: But that gets into the question of what to do when alien code trips the auto-gc trigger. 2017-03-16T22:14:50Z phoe quit (Ping timeout: 240 seconds) 2017-03-16T22:20:46Z gargaml quit (Quit: WeeChat 1.7) 2017-03-16T22:22:24Z phoe joined #sbcl 2017-03-16T22:23:01Z gingerale quit (Remote host closed the connection) 2017-03-16T22:26:03Z dougk_: i wonder why i thought i had fixed it, but didn't want to be bothered to examine why i hadn't just now when testing some disassembler cleanups, so i bailed and rebuild without cheneygc. 2017-03-16T22:26:16Z dougk_: that was kind of sad because occasionally i like to build with it to try it out 2017-03-16T22:28:28Z nyef: I've said it before, and I'll say it again: I don't mind us getting rid of cheneygc, so long as we don't drop below having two GCs. 2017-03-16T22:29:20Z foom: Quick, add a new GC. :) 2017-03-16T22:29:29Z nyef: Actually... what if we replace the WP-based auto-gc trigger with an overflow model similar to gencgc? 2017-03-16T22:29:42Z nyef: Down to using an alloc-region type structure? 2017-03-16T22:32:35Z BusFactor1 quit (Ping timeout: 240 seconds) 2017-03-16T22:37:28Z cromachina joined #sbcl 2017-03-16T22:41:16Z nyef: ... Adding that to my project queue. Getting one or two of the cheneygc backends moved over should be quite doable, maybe PPC, MIPS, and ARM? And after that it "just" becomes a longer-term cleanup thing. 2017-03-16T22:57:12Z dougk_: omg. I just realized that if you write, in SBCL source: (let ((foo (+ (something) (* sb-vm:const sb-vm:otherconst))))) we can totally NOT report that as a cold-loader failure 2017-03-16T22:57:21Z dougk_: because the constant was folded out ! wtf ! 2017-03-16T22:57:40Z dougk_: in other words, we don't say "symbol in wrong package for cold load" 2017-03-16T22:58:16Z dougk_: i almost wonder if we need to catch these at read time somehow 2017-03-16T23:05:37Z phoe quit (Ping timeout: 258 seconds) 2017-03-16T23:07:29Z phoe joined #sbcl 2017-03-16T23:08:21Z nyef: uncross-time, maybe? 2017-03-16T23:09:36Z dougk_: i think it's got to do with blind allowance of any host-defined constant maybe? otherwise it should have been a style-warning 2017-03-16T23:09:45Z dougk_: either that, or "parallel build strikes again" 2017-03-16T23:10:06Z nyef: That last should be easy to rule out... or in. 2017-03-16T23:10:09Z dougk_: i have a fix for that - fasls need to store their warnings so that LOAD of a fasl can slurp in a record of what happened 2017-03-16T23:10:42Z dougk_: this is, in general, something we should expose. Fare did something like that in ASDF 2017-03-16T23:11:39Z dougk_: i also want to use sbcl to define a "maximally hostile" host Lisp where everything is wrong, and see if it can build under that 2017-03-16T23:11:57Z dougk_: like 16-bit fixnums, every array promotes to T except string/bit, etc 2017-03-16T23:11:58Z nyef: I was thinking to try to do a cross-build from 2017-03-16T23:11:59Z nyef: Genera. 2017-03-16T23:12:45Z dougk_: of course it is literally impossible to build unless the host supports (simple-array (unsigned-byte 8) (*)), so we need that 2017-03-16T23:13:04Z dougk_: otoh, maybe the host only needs that for stream output and not internal representation ? 2017-03-16T23:15:13Z vydd joined #sbcl 2017-03-16T23:15:42Z vydd quit (Remote host closed the connection) 2017-03-16T23:16:14Z vydd joined #sbcl 2017-03-16T23:20:46Z vydd quit (Ping timeout: 264 seconds) 2017-03-16T23:35:48Z Xof: dougk_: I'm fairly sure I've corrected some of those accidental host constant problems in my time 2017-03-16T23:36:18Z Xof: (they're not as hard to spot as `(... ,most-positive-fixnum) problems) 2017-03-16T23:37:33Z Xof: the maximally hostile lisp implementation thing is potentially fun 2017-03-16T23:38:03Z Xof: low array-dimension-limit, low call-arguments-limit ... 2017-03-16T23:38:13Z Xof: I met an sbcl user today 2017-03-16T23:38:56Z Xof: at a concert of 17th- and 21st-century music 2017-03-16T23:39:03Z dougk_: speaking of low array-dimension-limit, anyone care to speculate about why gencgc has "transother[SIMPLE_ARRAY_WIDETAG] = trans_boxed_large" (around line 4060) ? 2017-03-16T23:39:16Z dougk_: this is like saying you have an array HEADER that is so large it would go on a large object page 2017-03-16T23:40:07Z dougk_: but you can only do that if your array has, O(1000) dimensions on 32-bit and 4000 dimensions on 64-bit 2017-03-16T23:44:55Z Xof: * array-dimension-limit 2017-03-16T23:44:55Z Xof: 4611686018427387901 2017-03-16T23:45:51Z Xof: pdfietz, are you listening? 2017-03-16T23:48:24Z dougk_: but the idea of special-casing it in GC is ... distasteful 2017-03-16T23:48:48Z dougk_: like, oh we want to make sure we pay no copy cost for this thing that for all practical purposes you can't use anyway 2017-03-16T23:49:16Z dougk_: just try to compute the row-major-index efficiently - you're dead anyway for practical purposes 2017-03-17T00:03:39Z sjl quit (Read error: Connection reset by peer) 2017-03-17T00:27:03Z scymtym joined #sbcl 2017-03-17T00:37:51Z csziacobus joined #sbcl 2017-03-17T01:00:50Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-17T01:20:40Z daviid joined #sbcl 2017-03-17T01:37:52Z jibanes quit (Ping timeout: 260 seconds) 2017-03-17T01:39:19Z pkhuong: yeah. set adl to 255, win. 2017-03-17T01:39:41Z stassats: 1024 is the lower limit 2017-03-17T01:39:43Z jibanes joined #sbcl 2017-03-17T01:39:59Z pkhuong: how could I forget :( 2017-03-17T02:16:17Z jack_rabbit joined #sbcl 2017-03-17T02:21:16Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-17T02:42:47Z stassats quit (Read error: Connection reset by peer) 2017-03-17T03:05:28Z daviid quit (Ping timeout: 246 seconds) 2017-03-17T03:52:08Z daviid joined #sbcl 2017-03-17T04:20:00Z neuronsong quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-17T04:24:27Z daviid quit (Ping timeout: 260 seconds) 2017-03-17T04:42:20Z neuronsong joined #sbcl 2017-03-17T05:13:54Z rpg quit (Ping timeout: 256 seconds) 2017-03-17T05:36:37Z Bike quit (Ping timeout: 258 seconds) 2017-03-17T05:40:41Z Bike joined #sbcl 2017-03-17T05:45:06Z neuronsong quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-17T05:54:00Z shka_ joined #sbcl 2017-03-17T06:14:56Z neuronsong joined #sbcl 2017-03-17T06:19:36Z angavrilov joined #sbcl 2017-03-17T06:27:47Z neuronsong quit (Quit: Textual IRC Client: www.textualapp.com) 2017-03-17T06:41:36Z igajsin joined #sbcl 2017-03-17T06:44:51Z shka_ quit (Quit: Konversation terminated!) 2017-03-17T06:45:05Z shka_ joined #sbcl 2017-03-17T07:14:26Z schjetne quit (Read error: Connection reset by peer) 2017-03-17T07:15:44Z schjetne joined #sbcl 2017-03-17T07:21:17Z shka_ quit (Quit: Konversation terminated!) 2017-03-17T08:32:51Z vydd joined #sbcl 2017-03-17T08:32:51Z vydd quit (Changing host) 2017-03-17T08:32:51Z vydd joined #sbcl 2017-03-17T08:59:22Z jack_rabbit joined #sbcl 2017-03-17T09:04:36Z vydd quit 2017-03-17T09:13:00Z i5um41ru joined #sbcl 2017-03-17T09:23:16Z gingerale joined #sbcl 2017-03-17T09:37:33Z Bike quit (Quit: slerp) 2017-03-17T09:50:41Z mateuszb quit (Read error: Connection reset by peer) 2017-03-17T09:51:08Z mateuszb joined #sbcl 2017-03-17T09:53:27Z i5um41ru quit (Ping timeout: 260 seconds) 2017-03-17T11:08:11Z jamtho joined #sbcl 2017-03-17T11:21:43Z schjetne quit (Read error: Connection reset by peer) 2017-03-17T11:23:07Z schjetne joined #sbcl 2017-03-17T11:23:46Z phoe quit (Ping timeout: 264 seconds) 2017-03-17T11:34:28Z m00natic joined #sbcl 2017-03-17T11:40:38Z stassats joined #sbcl 2017-03-17T11:42:15Z sjl joined #sbcl 2017-03-17T12:28:20Z rpg joined #sbcl 2017-03-17T12:33:56Z jamtho quit (Remote host closed the connection) 2017-03-17T12:34:31Z jamtho joined #sbcl 2017-03-17T12:39:44Z phoe joined #sbcl 2017-03-17T13:09:28Z jamtho quit (Ping timeout: 260 seconds) 2017-03-17T13:33:28Z csziacobus quit (Ping timeout: 240 seconds) 2017-03-17T13:34:55Z gargaml joined #sbcl 2017-03-17T14:01:25Z foom: Haha, a 1024 dimension array is the minimum allowed by the standard? What's even the point of a 1024 dimension array. 2017-03-17T14:02:08Z foom: er, minimum maximum allowed, I meant. :) 2017-03-17T14:04:09Z jdz: foom: that's the size of a dimension, not number of dimensions, right? 2017-03-17T14:04:45Z Shinmera: Yes. 2017-03-17T14:04:54Z Shinmera: "The upper exclusive bound on each individual dimension of an array." 2017-03-17T14:05:30Z Shinmera: ARRAY-RANK-LIMIT's minimal maximum is 8. 2017-03-17T14:05:55Z foom: *that* makes more sense. :) 2017-03-17T14:06:07Z jdz: Glossary also says «When multiple array dimensions are to be used together to specify a multi-dimensional array, there is also an implied constraint that the product of all of the dimensions be less than the value of array-total-size-limit.» 2017-03-17T14:07:05Z Shinmera wonders how many applications would break with the array size limit at 1024 2017-03-17T14:07:11Z foom: all of them 2017-03-17T14:07:29Z foom: (approximately) :) 2017-03-17T14:09:54Z Shinmera: Makes one wonder why they set it to that. Can't find an issue related to it on the quick. 2017-03-17T14:11:01Z foom: I mean, what if you want to run common lisp on a device that only has 1K of ram? 2017-03-17T14:11:29Z Shinmera: You won't be able to because the symbol names alone explode that memory. 2017-03-17T14:11:50Z foom: Those all go in ROM! 2017-03-17T14:12:10Z Shinmera: 100Mb ROM, 1K RAM? :) 2017-03-17T14:13:31Z |3b|: could still fit a few 1k element bitvectors :) 2017-03-17T14:15:13Z flip214: pforth, anyone? works with 8kB ROM and a few bytes RAM... 2017-03-17T14:15:30Z flip214: perhaps a CL should be done via a forth interpreter for small systems ;) 2017-03-17T14:20:37Z foom quit (Ping timeout: 246 seconds) 2017-03-17T14:26:54Z BusFactor1 joined #sbcl 2017-03-17T14:31:08Z rpg quit (Ping timeout: 260 seconds) 2017-03-17T14:32:57Z foom joined #sbcl 2017-03-17T14:33:17Z foom: sbcl sets array-rank-limit = 65529. You'd better have almost all those dimensions have a size of 1, or there's no way you're fitting in array-total-size-limit, though! 2017-03-17T14:36:15Z flip214: is there a name for that (arrays where most dimensions are 1)? something like "sparse array"? 2017-03-17T14:37:09Z |3b| would expect something else from 'sparse', maybe 'degenerate'? 2017-03-17T14:38:15Z flip214: "slim" array? 2017-03-17T14:38:27Z flip214: "mostly dimensionless"? 2017-03-17T14:38:41Z flip214 yeah, "mostly harmless". go figure ;) 2017-03-17T14:39:41Z foom: I'd call it "not *quite* as useless as an n-D array with a 0-size dimension." 2017-03-17T14:46:35Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-17T14:50:22Z nyef: Don't forget that 0-D arrays are occasionally useful. (-: 2017-03-17T14:52:44Z flip214: (make-array '(1 3 0 5 6) :element-type '(unsigned-byte 1)) 2017-03-17T14:52:45Z flip214: ; #5A((() () ())) 2017-03-17T14:52:45Z flip214: (read-from-string "#5A((() () ()))") 2017-03-17T14:52:45Z flip214: ; #5A((() () ())) 2017-03-17T14:52:45Z flip214: ; 15 2017-03-17T14:52:59Z flip214: funny, or at least interesting. 2017-03-17T14:54:17Z flip214: sadly the (ARRAY-DIMENSIONS) can't be reconstructed from the printed representation any more. 2017-03-17T14:54:43Z flip214: I get (1 3 0 0 0) ... 2017-03-17T14:54:54Z flip214: yeah, was to be expected. 2017-03-17T15:10:02Z rpg joined #sbcl 2017-03-17T15:21:59Z rpg quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-17T15:26:57Z Bike joined #sbcl 2017-03-17T15:38:39Z decuser joined #sbcl 2017-03-17T15:39:00Z rpg joined #sbcl 2017-03-17T15:39:33Z decuser: trying to figure out how to enter single-step mode in the debugger. Every time I type start, the program continues, but it doesn't single step. 2017-03-17T15:40:01Z decuser: am I supposed to set something up so that the debugger breaks at every "line" or something? 2017-03-17T15:41:10Z decuser: here is the function I wanna debug:(defun sum (n1 n2) (if (zerop n1) n2 (sum (1- n1) (1+ n2)))) 2017-03-17T15:42:42Z decuser: well, that was ugly, a function that calcs the sum by recursion. I call it with a giant first arg and then C-c to enter the debugger. I woudl like to single step from the if the the recursion and see what's going on at each step. Is this possible? 2017-03-17T15:43:46Z dougk_: you need to compile with DEBUG greater than the maximum of SPEED,SPACE, and COMPILATION-SPEED optimize qualities 2017-03-17T15:44:15Z dougk_: and probably with a recursive function you *might* need to declare it as notinline, but i'm not sure about that. Try just the optimize decls first 2017-03-17T15:44:46Z decuser: dougk, that's basically, what the manual says, so I buy it, how do you "compile with DEBUG greater..." 2017-03-17T15:45:17Z dougk_: (declare (optimize debug)) should do it, unless you've also changed the others from their defaults 2017-03-17T15:46:07Z decuser: ok, do I do the declaration in the function definition or interactively with the toplevel? 2017-03-17T15:47:36Z decuser: and is declaim what you mean? 2017-03-17T15:47:46Z decuser: declare generates an error 2017-03-17T15:47:50Z dougk_: in the function, use (declare). at toplevel use (declaim) 2017-03-17T15:47:56Z decuser: ah 2017-03-17T15:48:48Z milanj joined #sbcl 2017-03-17T15:49:18Z decuser: ok, no errors, but same behavior. how to I declare it as not inline? 2017-03-17T15:50:32Z rpg quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-17T15:52:14Z rpg joined #sbcl 2017-03-17T15:56:20Z decuser quit (Ping timeout: 260 seconds) 2017-03-17T16:00:08Z dougk_: you don't need to 2017-03-17T16:09:40Z BitPuffin|osx quit (Ping timeout: 268 seconds) 2017-03-17T16:28:56Z sjl quit (Ping timeout: 260 seconds) 2017-03-17T16:55:19Z daviid joined #sbcl 2017-03-17T17:20:46Z vydd joined #sbcl 2017-03-17T17:27:02Z jamtho joined #sbcl 2017-03-17T17:39:40Z gargaml quit (Quit: WeeChat 1.7) 2017-03-17T17:43:30Z fridim_ joined #sbcl 2017-03-17T17:44:44Z jamtho quit (Ping timeout: 256 seconds) 2017-03-17T17:44:55Z fridim_ quit (Client Quit) 2017-03-17T17:46:02Z fridim_ joined #sbcl 2017-03-17T17:46:12Z fridim_ quit (Read error: Connection reset by peer) 2017-03-17T17:51:30Z m00natic quit (Remote host closed the connection) 2017-03-17T18:12:52Z BitPuffin|osx joined #sbcl 2017-03-17T18:14:36Z BitPuffin|osx quit (Remote host closed the connection) 2017-03-17T18:26:36Z DeadTrickster joined #sbcl 2017-03-17T18:27:24Z milanj quit (Quit: This computer has gone to sleep) 2017-03-17T19:11:10Z scymtym quit (Ping timeout: 264 seconds) 2017-03-17T19:45:35Z luis quit (Ping timeout: 252 seconds) 2017-03-17T19:49:01Z luis joined #sbcl 2017-03-17T20:12:55Z sjl joined #sbcl 2017-03-17T20:15:14Z attila_lendvai joined #sbcl 2017-03-17T20:15:14Z attila_lendvai quit (Changing host) 2017-03-17T20:15:14Z attila_lendvai joined #sbcl 2017-03-17T20:19:03Z igajsin quit (Ping timeout: 258 seconds) 2017-03-17T20:29:35Z DGASAU quit (Ping timeout: 240 seconds) 2017-03-17T20:49:56Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-17T20:50:01Z attila_lendvai1 joined #sbcl 2017-03-17T20:50:01Z attila_lendvai1 is now known as attila_lendvai 2017-03-17T20:50:01Z attila_lendvai quit (Changing host) 2017-03-17T20:50:01Z attila_lendvai joined #sbcl 2017-03-17T20:54:46Z attila_lendvai quit (Read error: Connection reset by peer) 2017-03-17T20:55:20Z attila_lendvai joined #sbcl 2017-03-17T20:55:20Z attila_lendvai quit (Changing host) 2017-03-17T20:55:20Z attila_lendvai joined #sbcl 2017-03-17T20:58:54Z attila_lendvai quit (Client Quit) 2017-03-17T20:58:59Z attila_lendvai1 joined #sbcl 2017-03-17T20:58:59Z attila_lendvai1 is now known as attila_lendvai 2017-03-17T20:58:59Z attila_lendvai quit (Changing host) 2017-03-17T20:58:59Z attila_lendvai joined #sbcl 2017-03-17T21:00:12Z milanj joined #sbcl 2017-03-17T21:44:27Z DGASAU joined #sbcl 2017-03-17T21:47:08Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-17T22:00:39Z angavrilov quit (Remote host closed the connection) 2017-03-17T22:10:25Z phoe quit (Ping timeout: 268 seconds) 2017-03-17T22:10:56Z gingerale quit (Remote host closed the connection) 2017-03-17T22:11:22Z milanj quit (Quit: Leaving) 2017-03-17T22:17:04Z BusFactor1 quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-17T22:31:12Z prxq joined #sbcl 2017-03-17T22:45:17Z prxq quit (Remote host closed the connection) 2017-03-17T22:49:02Z phoe_ joined #sbcl 2017-03-17T23:47:52Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-18T00:14:44Z scymtym joined #sbcl 2017-03-18T00:17:29Z rumbler31 joined #sbcl 2017-03-18T00:26:11Z rumbler31 quit 2017-03-18T00:27:11Z rumbler31 joined #sbcl 2017-03-18T00:54:33Z rumbler31 quit (Remote host closed the connection) 2017-03-18T01:17:25Z vydd quit (Remote host closed the connection) 2017-03-18T01:17:59Z vydd joined #sbcl 2017-03-18T01:22:36Z vydd quit (Ping timeout: 256 seconds) 2017-03-18T01:36:32Z jamtho joined #sbcl 2017-03-18T01:40:57Z jamtho quit (Ping timeout: 240 seconds) 2017-03-18T01:44:42Z stassats quit (Ping timeout: 256 seconds) 2017-03-18T01:45:20Z jamtho joined #sbcl 2017-03-18T02:21:28Z jamtho quit (Ping timeout: 240 seconds) 2017-03-18T02:55:04Z rumbler31 joined #sbcl 2017-03-18T02:59:38Z rumbler31 quit (Ping timeout: 268 seconds) 2017-03-18T03:09:30Z sjl quit (Ping timeout: 268 seconds) 2017-03-18T03:19:16Z vydd joined #sbcl 2017-03-18T03:23:28Z vydd quit (Ping timeout: 240 seconds) 2017-03-18T04:02:28Z daviid quit (Ping timeout: 240 seconds) 2017-03-18T04:20:13Z vydd joined #sbcl 2017-03-18T04:25:21Z vydd quit (Ping timeout: 268 seconds) 2017-03-18T04:51:16Z rszeno joined #sbcl 2017-03-18T04:53:59Z rszeno quit (Client Quit) 2017-03-18T05:09:59Z rpg quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-18T05:54:48Z DGASAU quit (Ping timeout: 240 seconds) 2017-03-18T06:32:49Z angavrilov joined #sbcl 2017-03-18T06:42:44Z vydd joined #sbcl 2017-03-18T06:43:13Z vydd quit (Remote host closed the connection) 2017-03-18T07:14:06Z gingerale joined #sbcl 2017-03-18T07:15:20Z igajsin joined #sbcl 2017-03-18T07:51:14Z attila_lendvai joined #sbcl 2017-03-18T07:51:14Z attila_lendvai quit (Changing host) 2017-03-18T07:51:14Z attila_lendvai joined #sbcl 2017-03-18T08:45:04Z vydd joined #sbcl 2017-03-18T08:50:23Z Bike quit (Quit: leaving) 2017-03-18T08:51:10Z vydd quit (Ping timeout: 246 seconds) 2017-03-18T08:56:25Z rumbler31 joined #sbcl 2017-03-18T08:59:08Z jack_rabbit joined #sbcl 2017-03-18T09:00:37Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-18T09:02:56Z DGASAU joined #sbcl 2017-03-18T09:32:47Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-18T09:36:08Z daviid joined #sbcl 2017-03-18T09:53:09Z vydd joined #sbcl 2017-03-18T09:53:09Z vydd quit (Changing host) 2017-03-18T09:53:09Z vydd joined #sbcl 2017-03-18T10:04:34Z schjetne quit (Ping timeout: 264 seconds) 2017-03-18T11:14:33Z schjetne joined #sbcl 2017-03-18T11:17:59Z jamtho joined #sbcl 2017-03-18T11:21:08Z schjetne quit (Ping timeout: 240 seconds) 2017-03-18T11:22:43Z jamtho quit (Ping timeout: 246 seconds) 2017-03-18T11:25:08Z schjetne joined #sbcl 2017-03-18T11:48:39Z clintm joined #sbcl 2017-03-18T12:02:03Z rpg joined #sbcl 2017-03-18T12:29:26Z rpg quit (Ping timeout: 260 seconds) 2017-03-18T12:41:13Z sjl joined #sbcl 2017-03-18T12:41:27Z rumbler31 joined #sbcl 2017-03-18T12:45:40Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-18T14:05:43Z carvite quit (Remote host closed the connection) 2017-03-18T16:46:44Z Bike joined #sbcl 2017-03-18T17:53:45Z gabnet joined #sbcl 2017-03-18T17:54:44Z sjl quit (Ping timeout: 260 seconds) 2017-03-18T18:04:43Z stassats joined #sbcl 2017-03-18T18:10:02Z vydd quit (Read error: No route to host) 2017-03-18T18:10:25Z vydd joined #sbcl 2017-03-18T18:10:27Z vydd quit (Changing host) 2017-03-18T18:10:27Z vydd joined #sbcl 2017-03-18T18:12:27Z carvite joined #sbcl 2017-03-18T18:21:28Z vydd quit 2017-03-18T18:37:45Z Suzuran quit (Quit: POPJ P,) 2017-03-18T19:21:11Z gabnet quit (Quit: Ex-Chat) 2017-03-18T19:27:17Z angavrilov quit (Remote host closed the connection) 2017-03-18T19:48:25Z csziacobus joined #sbcl 2017-03-18T19:52:16Z csziacobus: so i have a macro foo defined with defmacro in a.lisp which is built before b.lisp in the sbcl build process, where a.lisp is both in the cross compiler and compiled by the cross compiler, whereas b.lisp is only compiled by the cross compiler. during make-host-2, b.lisp complains that foo is an undefined function, whereas when i have foo defined with sb-xc:defmacro, b.lisp uses the definition of the macro in make-host-1 instead of the 2017-03-18T19:52:16Z csziacobus: updated definition in make-host-2. can someone shed some light on how this works? 2017-03-18T19:54:05Z nyef: Magic. 2017-03-18T19:54:12Z nyef: Do we still have def!macro ? 2017-03-18T20:08:53Z csziacobus quit (Ping timeout: 258 seconds) 2017-03-18T20:11:13Z sjl joined #sbcl 2017-03-18T20:31:23Z attila_lendvai joined #sbcl 2017-03-18T20:42:20Z daviid quit (Ping timeout: 260 seconds) 2017-03-18T20:42:46Z schjetne quit (Remote host closed the connection) 2017-03-18T21:03:03Z schjetne joined #sbcl 2017-03-18T21:30:20Z igajsin quit (Ping timeout: 260 seconds) 2017-03-18T21:47:02Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-18T22:54:22Z gingerale quit (Remote host closed the connection) 2017-03-18T23:12:56Z daviid joined #sbcl 2017-03-18T23:28:36Z christoph_debian quit (Ping timeout: 256 seconds) 2017-03-18T23:39:32Z carvite_ joined #sbcl 2017-03-19T00:28:53Z csziacobus joined #sbcl 2017-03-19T00:32:57Z jamtho joined #sbcl 2017-03-19T00:34:03Z jamtho quit (Read error: Connection reset by peer) 2017-03-19T01:20:24Z stassats quit (Ping timeout: 260 seconds) 2017-03-19T01:37:40Z sjl quit (Ping timeout: 260 seconds) 2017-03-19T02:22:00Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-19T02:25:59Z csziacobus: nyef: def!macro is long gone 2017-03-19T02:26:06Z csziacobus: i figured out the issue though 2017-03-19T02:26:27Z csziacobus: deftransforms expanding into macros act strange across build stage boundraries 2017-03-19T02:27:45Z nyef: Okay then. 2017-03-19T02:36:14Z JoshS joined #sbcl 2017-03-19T02:43:28Z daviid quit (Ping timeout: 260 seconds) 2017-03-19T05:50:37Z edgar-rft quit (Quit: edgar-rft) 2017-03-19T06:27:47Z igajsin joined #sbcl 2017-03-19T06:51:37Z carvite__ joined #sbcl 2017-03-19T07:23:06Z angavrilov joined #sbcl 2017-03-19T07:34:34Z gingerale joined #sbcl 2017-03-19T08:52:52Z Bike quit (Quit: leaving) 2017-03-19T09:27:35Z carvite_1 joined #sbcl 2017-03-19T09:31:58Z mateuszb quit (Ping timeout: 258 seconds) 2017-03-19T09:33:29Z mateuszb joined #sbcl 2017-03-19T10:23:54Z shka joined #sbcl 2017-03-19T10:46:37Z gargaml joined #sbcl 2017-03-19T11:11:40Z vydd joined #sbcl 2017-03-19T11:45:16Z shka quit (Read error: No route to host) 2017-03-19T11:45:59Z shka joined #sbcl 2017-03-19T12:02:39Z rpg_ joined #sbcl 2017-03-19T12:08:25Z sjl joined #sbcl 2017-03-19T12:25:21Z sjl quit (Quit: WeeChat 1.3) 2017-03-19T12:31:34Z rpg_ quit (Ping timeout: 264 seconds) 2017-03-19T12:32:32Z sjl joined #sbcl 2017-03-19T13:16:16Z gargaml quit (Ping timeout: 260 seconds) 2017-03-19T13:24:11Z IRCFrEAK joined #sbcl 2017-03-19T13:28:28Z IRCFrEAK quit (K-Lined) 2017-03-19T15:20:47Z stassats joined #sbcl 2017-03-19T15:30:29Z decuser joined #sbcl 2017-03-19T15:30:45Z Bike joined #sbcl 2017-03-19T15:33:41Z decuser: I'm still trying to figure out debugging in sbcl. I wrote a recursive function to sum up two numbers. I then called it (9999999 0) and hit C-c to enter the debugger. 2017-03-19T15:34:23Z decuser: I was able to see the state of my two variables. I would like to single step, but it appears that I can only restart and break again. 2017-03-19T15:35:00Z decuser: I have used (declare (inline sum)) and (declare (optimize (debug x))) to seemingly no avail. 2017-03-19T15:35:09Z decuser: How do you single step in code? 2017-03-19T15:36:11Z decuser: Also, a second question, C-c randomly breaks. Sometimes, it's in my code, and sometimes its way off in system code and I don't see anything other than my toplevel call in the backtrace. Is there a better way to break into my own code? 2017-03-19T15:40:16Z stassats: your code has to be running 2017-03-19T15:43:16Z attila_lendvai joined #sbcl 2017-03-19T16:00:39Z BusFactor1 joined #sbcl 2017-03-19T16:07:19Z BusFactor1 quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-19T16:09:56Z decuser quit (Quit: Page closed) 2017-03-19T16:12:20Z stassats: removed the bogus stack slot on x86-64, now, can we shuffle around return-pc and OCFP slots to avoid POP QWORD PTR [RBP+8]? 2017-03-19T16:13:47Z nyef: Didn't we already swap the return-pc and OCFP once? 2017-03-19T16:14:07Z stassats: but having them in the standard C locations is a good idea 2017-03-19T16:14:28Z stassats: for making gdb slightly less oblivious to lisp code 2017-03-19T16:14:55Z stassats: so that means a different call sequence 2017-03-19T16:15:04Z nyef: Part of the problem is that C frames are built with the arguments "below" the return address, allocated by the caller, and then the new frame is built by the callee, which puts the return address from CALL in the "right place". 2017-03-19T16:15:08Z stassats: supposedly, there would be more calls to functions than functions 2017-03-19T16:15:13Z nyef: Hrm. 2017-03-19T16:15:26Z nyef: Actually, do we have an available SC-number? 2017-03-19T16:15:29Z stassats: so, supposedly, increasing prolog size would be a better idea than increasing call size 2017-03-19T16:16:04Z nyef: What if our incoming args are in a separate SC? 2017-03-19T16:16:11Z stassats: but it all has to be compatible with TC 2017-03-19T16:16:25Z nyef: Turbo C? 2017-03-19T16:16:30Z nyef: Think C? 2017-03-19T16:16:34Z stassats: nyef: it's not args, just OCFP/PC 2017-03-19T16:16:44Z stassats: nyef: tail calls 2017-03-19T16:16:50Z nyef: Ah, right. 2017-03-19T16:16:50Z BusFactor1 joined #sbcl 2017-03-19T16:17:04Z nyef: Yeah, that's a problem. 2017-03-19T16:17:33Z nyef: Well, yes-and-no. We can find the end-of-arguments easily enough, since we know where the start is and we have NARGS. 2017-03-19T16:18:15Z stassats: on an unrelated note, can you have MMU map different threads to different pages? 2017-03-19T16:18:35Z stassats: was thinking of abusing that for thread local special biindings 2017-03-19T16:19:35Z nyef: You want thread-local mappings? 2017-03-19T16:20:05Z nyef: That'd be OS-kernel dependent. 2017-03-19T16:20:19Z stassats: just pulling ideas out of a hat 2017-03-19T16:23:46Z stassats: i forgot, what is the noise sc used for? 2017-03-19T16:23:52Z stassats: i think i tried removing it once already 2017-03-19T16:25:06Z nyef: It's to hold the NFP and NSP TNs for c-stack-is-control-stack. 2017-03-19T16:25:24Z stassats: ok, rather, not what but why 2017-03-19T16:25:57Z nyef: Because they're component-live TNs, and it was easier to dummy them out on x86oids than to conditionally remove them throughout the compiler? 2017-03-19T16:27:50Z stassats: ok, i think i tried to do that 2017-03-19T16:28:07Z stassats: and it wasn't a win 2017-03-19T16:28:50Z nyef: Hrm. Maybe I should try? 2017-03-19T16:29:18Z stassats: conditionalizing a lot of stuff was ugly and no core reduction size or anything measurable 2017-03-19T16:29:29Z nyef: Right, right. 2017-03-19T17:02:35Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-19T17:14:10Z gargaml joined #sbcl 2017-03-19T17:40:08Z gargaml quit (Ping timeout: 240 seconds) 2017-03-19T17:43:17Z gargaml joined #sbcl 2017-03-19T18:04:20Z stassats: fatal error encountered in SBCL pid 16684(tid 0xb000a000): %PRIMITIVE HALT called; the party is over. 2017-03-19T18:04:22Z stassats: what error, where 2017-03-19T18:04:23Z stassats: why 2017-03-19T18:05:12Z nyef: backtrace? 2017-03-19T18:05:30Z stassats: that's from some other thread 2017-03-19T18:06:16Z stassats: ok, got the main thread backtrace, nothing illuminating so far 2017-03-19T18:09:13Z stassats: more than 100 frames, that's why 2017-03-19T18:09:39Z stassats: still no idea 2017-03-19T18:27:06Z rpg_ joined #sbcl 2017-03-19T18:29:05Z _rpg joined #sbcl 2017-03-19T18:31:34Z rpg_ quit (Ping timeout: 256 seconds) 2017-03-19T19:19:20Z csziacobus quit (Ping timeout: 260 seconds) 2017-03-19T20:06:09Z angavrilov quit (Remote host closed the connection) 2017-03-19T20:47:08Z stassats: condition-wait on -sb-futex is wasteful 2017-03-19T20:47:35Z stassats: i just reduced idle consumption of slime from 400 wakeups a second to 40, but that's still wasting energy 2017-03-19T20:51:51Z daviid joined #sbcl 2017-03-19T20:52:10Z stassats: why isn't pthread_cond_wait used? 2017-03-19T20:53:11Z stassats: and wait-for-mutex is just a spinlock with some sleep sprinkled 2017-03-19T20:53:26Z stassats: that's really poor 2017-03-19T20:55:22Z stassats: why have futexes at all? just how bad is pthread_cond_wait / pthread_mutex_lock? 2017-03-19T20:55:59Z prxq joined #sbcl 2017-03-19T21:10:49Z attila_lendvai joined #sbcl 2017-03-19T21:10:57Z attila_lendvai quit (Changing host) 2017-03-19T21:10:57Z attila_lendvai joined #sbcl 2017-03-19T21:17:11Z pkhuong: when futexes are available, they're nice because we don't have to destroy them. 2017-03-19T21:17:40Z stassats: found out there's pthread-futex.c 2017-03-19T21:17:55Z stassats: (bitrotten, of course) 2017-03-19T21:19:02Z nyef: Xb 2017-03-19T21:19:06Z nyef: Err... Oops. 2017-03-19T21:19:57Z christoph_debian joined #sbcl 2017-03-19T21:20:14Z stassats: and where did sb-lutex go? 2017-03-19T21:21:11Z stassats: "killing lutexes, adding timeouts" 2017-03-19T21:22:02Z stassats: there are claims that freebsd uses umtx, but i don't see sb-futex pushed there 2017-03-19T21:22:32Z stassats: sigh 2017-03-19T21:27:50Z stassats: latest macos has dispatch_semaphore and other stuff 2017-03-19T21:33:22Z stassats: and pthread-futex doesn't survive sb-concurrency 2017-03-19T21:34:52Z stassats: and sb-thread:condition-wait still spins 2017-03-19T21:41:48Z igajsin quit (Ping timeout: 240 seconds) 2017-03-19T21:46:46Z prxq quit (Remote host closed the connection) 2017-03-19T22:28:09Z shka quit (Ping timeout: 260 seconds) 2017-03-19T22:52:45Z stassats: defining (defun make-sem () (alien-funcall (extern-alien "dispatch_semaphore_create" (function (* void) long)) 0)) twice does not end well 2017-03-19T22:52:46Z rumbler31 joined #sbcl 2017-03-19T22:52:52Z stassats: cannot use values types here 2017-03-19T22:53:11Z stassats: actually, can't do (describe 'make-sem) 2017-03-19T22:54:37Z stassats: and The function SB-KERNEL:%SIMPLE-FUN-SELF is undefined. 2017-03-19T22:59:50Z carvite_1 quit (Quit: leaving) 2017-03-19T23:03:44Z gingerale quit (Read error: Connection reset by peer) 2017-03-19T23:05:47Z gargaml quit (Ping timeout: 258 seconds) 2017-03-19T23:06:47Z rumbler31 quit (Remote host closed the connection) 2017-03-19T23:12:28Z daviid quit (Ping timeout: 240 seconds) 2017-03-19T23:32:32Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-19T23:38:58Z stassats: using dispatch_semaphore for slime synchronization, reduces wakeup count to 5/s 2017-03-19T23:39:08Z stassats: that's from 40, which is originally from 400 2017-03-19T23:41:24Z stassats: can't exactly use them to implement condition variables, since they don't need a mutex, maybe a new API 2017-03-19T23:44:55Z nyef: Heck with it, just bloody select(3) on a pipe already! 2017-03-19T23:45:42Z stassats: that's in-process communication 2017-03-19T23:45:55Z stassats: and i'd expect select to be broken in some way on macos 2017-03-19T23:46:25Z stassats: and why doesn't it go completely idle 2017-03-19T23:49:23Z nyef: MacOS isn't worth bothering with after about version 7. Maybe version 8. 2017-03-19T23:50:01Z stassats: maybe true if there were any other non-broken oses 2017-03-19T23:50:14Z nyef: There are no non-broken OSes. 2017-03-19T23:50:21Z nyef: All OSes are broken in some respect. 2017-03-19T23:52:35Z stassats: nyef: and do you mean programming for or using? 2017-03-19T23:53:26Z nyef: Oh, using. 2017-03-19T23:53:50Z nyef: MacOS 8 still used a single address space and co-operative multitasking. 2017-03-19T23:54:04Z stassats: found the last energy sapping place in slime, autoflush every 200ms, even if there's no output 2017-03-19T23:54:15Z stassats: it's like slime was designed to be as wasteful as possible 2017-03-19T23:54:47Z stassats: nyef: i like using the current macOS, not so much dealing with low level APIs which nobody else uses 2017-03-19T23:54:53Z vydd quit (Remote host closed the connection) 2017-03-19T23:55:06Z vydd joined #sbcl 2017-03-19T23:55:38Z vydd quit (Remote host closed the connection) 2017-03-19T23:56:22Z nyef: Current macOS is already devolving to unusability. 2017-03-19T23:56:37Z nyef: SLIME wasn't designed. It was bodged together. 2017-03-19T23:57:19Z nyef: I said it yesterday with respect to SLIME, "It's amazing how forward and backward compatible a versioned protocol can be, isn't it?" 2017-03-19T23:58:32Z stassats: well, the protocol isn't really compatible in any direction 2017-03-19T23:58:34Z nyef: Essentially, it's someone's hack that got canonized, but it's still a hack. 2017-03-19T23:58:47Z stassats: but swank tries hard to be across different implementations 2017-03-19T23:59:44Z nyef: Yes, that's my point: It's not designed. If it were designed, it'd be able to deal with an elisp vs. backend mismatch far, far better than it currently mangles^H^H^H^Hages. 2017-03-20T00:00:14Z stassats: and i'm writing a slime replacement, what is it, for four years now? 2017-03-20T00:00:58Z stassats: most time spend on making an emacs replacement 2017-03-20T00:16:34Z stassats: ok, solved the wake up problem completely 2017-03-20T00:17:06Z stassats: and now auto-flushing is even more responsive, since i schedule it only when something hits the buffer, i can set a lower timeout 2017-03-20T00:17:11Z stassats: and it won't spin 2017-03-20T00:17:20Z stassats: too hacky too publish 2017-03-20T00:18:29Z stassats: but no more useless battery drain for me 2017-03-20T00:31:19Z edgar-rft joined #sbcl 2017-03-20T00:34:23Z daviid joined #sbcl 2017-03-20T00:53:39Z carvite_ quit (Remote host closed the connection) 2017-03-20T00:53:44Z carvite__ quit (Remote host closed the connection) 2017-03-20T00:53:48Z carvite quit (Remote host closed the connection) 2017-03-20T00:55:54Z carvite joined #sbcl 2017-03-20T00:56:28Z vydd joined #sbcl 2017-03-20T01:01:10Z vydd quit (Ping timeout: 259 seconds) 2017-03-20T01:18:53Z sjl quit (Read error: Connection reset by peer) 2017-03-20T01:47:16Z stassats quit (Ping timeout: 240 seconds) 2017-03-20T02:57:24Z vydd joined #sbcl 2017-03-20T03:01:48Z vydd quit (Ping timeout: 260 seconds) 2017-03-20T03:08:44Z daviid quit (Ping timeout: 260 seconds) 2017-03-20T03:43:49Z daviid joined #sbcl 2017-03-20T03:58:08Z vydd joined #sbcl 2017-03-20T03:58:48Z edgar-rft quit (Quit: edgar-rft) 2017-03-20T04:02:46Z vydd quit (Ping timeout: 240 seconds) 2017-03-20T04:30:34Z angavrilov joined #sbcl 2017-03-20T04:51:06Z jshjsh joined #sbcl 2017-03-20T04:51:18Z JoshS quit (Disconnected by services) 2017-03-20T04:51:28Z jshjsh is now known as JoshS 2017-03-20T05:09:57Z _rpg quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-20T05:11:21Z edgar-rft joined #sbcl 2017-03-20T05:17:16Z daviid quit (Ping timeout: 240 seconds) 2017-03-20T05:44:57Z igajsin joined #sbcl 2017-03-20T05:59:09Z vydd joined #sbcl 2017-03-20T06:03:48Z vydd quit (Ping timeout: 260 seconds) 2017-03-20T06:07:46Z shka joined #sbcl 2017-03-20T06:47:25Z shka quit (Ping timeout: 246 seconds) 2017-03-20T06:57:03Z vydd joined #sbcl 2017-03-20T06:59:42Z jshjsh joined #sbcl 2017-03-20T07:02:48Z JoshS quit (Ping timeout: 240 seconds) 2017-03-20T07:12:05Z UserJosh joined #sbcl 2017-03-20T07:15:23Z jshjsh quit (Ping timeout: 268 seconds) 2017-03-20T07:23:07Z scymtym quit (Ping timeout: 246 seconds) 2017-03-20T07:42:08Z vydd quit (Remote host closed the connection) 2017-03-20T08:07:32Z vydd joined #sbcl 2017-03-20T08:15:02Z vydd quit 2017-03-20T08:19:45Z UserJosh is now known as JoshS 2017-03-20T08:29:58Z gingerale joined #sbcl 2017-03-20T08:32:07Z attila_lendvai joined #sbcl 2017-03-20T08:32:07Z attila_lendvai quit (Changing host) 2017-03-20T08:32:07Z attila_lendvai joined #sbcl 2017-03-20T08:32:16Z scymtym joined #sbcl 2017-03-20T08:45:23Z gingerale quit (Ping timeout: 258 seconds) 2017-03-20T08:45:26Z Bike quit (Quit: leaving) 2017-03-20T09:06:25Z gargaml joined #sbcl 2017-03-20T09:24:53Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-20T10:23:51Z AndreasO joined #sbcl 2017-03-20T10:23:58Z AndreasO quit (Client Quit) 2017-03-20T10:35:01Z edgar-rft quit (Quit: edgar-rft) 2017-03-20T10:35:59Z m00natic joined #sbcl 2017-03-20T11:28:28Z edgar-rft joined #sbcl 2017-03-20T11:34:08Z sjl joined #sbcl 2017-03-20T11:45:08Z milanj joined #sbcl 2017-03-20T12:00:22Z rpg joined #sbcl 2017-03-20T12:25:26Z rpg quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-20T12:25:48Z gingerale joined #sbcl 2017-03-20T12:59:05Z attila_lendvai joined #sbcl 2017-03-20T13:00:08Z sjl quit (Ping timeout: 240 seconds) 2017-03-20T13:26:12Z cromachina quit (Read error: Connection reset by peer) 2017-03-20T13:29:15Z rpg_ joined #sbcl 2017-03-20T13:31:05Z sjl joined #sbcl 2017-03-20T13:35:49Z BusFactor1 quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-20T13:57:05Z sjl quit (Ping timeout: 240 seconds) 2017-03-20T14:00:45Z sjl joined #sbcl 2017-03-20T14:25:41Z BusFactor1 joined #sbcl 2017-03-20T14:28:32Z daviid joined #sbcl 2017-03-20T14:39:58Z jibanes quit (Ping timeout: 258 seconds) 2017-03-20T14:41:47Z jibanes joined #sbcl 2017-03-20T14:42:57Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-20T14:45:46Z DeadTrickster joined #sbcl 2017-03-20T15:04:03Z phoe_ quit (Ping timeout: 268 seconds) 2017-03-20T15:17:22Z solene quit (Remote host closed the connection) 2017-03-20T15:36:10Z jack_rabbit joined #sbcl 2017-03-20T15:41:04Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-20T15:47:36Z daviid quit (Ping timeout: 260 seconds) 2017-03-20T15:47:50Z galdor joined #sbcl 2017-03-20T15:48:48Z galdor: fyi, one cannot clone either the official git repository or the github mirror: error: object e21fb12b6bfe76a1fbe652a5aaca7779612fe766: missingSpaceBeforeDate: invalid author/committer line - missing space before date 2017-03-20T15:52:12Z foom: galdor: Which repo is that in? 2017-03-20T15:52:36Z galdor: both git://git.code.sf.net/p/sbcl/sbcl and https://github.com/sbcl/sbcl.git 2017-03-20T15:54:18Z flip214: galdor: after "git remote update" a "git show e21fb12b6bfe76a1fbe652a5aaca7779612fe766" works for me just fine 2017-03-20T15:54:36Z Shinmera: galdor: WOMM™ 2017-03-20T15:54:38Z foom: I think you have fsck turned on 2017-03-20T15:54:40Z flip214: and I can pull just fine, too 2017-03-20T15:54:43Z foom: (which is not the default in git) 2017-03-20T15:54:51Z foom: That tag has been there since ages ago. 2017-03-20T15:55:17Z flip214: error in tag e21fb12b6bfe76a1fbe652a5aaca7779612fe766: missingSpaceBeforeDate: invalid author/committer line - missing space before date 2017-03-20T15:55:20Z flip214: error in tag 011f648795b86569462be4974bd7bcf2aea21ca4: missingSpaceBeforeDate: invalid author/committer line - missing space before date 2017-03-20T15:55:29Z flip214: this is what I get on "git fsck" 2017-03-20T15:55:33Z flip214: but otherwise, it works fine 2017-03-20T15:55:52Z foom: yea, someone should delete the tags "start" and "tabby" 2017-03-20T15:55:55Z foom: they're useless anyways. 2017-03-20T15:58:39Z foom: galdor: anyways, you have one of the "*.fsckObjects" configs enabled in your git -- which is not default. You can disable that to get the clone to work. 2017-03-20T15:58:43Z foom: (although, this should be fixed in the repo, too) 2017-03-20T15:59:10Z galdor: yep, I have fsckObjects = true 2017-03-20T15:59:14Z galdor: silly me for being careful ;) 2017-03-20T15:59:53Z sjl quit (Ping timeout: 260 seconds) 2017-03-20T16:00:59Z galdor: thank you for the help 2017-03-20T16:04:47Z shka_ joined #sbcl 2017-03-20T16:07:51Z Bike joined #sbcl 2017-03-20T16:36:14Z milanj quit (Quit: This computer has gone to sleep) 2017-03-20T16:39:47Z sjl joined #sbcl 2017-03-20T17:07:42Z galdor left #sbcl 2017-03-20T17:09:43Z daviid joined #sbcl 2017-03-20T17:32:27Z stassats joined #sbcl 2017-03-20T17:36:18Z jibanes quit (Ping timeout: 258 seconds) 2017-03-20T17:38:02Z jibanes joined #sbcl 2017-03-20T17:44:01Z macdavid joined #sbcl 2017-03-20T17:50:24Z karswell joined #sbcl 2017-03-20T17:51:50Z m00natic quit (Remote host closed the connection) 2017-03-20T17:52:12Z gargaml quit (Ping timeout: 260 seconds) 2017-03-20T18:15:39Z milanj joined #sbcl 2017-03-20T19:00:20Z sjl quit (Ping timeout: 260 seconds) 2017-03-20T19:13:31Z stassats quit (Remote host closed the connection) 2017-03-20T19:13:46Z stassats joined #sbcl 2017-03-20T19:53:45Z daviid quit (Ping timeout: 240 seconds) 2017-03-20T20:10:28Z rumbler31 joined #sbcl 2017-03-20T20:13:26Z neuronsong joined #sbcl 2017-03-20T20:14:03Z jshjsh joined #sbcl 2017-03-20T20:15:13Z aeth_ joined #sbcl 2017-03-20T20:17:45Z JoshS quit (Disconnected by services) 2017-03-20T20:17:48Z jshjsh is now known as JoshS 2017-03-20T20:18:08Z macdavid quit (Ping timeout: 240 seconds) 2017-03-20T20:20:13Z White__Flame joined #sbcl 2017-03-20T20:20:27Z PuercoPope joined #sbcl 2017-03-20T20:23:39Z mood_ joined #sbcl 2017-03-20T20:25:00Z leah2_ joined #sbcl 2017-03-20T20:25:14Z dim` joined #sbcl 2017-03-20T20:25:58Z leah2 quit (Disconnected by services) 2017-03-20T20:26:06Z luis` joined #sbcl 2017-03-20T20:26:12Z leah2_ is now known as leah2 2017-03-20T20:29:11Z luis quit (*.net *.split) 2017-03-20T20:29:11Z aeth quit (*.net *.split) 2017-03-20T20:29:11Z nyef quit (*.net *.split) 2017-03-20T20:29:12Z PuercoPop quit (*.net *.split) 2017-03-20T20:29:13Z mood quit (*.net *.split) 2017-03-20T20:29:13Z White_Flame quit (*.net *.split) 2017-03-20T20:29:13Z dim quit (*.net *.split) 2017-03-20T20:29:14Z PuercoPope is now known as PuercoPop 2017-03-20T20:29:14Z dim` is now known as dim 2017-03-20T20:34:49Z gargaml joined #sbcl 2017-03-20T20:35:33Z stassats quit (Quit: ERC (IRC client for Emacs 25.1.2)) 2017-03-20T20:35:40Z stassats` joined #sbcl 2017-03-20T20:40:26Z scymtym_ joined #sbcl 2017-03-20T20:44:08Z scymtym quit (Ping timeout: 258 seconds) 2017-03-20T20:46:45Z scymtym_ quit (Ping timeout: 240 seconds) 2017-03-20T20:48:09Z igajsin quit (Ping timeout: 268 seconds) 2017-03-20T21:00:30Z luis` quit (Ping timeout: 256 seconds) 2017-03-20T21:02:13Z luis joined #sbcl 2017-03-20T21:11:08Z angavrilov quit (Remote host closed the connection) 2017-03-20T21:14:32Z macdavid joined #sbcl 2017-03-20T21:19:08Z macdavid quit (Ping timeout: 240 seconds) 2017-03-20T21:20:18Z rumbler31 quit (Remote host closed the connection) 2017-03-20T21:20:54Z mood_ is now known as mood 2017-03-20T21:23:37Z fjl joined #sbcl 2017-03-20T21:24:27Z fjl__ quit (Ping timeout: 240 seconds) 2017-03-20T21:26:59Z rumbler31 joined #sbcl 2017-03-20T21:30:30Z Intensity quit (Changing host) 2017-03-20T21:30:30Z Intensity joined #sbcl 2017-03-20T21:30:47Z gingerale quit (Remote host closed the connection) 2017-03-20T21:34:59Z daviid joined #sbcl 2017-03-20T21:35:12Z scymtym joined #sbcl 2017-03-20T21:36:40Z gargaml quit (Quit: WeeChat 1.7) 2017-03-20T21:46:31Z nyef joined #sbcl 2017-03-20T21:54:42Z BusFactor1 quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-20T22:00:46Z aeth_ is now known as aeth 2017-03-20T22:10:14Z daviid` joined #sbcl 2017-03-20T22:12:14Z daviid quit (Read error: No route to host) 2017-03-20T22:12:58Z daviid`` joined #sbcl 2017-03-20T22:14:28Z daviid` quit (Ping timeout: 240 seconds) 2017-03-20T22:15:31Z macdavid joined #sbcl 2017-03-20T22:20:46Z macdavid quit (Ping timeout: 264 seconds) 2017-03-20T22:48:04Z cromachina joined #sbcl 2017-03-20T22:50:03Z JoshS quit (Quit: Leaving) 2017-03-20T22:57:24Z shka_ quit (Ping timeout: 260 seconds) 2017-03-20T23:02:10Z daviid`` quit (Ping timeout: 264 seconds) 2017-03-20T23:11:31Z attila_lendvai quit (Quit: Leaving.) 2017-03-20T23:11:41Z attila_lendvai joined #sbcl 2017-03-20T23:15:58Z macdavid joined #sbcl 2017-03-20T23:21:22Z macdavid quit (Ping timeout: 264 seconds) 2017-03-21T00:04:28Z milanj quit (Quit: Leaving) 2017-03-21T00:16:55Z macdavid joined #sbcl 2017-03-21T00:21:58Z macdavid quit (Ping timeout: 264 seconds) 2017-03-21T00:22:19Z attila_lendvai quit (Quit: Leaving.) 2017-03-21T00:22:28Z attila_lendvai joined #sbcl 2017-03-21T00:22:28Z attila_lendvai quit (Changing host) 2017-03-21T00:22:28Z attila_lendvai joined #sbcl 2017-03-21T00:35:21Z daviid joined #sbcl 2017-03-21T00:39:32Z stassats`: on x86-64, what if we have thread allocation regions 64K aligned 2017-03-21T00:39:52Z stassats`: and simple have an m16 add 2017-03-21T00:40:02Z stassats`: and jump on overflow 2017-03-21T00:44:06Z stassats`: but wait, right now they are 32768 aligned, but that can mean signed overflow 2017-03-21T00:45:26Z stassats`: or is it 2017-03-21T00:51:16Z nyef: Signed overflow doesn't occur from -1 to 0, so it'd amount to 64K pages that are aligned on a 32K boundary but not a 64K boundary. 2017-03-21T00:51:54Z stassats`: and it's not going to be general, because one allocation can be larger than 64K 2017-03-21T00:52:04Z stassats`: i'm trying it with a constant size 2017-03-21T00:54:54Z Cthulhux left #sbcl 2017-03-21T01:00:16Z stassats`: doesn't quite work 2017-03-21T01:02:15Z stassats`: ok, something's not overflowing 2017-03-21T01:17:54Z macdavid joined #sbcl 2017-03-21T01:22:28Z macdavid quit (Ping timeout: 240 seconds) 2017-03-21T01:26:07Z stassats`: ok, it looks like we check for > region end, and overflowing checks for >= 2017-03-21T01:27:31Z stassats`: but i can check for zf too 2017-03-21T01:29:56Z stassats`: but there's no jump for CF=1 && ZF==0 2017-03-21T01:37:38Z stassats`: oh, and that doesn't mater anyway, i'm losing a bit 2017-03-21T01:41:34Z stassats`: well, it was a nice idea 2017-03-21T01:42:34Z jack_rabbit joined #sbcl 2017-03-21T01:51:57Z daviid quit (Ping timeout: 240 seconds) 2017-03-21T02:09:22Z stassats` quit (Ping timeout: 264 seconds) 2017-03-21T02:15:46Z jack_rabbit quit (Remote host closed the connection) 2017-03-21T02:18:51Z macdavid joined #sbcl 2017-03-21T02:22:30Z pkhuong: an m16 add would take a strong hit for the narrow to wide forwarding. 2017-03-21T02:22:54Z nyef: Clearly, then, we need 4G alocation regions. d-: 2017-03-21T02:23:09Z pkhuong: if you want to lower the size of the thread-local allocation struct, you could still check if (new ^ old) >= constant 2017-03-21T02:23:28Z macdavid quit (Ping timeout: 240 seconds) 2017-03-21T02:23:48Z pkhuong: nyef: if only we had the address space. 2017-03-21T02:48:17Z Guest45792 joined #sbcl 2017-03-21T03:00:52Z jack_rabbit joined #sbcl 2017-03-21T03:01:55Z Guest45792 quit (Quit: WeeChat 0.3.8) 2017-03-21T03:19:20Z macdavid joined #sbcl 2017-03-21T03:24:22Z macdavid quit (Ping timeout: 264 seconds) 2017-03-21T03:31:08Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-21T04:20:18Z macdavid joined #sbcl 2017-03-21T04:24:48Z macdavid quit (Ping timeout: 240 seconds) 2017-03-21T04:45:08Z neuronsong quit (Quit: Textual IRC Client: www.textualapp.com) 2017-03-21T05:09:48Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-21T05:09:57Z rpg_ quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-21T05:11:38Z jack_rabbit joined #sbcl 2017-03-21T05:21:17Z macdavid joined #sbcl 2017-03-21T05:26:10Z macdavid quit (Ping timeout: 264 seconds) 2017-03-21T05:55:47Z fiddlerwoaroof quit (Ping timeout: 260 seconds) 2017-03-21T05:58:54Z fiddlerwoaroof joined #sbcl 2017-03-21T06:15:39Z igajsin joined #sbcl 2017-03-21T06:21:43Z macdavid joined #sbcl 2017-03-21T06:22:46Z angavrilov joined #sbcl 2017-03-21T06:26:46Z macdavid quit (Ping timeout: 264 seconds) 2017-03-21T06:51:16Z macdavid joined #sbcl 2017-03-21T06:58:37Z scymtym quit (Ping timeout: 258 seconds) 2017-03-21T07:10:42Z jsnell: has anyone been working on AVX/AVX2 support? 2017-03-21T07:15:46Z rumbler31 quit (Read error: Connection reset by peer) 2017-03-21T07:17:59Z rumbler31 joined #sbcl 2017-03-21T07:23:28Z macdavid quit (Ping timeout: 240 seconds) 2017-03-21T07:24:03Z gingerale joined #sbcl 2017-03-21T07:44:28Z flip214: foom: deleting the tags locally doesn't help much... they're fetched again. 2017-03-21T07:45:17Z flip214: so someone with access might want to _set_ these two tags to new commits... then (deleting and) fetching them again would fix the error messages 2017-03-21T08:19:45Z scymtym joined #sbcl 2017-03-21T08:19:48Z jdz quit (Ping timeout: 240 seconds) 2017-03-21T08:23:47Z jdz joined #sbcl 2017-03-21T08:24:07Z Bike quit (Quit: leaving) 2017-03-21T08:31:24Z scymtym_ joined #sbcl 2017-03-21T08:35:36Z scymtym quit (Ping timeout: 258 seconds) 2017-03-21T08:36:57Z gargaml joined #sbcl 2017-03-21T08:54:09Z scymtym_ quit (Remote host closed the connection) 2017-03-21T08:59:26Z scymtym joined #sbcl 2017-03-21T10:34:59Z rumbler31 quit (Remote host closed the connection) 2017-03-21T10:39:44Z m00natic joined #sbcl 2017-03-21T11:06:29Z rumbler31 joined #sbcl 2017-03-21T11:07:09Z rumbler31 quit (Remote host closed the connection) 2017-03-21T11:10:16Z rumbler31 joined #sbcl 2017-03-21T11:50:04Z attila_lendvai joined #sbcl 2017-03-21T12:01:46Z rpg joined #sbcl 2017-03-21T12:15:35Z sjl joined #sbcl 2017-03-21T12:17:11Z rumbler31 quit (Remote host closed the connection) 2017-03-21T12:29:16Z rpg quit (Ping timeout: 256 seconds) 2017-03-21T13:02:43Z stassats joined #sbcl 2017-03-21T13:03:56Z scymtym quit (Remote host closed the connection) 2017-03-21T13:04:08Z scymtym_ joined #sbcl 2017-03-21T13:04:20Z BusFacto_ joined #sbcl 2017-03-21T13:18:58Z rpg joined #sbcl 2017-03-21T13:40:07Z stassats quit (Ping timeout: 260 seconds) 2017-03-21T13:45:38Z stassats joined #sbcl 2017-03-21T13:46:38Z cromachina quit (Read error: Connection reset by peer) 2017-03-21T13:49:47Z foom: flip214: No, deleting them from upstream I mean 2017-03-21T13:50:17Z flip214: foom: I understand you. 2017-03-21T13:50:43Z flip214: but then _all_ committers need to delete them at the same time, else any push will wreak havoc again. 2017-03-21T13:52:12Z foom: It doesn't push tags by default 2017-03-21T14:34:20Z daviid joined #sbcl 2017-03-21T14:35:32Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-21T15:09:16Z gingerale quit (Remote host closed the connection) 2017-03-21T15:11:51Z gingerale joined #sbcl 2017-03-21T15:35:08Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-21T15:45:02Z BitPuffin|osx joined #sbcl 2017-03-21T16:01:59Z attila_lendvai joined #sbcl 2017-03-21T16:01:59Z attila_lendvai quit (Changing host) 2017-03-21T16:01:59Z attila_lendvai joined #sbcl 2017-03-21T16:46:03Z Bike joined #sbcl 2017-03-21T16:48:53Z shka_ joined #sbcl 2017-03-21T17:18:27Z sjl quit (Ping timeout: 240 seconds) 2017-03-21T17:21:42Z Xof: who screams if we freeze nowish? 2017-03-21T17:22:22Z stassats: i had nothing planned 2017-03-21T17:22:55Z stassats: but can we have a pre-freeze, like, no new features, only fixes? 2017-03-21T17:23:29Z stassats: though some fixes require more effort than some features 2017-03-21T17:23:33Z Xof: they do 2017-03-21T17:24:06Z Xof: I don't have the energy to police ruthelessly the "no fixes if it's not a regression" rule 2017-03-21T17:24:11Z Xof: possibly I shouldn't admit that 2017-03-21T17:24:32Z Xof: but it's the last week of term and I have Just About managed to keep one step ahead of the students 2017-03-21T17:24:36Z stassats: do you plan to release earlier or just a longer freeze period? 2017-03-21T17:24:40Z nyef: "It's a regression from next May's release?" 2017-03-21T17:24:49Z Xof: bonus: I now know a lot of that Computer Science stuff that I've been ignorant of 2017-03-21T17:25:01Z Xof: release on Sunday/Monday 2017-03-21T17:25:26Z Xof: s/a lot/a small amount/ 2017-03-21T17:25:29Z stassats: ok then 2017-03-21T17:25:45Z stassats: i got the windows fix today just in time 2017-03-21T17:26:29Z stassats: (though i would have pushed it even during a freeze) 2017-03-21T17:31:47Z m00natic quit (Remote host closed the connection) 2017-03-21T18:35:46Z gargaml quit (Ping timeout: 264 seconds) 2017-03-21T18:43:37Z vydd joined #sbcl 2017-03-21T19:33:25Z BusFact__ joined #sbcl 2017-03-21T19:35:57Z BusFacto_ quit (Ping timeout: 240 seconds) 2017-03-21T20:07:18Z milanj joined #sbcl 2017-03-21T20:10:14Z karswell` joined #sbcl 2017-03-21T20:10:32Z karswell quit (Read error: Connection reset by peer) 2017-03-21T20:39:43Z scymtym_ quit (Ping timeout: 258 seconds) 2017-03-21T20:52:08Z BusFact__ is now known as burtons 2017-03-21T20:57:51Z stassats: room takes its time when there's a lot of conses 2017-03-21T20:58:05Z stassats: oh, it even exhausted heap, handy 2017-03-21T21:01:09Z stassats: * (defvar *x* (make-list 200000000)) (time (sb-ext:gc :full t)) 2017-03-21T21:01:10Z stassats: 2017-03-21T21:01:10Z stassats: Evaluation took: 2017-03-21T21:01:12Z stassats: 7.014 seconds of real time 2017-03-21T21:02:01Z gingerale quit (Read error: Connection reset by peer) 2017-03-21T21:02:14Z stassats: that's a hefty pause... 2017-03-21T21:03:13Z stassats: second gc takes 2 seconds 2017-03-21T21:06:36Z stassats: ROOM tends to exhaust heap when memory pressure is high 2017-03-21T21:11:27Z nyef: That's... not good. 2017-03-21T21:11:38Z stassats: probably because of without-gcing 2017-03-21T21:12:01Z stassats: i'm trying to break the allocation sequence on x86 2017-03-21T21:12:04Z stassats: on a corner case 2017-03-21T21:12:05Z nyef: Fair point, and there's an argument to at least reduce the scope of the without-gcing. 2017-03-21T21:13:19Z stassats: if the allocation region is high up the dynamic space a new large allocation can overflow the addition 2017-03-21T21:14:14Z stassats: now, how to move the thread allocation region up 2017-03-21T21:16:22Z stassats: failed AVER: (<= (SB-KERNEL:GET-LISP-OBJ-ADDRESS SB-VM::START) (SB-KERNEL:GET-LISP-OBJ-ADDRESS SB-VM::END)) 2017-03-21T21:16:23Z stassats: that's from ROOM 2017-03-21T21:16:31Z stassats: SB-VM::MAP-OBJECTS-IN-RANGE 2017-03-21T21:16:55Z nyef: That's no good. 2017-03-21T21:17:22Z nyef: Smack the alloc_start_page for gen 0, close the allocation region, and cons something to force a new allocation region to open? 2017-03-21T21:17:57Z nyef: The search for a new region begins at the alloc_start_page, so you can push it arbitrarily far into the heap space. 2017-03-21T21:18:40Z nyef: We overflow trap before writing the header, right? 2017-03-21T21:20:11Z stassats: it's add region-start, size, then cmp region-end, region-start 2017-03-21T21:20:21Z stassats: and the add can overflow breaking the comparison 2017-03-21T21:21:10Z stassats: i set out to reduce allocation overhead, i may end up making it worse due to correctness... 2017-03-21T21:21:12Z nyef: "overflow" how? 2017-03-21T21:21:43Z nyef: Unsigned comparison, right? 2017-03-21T21:22:41Z stassats: nyef: there's no check after the ADD 2017-03-21T21:23:04Z stassats: so it can wrap around and the CMP will succeed 2017-03-21T21:23:44Z shka_ quit (Ping timeout: 260 seconds) 2017-03-21T21:24:02Z stassats: ok, i don't want to muck with alloc_start_page, i want the test case to be self-sufficient 2017-03-21T21:26:14Z stassats: ./run-sbcl.sh --dynamic-space-size 3821 (defvar *x* (make-array 500000000)) (room) is enough to trigger failed AVER: (<= (SB-KERNEL:GET-LISP-OBJ-ADDRESS SB-VM::START) (SB-KERNEL:GET-LISP-OBJ-ADDRESS SB-VM::END)) 2017-03-21T21:26:53Z stassats: (SB-VM::MAP-OBJECTS-IN-RANGE # -527981372 545659138) 2017-03-21T21:31:01Z stassats: and full gc with a (make-array 500000000) is slow as well, is the array being moved? 2017-03-21T21:31:57Z stassats: it's not, but that's still 250ms, compared to 18ms with no array 2017-03-21T21:32:08Z scymtym joined #sbcl 2017-03-21T21:34:10Z stassats: or even 1.4 seconds, what does it try to do with that vector? 2017-03-21T21:34:59Z nyef: Scan it for references? 2017-03-21T21:35:16Z stassats: ok, right 2017-03-21T21:35:31Z stassats: the profiler shows 99% is spent in scavenge 2017-03-21T21:40:00Z stassats: (and slime is silly) 2017-03-21T21:40:32Z igajsin quit (Ping timeout: 260 seconds) 2017-03-21T21:44:03Z stassats: even with a large array the alloc region is pretty low 2017-03-21T21:46:13Z stassats: ok, bingo 2017-03-21T21:54:14Z stassats: and a memory fault 2017-03-21T21:55:46Z stassats: http://paste.lisp.org/display/342029 2017-03-21T21:56:01Z stassats: i doubt i can replicated that on x86-64 as easily, though 2017-03-21T21:58:53Z nyef: Well found. 2017-03-21T21:59:50Z stassats: so that add should check for overflow or some other comparison strategy has to be employed 2017-03-21T22:00:47Z stassats: maybe we should maintain size left instead of region-end? 2017-03-21T22:01:36Z pkhuong: or always go to C code if the allocation is large enough? 2017-03-21T22:01:56Z stassats: pkhuong: and it always do 2017-03-21T22:02:04Z pkhuong: before we check for overflow. 2017-03-21T22:02:58Z stassats: pkhuong: a jump on overflow after ADD will achieve that at a lesser cost 2017-03-21T22:09:05Z stassats: filed as https://bugs.launchpad.net/sbcl/+bug/1674839 2017-03-21T22:09:16Z stassats: and easy fix is just to jump on overflow 2017-03-21T22:09:39Z stassats: but all backends need to be audited 2017-03-21T22:12:26Z stassats: perhaps the fact that the allocation regions are page sized can somehow be exploited 2017-03-21T22:13:15Z stassats: i think i can trigger this on x86-64 easily, zero-filled array don't take up any space 2017-03-21T22:13:41Z stassats: i doubt i can make dynamic-space-size as expansive, though 2017-03-21T22:14:59Z stassats: and fixnums are larger, which will help 2017-03-21T22:17:12Z stassats: i'm able to allocate a 5TB vector 2017-03-21T22:17:44Z angavrilov quit (Remote host closed the connection) 2017-03-21T22:24:50Z stassats: ./run-sbcl.sh --dynamic-space-size 80024GB GC invariant lost, file "gencgc.c", line 4041 2017-03-21T22:24:55Z stassats: gotta love corner cases 2017-03-21T22:25:32Z stassats: that's calloc(page_table_pages, sizeof(struct page)); 2017-03-21T22:32:43Z sjl joined #sbcl 2017-03-21T22:37:00Z Reinisch quit (Ping timeout: 260 seconds) 2017-03-21T22:48:09Z stassats: 50TB heap makes everything quite slow 2017-03-21T22:50:06Z stassats: trying with 2MB pages 2017-03-21T22:50:23Z vydd quit (Remote host closed the connection) 2017-03-21T22:50:36Z vydd joined #sbcl 2017-03-21T22:50:36Z vydd quit (Changing host) 2017-03-21T22:50:36Z vydd joined #sbcl 2017-03-21T22:51:08Z vydd quit (Remote host closed the connection) 2017-03-21T22:51:26Z vydd joined #sbcl 2017-03-21T22:51:26Z vydd quit (Changing host) 2017-03-21T22:51:26Z vydd joined #sbcl 2017-03-21T22:51:56Z vydd quit (Remote host closed the connection) 2017-03-21T22:52:18Z vydd joined #sbcl 2017-03-21T22:52:44Z vydd quit (Remote host closed the connection) 2017-03-21T22:53:13Z stassats: can't build with 2MB pages... 2017-03-21T23:01:21Z stassats: (floor (gspace-byte-address gspace) sb!c:*backend-page-bytes*) fails on static space 2017-03-21T23:18:38Z stassats: going to 1MB, a runtime failure, 512KB, a different runtime failure 2017-03-21T23:33:06Z stassats: Dynamic space usage is: 109,952,245,733,840 bytes. 2017-03-21T23:35:50Z stassats: i don't think this can cause problems on x86-64, but i'm not 100% sure 2017-03-21T23:36:01Z cromachina joined #sbcl 2017-03-21T23:38:15Z stassats: or arm64 2017-03-21T23:38:19Z stassats: on linux, that is 2017-03-21T23:44:33Z stassats: all because the amount of bytes is signed 2017-03-21T23:55:27Z nyef: How about add, trap on carry set, then compare? 2017-03-21T23:55:43Z stassats: yeah, that's my first solution 2017-03-21T23:55:58Z vydd joined #sbcl 2017-03-22T00:10:46Z vydd quit (Remote host closed the connection) 2017-03-22T00:11:03Z vydd joined #sbcl 2017-03-22T00:11:32Z vydd quit (Remote host closed the connection) 2017-03-22T01:00:30Z milanj quit (Quit: Leaving) 2017-03-22T01:14:57Z stassats quit (Ping timeout: 240 seconds) 2017-03-22T01:59:11Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-22T02:12:22Z vydd joined #sbcl 2017-03-22T02:17:10Z vydd quit (Ping timeout: 256 seconds) 2017-03-22T02:28:20Z l04m33 joined #sbcl 2017-03-22T02:28:28Z l04m33 left #sbcl 2017-03-22T02:44:35Z leo_song joined #sbcl 2017-03-22T02:48:52Z sjl quit (Ping timeout: 258 seconds) 2017-03-22T03:13:18Z vydd joined #sbcl 2017-03-22T03:18:24Z vydd quit (Ping timeout: 260 seconds) 2017-03-22T05:14:10Z rpg quit (Ping timeout: 264 seconds) 2017-03-22T05:14:28Z vydd joined #sbcl 2017-03-22T05:19:04Z vydd quit (Ping timeout: 256 seconds) 2017-03-22T05:19:21Z shka_ joined #sbcl 2017-03-22T05:24:28Z edgar-rft quit (Quit: edgar-rft) 2017-03-22T05:30:27Z daviid quit (Ping timeout: 240 seconds) 2017-03-22T06:12:13Z igajsin joined #sbcl 2017-03-22T06:19:07Z angavrilov joined #sbcl 2017-03-22T06:20:16Z shka_ quit (Ping timeout: 256 seconds) 2017-03-22T07:15:04Z vydd joined #sbcl 2017-03-22T07:18:22Z scymtym quit (Ping timeout: 246 seconds) 2017-03-22T07:20:09Z vydd quit (Ping timeout: 268 seconds) 2017-03-22T07:49:11Z vydd joined #sbcl 2017-03-22T07:49:11Z vydd quit (Changing host) 2017-03-22T07:49:11Z vydd joined #sbcl 2017-03-22T08:06:57Z vydd quit (Remote host closed the connection) 2017-03-22T08:07:11Z vydd joined #sbcl 2017-03-22T08:07:11Z vydd quit (Changing host) 2017-03-22T08:07:11Z vydd joined #sbcl 2017-03-22T08:07:42Z vydd quit (Remote host closed the connection) 2017-03-22T08:11:00Z Bike quit (Quit: leaving) 2017-03-22T08:26:33Z scymtym joined #sbcl 2017-03-22T08:40:54Z vydd joined #sbcl 2017-03-22T08:40:57Z vydd quit (Changing host) 2017-03-22T08:40:57Z vydd joined #sbcl 2017-03-22T08:41:01Z gargaml joined #sbcl 2017-03-22T09:05:33Z vydd quit 2017-03-22T09:06:19Z gingerale joined #sbcl 2017-03-22T09:14:09Z daviid joined #sbcl 2017-03-22T09:49:28Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-22T09:51:36Z DeadTrickster joined #sbcl 2017-03-22T10:13:22Z DeadTrickster quit (Ping timeout: 246 seconds) 2017-03-22T10:15:13Z DeadTrickster joined #sbcl 2017-03-22T10:30:57Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-22T10:32:28Z DeadTrickster joined #sbcl 2017-03-22T10:45:56Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-22T10:47:54Z DeadTrickster joined #sbcl 2017-03-22T10:53:52Z myrkraverk quit (Remote host closed the connection) 2017-03-22T10:54:35Z myrkraverk joined #sbcl 2017-03-22T10:58:48Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-22T10:59:56Z DeadTrickster joined #sbcl 2017-03-22T11:04:21Z m00natic joined #sbcl 2017-03-22T11:09:44Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-22T11:10:22Z DeadTrickster joined #sbcl 2017-03-22T11:19:13Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-22T11:19:51Z DeadTrickster joined #sbcl 2017-03-22T11:27:24Z DeadTrickster quit (Ping timeout: 256 seconds) 2017-03-22T11:28:35Z DeadTrickster joined #sbcl 2017-03-22T11:53:14Z attila_lendvai joined #sbcl 2017-03-22T12:01:41Z rpg joined #sbcl 2017-03-22T12:24:34Z DeadTrickster quit (Read error: Connection timed out) 2017-03-22T12:27:00Z DeadTrickster joined #sbcl 2017-03-22T12:29:21Z fjl quit (Remote host closed the connection) 2017-03-22T12:29:33Z fjl joined #sbcl 2017-03-22T12:30:05Z rpg quit (Ping timeout: 240 seconds) 2017-03-22T12:31:26Z DeadTrickster quit (Ping timeout: 256 seconds) 2017-03-22T12:44:24Z fjl quit (Remote host closed the connection) 2017-03-22T12:44:35Z fjl joined #sbcl 2017-03-22T13:11:32Z daviid quit (Ping timeout: 260 seconds) 2017-03-22T13:25:17Z stassats joined #sbcl 2017-03-22T13:25:38Z sjl joined #sbcl 2017-03-22T14:06:57Z cromachina quit (Read error: Connection reset by peer) 2017-03-22T14:25:24Z rpg_ joined #sbcl 2017-03-22T14:37:48Z daviid joined #sbcl 2017-03-22T14:41:25Z DeadTrickster joined #sbcl 2017-03-22T14:51:29Z Reinisch joined #sbcl 2017-03-22T15:00:43Z sjl quit (Ping timeout: 240 seconds) 2017-03-22T15:34:17Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-22T15:35:55Z scymtym quit (Ping timeout: 258 seconds) 2017-03-22T15:37:17Z sjl joined #sbcl 2017-03-22T15:43:34Z daviid quit (Ping timeout: 264 seconds) 2017-03-22T16:26:08Z Bike joined #sbcl 2017-03-22T16:34:19Z attila_lendvai joined #sbcl 2017-03-22T16:56:32Z shka_ joined #sbcl 2017-03-22T17:01:08Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-22T17:16:57Z attila_lendvai quit (Read error: No route to host) 2017-03-22T17:18:33Z scymtym joined #sbcl 2017-03-22T17:19:02Z attila_lendvai joined #sbcl 2017-03-22T17:31:49Z daviid joined #sbcl 2017-03-22T17:33:24Z stassats: huh, checking for overflow on x86 doesn't help 2017-03-22T17:39:57Z stassats: i don't get it 2017-03-22T17:43:48Z stassats: stepping in gdb, the overflow branch is taken 2017-03-22T17:47:27Z m00natic quit (Remote host closed the connection) 2017-03-22T17:49:26Z stassats: /* Check whether there is room in the current alloc region. */ new_free_pointer = my_region->free_pointer + nbytes; 2017-03-22T17:49:28Z stassats: right... 2017-03-22T17:50:40Z vydd joined #sbcl 2017-03-22T17:52:28Z stassats: why does it even check again? 2017-03-22T17:54:44Z stassats: and wait, why doesn't (nbytes>=LARGE_OBJECT_SIZE) trigger 2017-03-22T18:20:46Z gargaml quit (Ping timeout: 264 seconds) 2017-03-22T18:42:20Z DeadTrickster joined #sbcl 2017-03-22T19:01:37Z milanj joined #sbcl 2017-03-22T19:16:00Z daviid quit (Ping timeout: 260 seconds) 2017-03-22T19:20:15Z edgar-rft joined #sbcl 2017-03-22T19:28:35Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-22T19:29:41Z attila_lendvai joined #sbcl 2017-03-22T19:29:41Z attila_lendvai quit (Changing host) 2017-03-22T19:29:41Z attila_lendvai joined #sbcl 2017-03-22T20:06:45Z angavrilov quit (Remote host closed the connection) 2017-03-22T20:07:13Z igajsin quit (Ping timeout: 240 seconds) 2017-03-22T20:26:13Z DeadTrickster quit (Ping timeout: 246 seconds) 2017-03-22T20:44:13Z edgar-rft quit (Quit: edgar-rft) 2017-03-22T20:55:26Z reinuseslisp joined #sbcl 2017-03-22T20:59:30Z pchrist quit (Remote host closed the connection) 2017-03-22T21:02:57Z sjl quit (Ping timeout: 240 seconds) 2017-03-22T21:07:28Z reinuseslisp quit (Ping timeout: 240 seconds) 2017-03-22T21:20:22Z burtons quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-22T21:25:47Z gingerale quit (Remote host closed the connection) 2017-03-22T21:28:39Z milanj quit (Quit: Leaving) 2017-03-22T21:48:49Z shka_ quit (Ping timeout: 246 seconds) 2017-03-22T21:58:16Z dougk_ quit (Ping timeout: 246 seconds) 2017-03-22T22:03:30Z daviid joined #sbcl 2017-03-22T22:03:45Z reinuseslisp joined #sbcl 2017-03-22T22:07:28Z les` quit (Ping timeout: 240 seconds) 2017-03-22T22:09:07Z nicdev quit (Ping timeout: 246 seconds) 2017-03-22T22:10:26Z dougk_ joined #sbcl 2017-03-22T22:22:10Z les joined #sbcl 2017-03-22T22:28:55Z burtons joined #sbcl 2017-03-22T22:40:51Z reinuseslisp quit (Quit: Leaving) 2017-03-22T22:41:22Z vydd quit 2017-03-22T22:42:26Z DeadTrickster joined #sbcl 2017-03-22T22:59:52Z DeadTrickster quit (Ping timeout: 246 seconds) 2017-03-22T23:09:56Z DeadTrickster joined #sbcl 2017-03-22T23:47:50Z jack_rabbit joined #sbcl 2017-03-22T23:50:05Z nicdev joined #sbcl 2017-03-23T00:05:50Z karswell` quit (Read error: Connection reset by peer) 2017-03-23T00:07:03Z karswell` joined #sbcl 2017-03-23T00:10:50Z wheelsucker joined #sbcl 2017-03-23T00:13:29Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-23T00:20:16Z cromachina joined #sbcl 2017-03-23T00:25:52Z karswell` quit (Ping timeout: 260 seconds) 2017-03-23T00:26:20Z BitPuffin|osx quit (Ping timeout: 260 seconds) 2017-03-23T01:09:21Z wheelsucker quit (Quit: Client Quit) 2017-03-23T01:19:35Z stassats quit (Ping timeout: 240 seconds) 2017-03-23T01:52:45Z edgar-rft joined #sbcl 2017-03-23T02:43:32Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-23T03:29:14Z nicdev` joined #sbcl 2017-03-23T03:31:27Z nicdev quit (Remote host closed the connection) 2017-03-23T03:55:59Z nicdev` is now known as nicdev 2017-03-23T04:48:36Z burtons quit (Ping timeout: 260 seconds) 2017-03-23T04:54:03Z shka_ joined #sbcl 2017-03-23T05:09:57Z rpg_ quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-23T05:44:22Z igajsin joined #sbcl 2017-03-23T06:08:44Z Ober_ joined #sbcl 2017-03-23T06:08:46Z shka_ quit (Ping timeout: 258 seconds) 2017-03-23T06:13:03Z Ober quit (Ping timeout: 240 seconds) 2017-03-23T06:13:05Z BitPuffin|osx joined #sbcl 2017-03-23T06:33:35Z angavrilov joined #sbcl 2017-03-23T06:38:51Z jack_rabbit joined #sbcl 2017-03-23T07:12:40Z scymtym quit (Ping timeout: 268 seconds) 2017-03-23T07:43:36Z fiddlerwoaroof quit (Ping timeout: 260 seconds) 2017-03-23T07:50:19Z fiddlerwoaroof joined #sbcl 2017-03-23T08:20:46Z fiddlerwoaroof quit (Ping timeout: 264 seconds) 2017-03-23T08:27:55Z scymtym joined #sbcl 2017-03-23T08:29:40Z attila_lendvai joined #sbcl 2017-03-23T08:29:40Z attila_lendvai quit (Changing host) 2017-03-23T08:29:40Z attila_lendvai joined #sbcl 2017-03-23T08:30:49Z fiddlerwoaroof joined #sbcl 2017-03-23T08:46:58Z Bike quit (Quit: leaving) 2017-03-23T08:52:55Z gingerale joined #sbcl 2017-03-23T09:04:40Z fiddlerwoaroof quit (Ping timeout: 246 seconds) 2017-03-23T09:12:01Z fiddlerwoaroof joined #sbcl 2017-03-23T09:26:34Z daviid quit (Ping timeout: 258 seconds) 2017-03-23T10:09:56Z pchrist joined #sbcl 2017-03-23T10:26:57Z angavrilov quit (Ping timeout: 240 seconds) 2017-03-23T10:27:24Z angavrilov joined #sbcl 2017-03-23T10:35:45Z sjl joined #sbcl 2017-03-23T10:39:14Z daviid joined #sbcl 2017-03-23T11:16:47Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-23T11:20:39Z attila_lendvai joined #sbcl 2017-03-23T11:20:39Z attila_lendvai quit (Changing host) 2017-03-23T11:20:39Z attila_lendvai joined #sbcl 2017-03-23T11:26:54Z milanj joined #sbcl 2017-03-23T11:28:13Z m00natic joined #sbcl 2017-03-23T11:33:50Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-23T12:01:49Z rpg_ joined #sbcl 2017-03-23T12:07:27Z BitPuffin|osx quit (Ping timeout: 240 seconds) 2017-03-23T12:26:40Z gargaml joined #sbcl 2017-03-23T13:39:25Z sjl quit (Ping timeout: 260 seconds) 2017-03-23T14:04:05Z cromachina quit (Read error: Connection reset by peer) 2017-03-23T14:04:58Z igajsin quit (Read error: Connection reset by peer) 2017-03-23T14:08:34Z igajsin joined #sbcl 2017-03-23T14:22:38Z igajsin quit (Read error: Connection reset by peer) 2017-03-23T14:22:39Z BitPuffin|osx joined #sbcl 2017-03-23T14:25:39Z igajsin joined #sbcl 2017-03-23T14:36:07Z burtons joined #sbcl 2017-03-23T14:47:02Z wheelsucker joined #sbcl 2017-03-23T15:29:14Z sjl joined #sbcl 2017-03-23T15:59:40Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-23T16:45:45Z shka_ joined #sbcl 2017-03-23T17:00:58Z Bike joined #sbcl 2017-03-23T17:38:42Z gargaml quit (Quit: WeeChat 1.7) 2017-03-23T17:44:31Z m00natic quit (Remote host closed the connection) 2017-03-23T19:35:10Z White__Flame is now known as White_Flame 2017-03-23T20:29:09Z prxq joined #sbcl 2017-03-23T20:49:40Z scymtym quit (Ping timeout: 258 seconds) 2017-03-23T21:15:35Z igajsin quit (Ping timeout: 240 seconds) 2017-03-23T21:16:58Z gingerale quit (Remote host closed the connection) 2017-03-23T21:20:09Z attila_lendvai joined #sbcl 2017-03-23T21:20:09Z attila_lendvai quit (Changing host) 2017-03-23T21:20:09Z attila_lendvai joined #sbcl 2017-03-23T21:23:54Z angavrilov quit (Remote host closed the connection) 2017-03-23T21:38:21Z scymtym joined #sbcl 2017-03-23T21:56:58Z shka_ quit (Ping timeout: 268 seconds) 2017-03-23T22:08:57Z burtons quit (Ping timeout: 240 seconds) 2017-03-23T22:59:18Z prxq quit (Remote host closed the connection) 2017-03-23T23:37:38Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-23T23:52:53Z sjl quit (Read error: Connection reset by peer) 2017-03-24T00:01:34Z wheelsucker quit (Quit: Client Quit) 2017-03-24T01:02:02Z stassats joined #sbcl 2017-03-24T01:04:44Z cromachina joined #sbcl 2017-03-24T01:13:34Z stassats quit (Ping timeout: 260 seconds) 2017-03-24T03:18:40Z xristos quit (Ping timeout: 255 seconds) 2017-03-24T03:19:56Z xristos joined #sbcl 2017-03-24T03:19:56Z xristos is now known as Guest118 2017-03-24T03:38:41Z jrm quit (Ping timeout: 240 seconds) 2017-03-24T03:41:04Z jrm joined #sbcl 2017-03-24T05:14:11Z rpg_ quit (Ping timeout: 240 seconds) 2017-03-24T05:18:14Z mkennedy joined #sbcl 2017-03-24T05:21:56Z shka_ joined #sbcl 2017-03-24T05:38:48Z mkennedy quit (Ping timeout: 240 seconds) 2017-03-24T05:41:06Z igajsin joined #sbcl 2017-03-24T05:55:56Z BitPuffin|osx quit (Ping timeout: 260 seconds) 2017-03-24T06:01:10Z shka_ quit (Ping timeout: 246 seconds) 2017-03-24T06:13:00Z dougk joined #sbcl 2017-03-24T06:14:18Z dougk: hi. I've got some test runs that show 15% overall speedup in gencgc from Mar. 9 to some as yet uncommitted changes. anyone have a favorite gc hammering test I should try? 2017-03-24T06:14:26Z shka_ joined #sbcl 2017-03-24T06:17:25Z milanj quit (Quit: Leaving) 2017-03-24T06:18:40Z milanj joined #sbcl 2017-03-24T06:24:02Z shka_ quit (Ping timeout: 256 seconds) 2017-03-24T06:26:45Z angavrilov joined #sbcl 2017-03-24T07:15:16Z scymtym quit (Ping timeout: 260 seconds) 2017-03-24T07:31:52Z jack_rabbit joined #sbcl 2017-03-24T07:48:17Z edgar-rft quit (Quit: edgar-rft) 2017-03-24T08:28:13Z attila_lendvai joined #sbcl 2017-03-24T08:28:13Z attila_lendvai quit (Changing host) 2017-03-24T08:28:13Z attila_lendvai joined #sbcl 2017-03-24T08:29:45Z scymtym joined #sbcl 2017-03-24T08:39:25Z angavrilov quit (Read error: Connection reset by peer) 2017-03-24T08:40:28Z gargaml joined #sbcl 2017-03-24T08:50:56Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-24T08:58:49Z Bike quit (Quit: leaving) 2017-03-24T10:25:04Z dougk quit (Ping timeout: 246 seconds) 2017-03-24T10:32:54Z angavrilov joined #sbcl 2017-03-24T10:34:09Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-24T10:36:06Z DeadTrickster joined #sbcl 2017-03-24T10:47:10Z m00natic joined #sbcl 2017-03-24T11:05:20Z gingerale joined #sbcl 2017-03-24T11:14:01Z stassats joined #sbcl 2017-03-24T11:18:37Z flip214: sbcl 1.3.14 on debian amd64, lots of boxing/unboxing and no tail call recursion? 2017-03-24T11:18:39Z flip214: http://paste.lisp.org/display/342260 2017-03-24T11:22:47Z flip214: (sb-ext:restrict-compiler-policy) 2017-03-24T11:22:48Z flip214: NIL 2017-03-24T11:26:13Z scymtym: flip214: the recursive calls are not in tail position 2017-03-24T11:27:36Z flip214: ah sorry, yes. misread. 2017-03-24T11:27:55Z flip214: would've needed an accumulator variable. 2017-03-24T11:28:03Z flip214: mis*thought*, more likely. 2017-03-24T11:28:15Z flip214: still, sbcl shouldn't be half as fast on that. 2017-03-24T11:32:05Z flip214: SBCL 1.3.15.107-c9e14929a (ie git HEAD) has the same boxing/unboxing 2017-03-24T11:34:54Z flip214: how can I get the beginning of a function disassembled as well? it starts at 0x12... is that a optimization issue? 2017-03-24T11:40:58Z scymtym: looks like a combination of poor representation selection and maybe suboptimal targeting during register allocation 2017-03-24T11:42:13Z flip214: so much phlogiston to me .... ;P 2017-03-24T11:42:41Z flip214: scymtym: thanks a lot for taking a look, BTW! 2017-03-24T11:54:07Z scymtym: gcc -O3 seems to inline the recursive calls to a certain depths 2017-03-24T12:01:06Z flip214: 00000000000006e0 g F .text 0000000000000328 lotto 2017-03-24T12:01:11Z flip214: yeah, got quite big.... 2017-03-24T12:01:36Z rpg joined #sbcl 2017-03-24T12:16:43Z Xof quit (Ping timeout: 246 seconds) 2017-03-24T12:44:35Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-24T12:45:30Z sjl joined #sbcl 2017-03-24T12:50:05Z DeadTrickster joined #sbcl 2017-03-24T12:54:21Z edgar-rft joined #sbcl 2017-03-24T12:58:12Z DeadTrickster_ joined #sbcl 2017-03-24T13:02:00Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-24T13:10:16Z foom quit (Ping timeout: 246 seconds) 2017-03-24T13:13:41Z cromachina quit (Read error: Connection reset by peer) 2017-03-24T13:21:28Z DeadTrickster_ quit (Ping timeout: 240 seconds) 2017-03-24T13:22:25Z foom joined #sbcl 2017-03-24T13:29:44Z Guest118 is now known as xristos 2017-03-24T13:29:52Z xristos quit (Changing host) 2017-03-24T13:29:52Z xristos joined #sbcl 2017-03-24T13:42:41Z dougk joined #sbcl 2017-03-24T14:00:21Z DeadTrickster joined #sbcl 2017-03-24T14:09:59Z edgar-rft quit (Quit: edgar-rft) 2017-03-24T14:21:46Z dougk__ joined #sbcl 2017-03-24T14:24:52Z dougk quit (Ping timeout: 268 seconds) 2017-03-24T14:38:27Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-24T14:38:43Z m00natic quit (Remote host closed the connection) 2017-03-24T14:42:16Z m00natic joined #sbcl 2017-03-24T14:46:29Z burtons joined #sbcl 2017-03-24T15:02:48Z BitPuffin|osx joined #sbcl 2017-03-24T15:10:01Z DeadTrickster joined #sbcl 2017-03-24T15:21:44Z dougk__ quit (Ping timeout: 260 seconds) 2017-03-24T15:28:22Z dougk joined #sbcl 2017-03-24T15:34:22Z scymtym quit (Ping timeout: 258 seconds) 2017-03-24T15:44:52Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-24T16:29:43Z gargaml quit (Quit: WeeChat 1.7) 2017-03-24T16:33:53Z Bike joined #sbcl 2017-03-24T16:34:51Z DeadTrickster quit (Read error: Connection reset by peer) 2017-03-24T16:41:38Z fiddlerwoaroof quit (Ping timeout: 260 seconds) 2017-03-24T16:46:15Z fiddlerwoaroof joined #sbcl 2017-03-24T16:52:12Z DeadTrickster joined #sbcl 2017-03-24T16:56:36Z shka_ joined #sbcl 2017-03-24T17:09:23Z scymtym joined #sbcl 2017-03-24T17:20:27Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-24T17:27:11Z m00natic quit (Remote host closed the connection) 2017-03-24T17:27:43Z vydd joined #sbcl 2017-03-24T17:27:43Z vydd quit (Changing host) 2017-03-24T17:27:43Z vydd joined #sbcl 2017-03-24T17:34:06Z DeadTrickster joined #sbcl 2017-03-24T19:21:58Z aeth quit (Ping timeout: 246 seconds) 2017-03-24T19:23:14Z aeth joined #sbcl 2017-03-24T19:37:52Z schjetne quit (Read error: Connection reset by peer) 2017-03-24T19:39:33Z schjetne joined #sbcl 2017-03-24T20:47:16Z vydd quit (Remote host closed the connection) 2017-03-24T20:47:31Z vydd joined #sbcl 2017-03-24T20:47:31Z vydd quit (Changing host) 2017-03-24T20:47:31Z vydd joined #sbcl 2017-03-24T20:48:01Z vydd quit (Remote host closed the connection) 2017-03-24T20:48:34Z vydd joined #sbcl 2017-03-24T20:48:34Z vydd quit (Changing host) 2017-03-24T20:48:34Z vydd joined #sbcl 2017-03-24T20:52:41Z vydd quit (Ping timeout: 240 seconds) 2017-03-24T20:55:36Z shka_ quit (Ping timeout: 258 seconds) 2017-03-24T21:05:34Z daviid quit (Ping timeout: 258 seconds) 2017-03-24T21:15:40Z attila_lendvai joined #sbcl 2017-03-24T21:15:40Z attila_lendvai quit (Changing host) 2017-03-24T21:15:40Z attila_lendvai joined #sbcl 2017-03-24T21:41:13Z igajsin quit (Ping timeout: 258 seconds) 2017-03-24T21:49:10Z vydd joined #sbcl 2017-03-24T21:53:48Z vydd quit (Ping timeout: 240 seconds) 2017-03-24T22:06:40Z gingerale quit (Remote host closed the connection) 2017-03-24T22:12:19Z burtons quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-24T22:24:22Z dougk quit (Ping timeout: 264 seconds) 2017-03-24T22:29:43Z daviid joined #sbcl 2017-03-24T22:48:43Z cromachina joined #sbcl 2017-03-24T22:51:45Z jack_rabbit joined #sbcl 2017-03-24T23:25:46Z dougk joined #sbcl 2017-03-24T23:36:18Z dougk quit (Read error: Connection reset by peer) 2017-03-24T23:36:36Z dougk joined #sbcl 2017-03-24T23:43:02Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-24T23:43:46Z dougk quit (Ping timeout: 246 seconds) 2017-03-24T23:46:48Z daviid quit (Ping timeout: 240 seconds) 2017-03-24T23:51:10Z vydd joined #sbcl 2017-03-24T23:55:19Z vydd quit (Ping timeout: 246 seconds) 2017-03-25T00:28:02Z BitPuffin joined #sbcl 2017-03-25T00:28:36Z vydd joined #sbcl 2017-03-25T00:34:04Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-25T01:04:09Z jack_rabbit joined #sbcl 2017-03-25T01:49:57Z stassats quit (Ping timeout: 240 seconds) 2017-03-25T02:06:50Z dougk joined #sbcl 2017-03-25T02:23:16Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-25T02:23:27Z vydd quit (Remote host closed the connection) 2017-03-25T02:36:11Z edgar-rft joined #sbcl 2017-03-25T03:18:48Z sjl quit (Ping timeout: 260 seconds) 2017-03-25T04:44:30Z milanj quit (Quit: This computer has gone to sleep) 2017-03-25T05:14:14Z rpg quit (Ping timeout: 256 seconds) 2017-03-25T06:53:25Z igajsin joined #sbcl 2017-03-25T07:00:40Z dougk quit (Ping timeout: 240 seconds) 2017-03-25T07:41:39Z gingerale joined #sbcl 2017-03-25T07:46:21Z salv0 joined #sbcl 2017-03-25T08:06:10Z shka_ joined #sbcl 2017-03-25T08:31:44Z shka_ quit (Quit: Konversation terminated!) 2017-03-25T08:32:04Z shka joined #sbcl 2017-03-25T08:41:22Z shka quit (Remote host closed the connection) 2017-03-25T08:58:32Z Bike quit (Quit: leaving) 2017-03-25T10:04:18Z milanj joined #sbcl 2017-03-25T10:16:25Z shka joined #sbcl 2017-03-25T11:07:05Z sjl joined #sbcl 2017-03-25T11:30:45Z igajsin quit (Ping timeout: 258 seconds) 2017-03-25T11:47:15Z igajsin joined #sbcl 2017-03-25T11:49:51Z shka quit (Remote host closed the connection) 2017-03-25T11:51:24Z shka joined #sbcl 2017-03-25T11:58:59Z vydd joined #sbcl 2017-03-25T11:58:59Z vydd quit (Changing host) 2017-03-25T11:58:59Z vydd joined #sbcl 2017-03-25T12:00:02Z rpg joined #sbcl 2017-03-25T12:01:57Z rpg_ joined #sbcl 2017-03-25T12:02:29Z shka quit (Quit: Konversation terminated!) 2017-03-25T12:04:34Z rpg quit (Ping timeout: 264 seconds) 2017-03-25T12:07:10Z sjl quit (Ping timeout: 246 seconds) 2017-03-25T12:23:34Z shka joined #sbcl 2017-03-25T12:38:42Z sjl joined #sbcl 2017-03-25T12:45:52Z igajsin quit (Ping timeout: 256 seconds) 2017-03-25T13:04:04Z igajsin joined #sbcl 2017-03-25T13:10:42Z BitPuffin|osx quit (Ping timeout: 260 seconds) 2017-03-25T13:11:43Z dougk joined #sbcl 2017-03-25T13:41:58Z salv0 quit (Ping timeout: 256 seconds) 2017-03-25T13:47:26Z salv0 joined #sbcl 2017-03-25T13:58:14Z dto joined #sbcl 2017-03-25T14:00:25Z sjl__ joined #sbcl 2017-03-25T14:01:43Z dto: hi. recently i started getting an error trying to run any of my games (based on lispbuilder-sdl, but i'm not sure this is strictly related to that) http://pastebin.com/raw/zXEjsgU6 it seems to happen in an interrupt handler. not sure if anything changed about my system since it worked----it even happens on previous SBCL based binary packages i'd made that worked before. furthermore, it doesn't manifest on CCL ---- i prefer to use SBCL 2017-03-25T14:01:43Z dto: for development so this is difficult for me. the SBCL-produced binaries work fine on other machines (windows, linux) but neither will run for me here anymore (either the EXE under Wine, or the linux version) 2017-03-25T14:02:10Z dto: can anyone help point me in the right direction for what might be causing this or how to troubleshoot SB-UNIX::RUN-HANDLER? 2017-03-25T14:02:40Z sjl quit (Ping timeout: 240 seconds) 2017-03-25T14:02:50Z dto: i'm upgrading GL libs now on distro to see if that's an issue 2017-03-25T14:04:38Z dto: wow. upgrading GL libs worked! 2017-03-25T14:04:48Z dto: it works again. hmm i guess that was silly. 2017-03-25T14:04:55Z dto: sorry for noise! :) 2017-03-25T14:09:30Z Shinmera: When in doubt blame nvidia? 2017-03-25T14:13:10Z dto: hi Shinmera . 99% of the time when i see a bug like this, it's drivers yeah 2017-03-25T14:13:13Z dto: i'm on an Intel chip 2017-03-25T14:13:38Z dto: Shinmera: i saw some tweets about your game engine. hows that going? 2017-03-25T14:14:00Z Shinmera: Still working on it. I'm currently stuck on questions about how to set up a shader pipeline system. 2017-03-25T14:14:22Z dto: hey ok 2017-03-25T14:14:30Z dto: several people on #lispgames working on such things. 2017-03-25T14:14:36Z BitPuffin|osx joined #sbcl 2017-03-25T14:14:40Z dto: maybe you'd find some help there 2017-03-25T14:15:04Z Shinmera: lispgames is a bit too noisy for me. 2017-03-25T14:15:40Z dto: ah ok :() 2017-03-25T14:15:41Z dto: :) 2017-03-25T14:17:11Z Shinmera: Not sure if I'll get the thing done before ELS, but I'm hoping I will. 2017-03-25T14:17:46Z dto: keep me posted. i am looking forward to seeing 2017-03-25T14:17:55Z Shinmera: Are you gonna be at ELS? 2017-03-25T14:18:47Z dto: no, but i'll try to see the vids 2017-03-25T14:53:41Z attila_lendvai joined #sbcl 2017-03-25T15:06:03Z JuanDaugherty joined #sbcl 2017-03-25T15:07:54Z JuanDaugherty: is the position on sbcl that it just provides AMOP and generic CLOS support without a specific CLOS as most implementations do or did I miss a presentation of the latter someplace? I believe it would have been PCL in cmucl 2017-03-25T15:10:34Z dto quit (Remote host closed the connection) 2017-03-25T15:16:18Z wheelsucker joined #sbcl 2017-03-25T15:50:36Z nyef quit (Ping timeout: 260 seconds) 2017-03-25T15:51:10Z DeadTrickster quit (Ping timeout: 256 seconds) 2017-03-25T15:51:15Z salv0 quit (Remote host closed the connection) 2017-03-25T15:52:59Z Xof joined #sbcl 2017-03-25T15:56:37Z DeadTrickster joined #sbcl 2017-03-25T16:02:30Z sjl__ quit (Ping timeout: 256 seconds) 2017-03-25T16:13:40Z drl joined #sbcl 2017-03-25T16:14:37Z JuanDaugherty quit (Quit: Hibernate, reboot, exeunt, etc.) 2017-03-25T16:15:42Z drl quit (Remote host closed the connection) 2017-03-25T16:16:15Z drl joined #sbcl 2017-03-25T16:22:43Z drl quit (Quit: Ex-Chat) 2017-03-25T16:27:14Z BitPuffin quit (Remote host closed the connection) 2017-03-25T16:30:35Z BitPuffin joined #sbcl 2017-03-25T16:34:21Z drl joined #sbcl 2017-03-25T16:44:48Z Bike joined #sbcl 2017-03-25T16:56:00Z drl quit (Quit: Ex-Chat) 2017-03-25T18:24:11Z milanj quit (Quit: This computer has gone to sleep) 2017-03-25T18:32:11Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-25T18:34:10Z daviid joined #sbcl 2017-03-25T19:10:06Z jack_rabbit joined #sbcl 2017-03-25T19:20:16Z nyef joined #sbcl 2017-03-25T19:47:42Z sjl__ joined #sbcl 2017-03-25T19:51:24Z sjl__ quit (Client Quit) 2017-03-25T19:57:00Z BitPuffin quit (Ping timeout: 260 seconds) 2017-03-25T20:23:42Z attila_lendvai joined #sbcl 2017-03-25T21:03:45Z angavrilov quit (Remote host closed the connection) 2017-03-25T21:07:56Z shka quit (Ping timeout: 256 seconds) 2017-03-25T21:54:39Z daviid quit (Ping timeout: 240 seconds) 2017-03-25T22:06:23Z stassats joined #sbcl 2017-03-25T22:19:47Z milanj joined #sbcl 2017-03-25T22:29:35Z igajsin quit (Ping timeout: 240 seconds) 2017-03-25T22:34:44Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-25T23:04:01Z whiteline quit (Quit: Leaving) 2017-03-25T23:05:07Z wheelsucker quit (Ping timeout: 260 seconds) 2017-03-25T23:08:50Z whiteline joined #sbcl 2017-03-25T23:10:29Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-25T23:43:59Z gingerale quit (Remote host closed the connection) 2017-03-25T23:58:57Z vydd quit 2017-03-26T00:11:11Z whiteline quit (Remote host closed the connection) 2017-03-26T00:28:45Z whiteline joined #sbcl 2017-03-26T00:58:43Z stassats quit (Ping timeout: 268 seconds) 2017-03-26T01:04:59Z daviid joined #sbcl 2017-03-26T01:05:34Z whiteline quit (Remote host closed the connection) 2017-03-26T01:06:51Z whiteline joined #sbcl 2017-03-26T03:08:10Z BitPuffin|osx quit (Ping timeout: 264 seconds) 2017-03-26T03:11:57Z Bike quit (Ping timeout: 260 seconds) 2017-03-26T03:12:47Z Bike joined #sbcl 2017-03-26T03:52:10Z fe[nl]ix quit (Remote host closed the connection) 2017-03-26T03:52:10Z Blkt quit (Read error: Connection reset by peer) 2017-03-26T03:53:30Z Blkt joined #sbcl 2017-03-26T03:53:51Z fe[nl]ix joined #sbcl 2017-03-26T04:05:33Z jack_rabbit joined #sbcl 2017-03-26T04:35:11Z milanj quit (Quit: This computer has gone to sleep) 2017-03-26T04:43:24Z myrkraverk quit (Remote host closed the connection) 2017-03-26T04:44:27Z myrkraverk joined #sbcl 2017-03-26T05:14:09Z rpg_ quit (Ping timeout: 240 seconds) 2017-03-26T05:16:39Z daviid quit (Ping timeout: 240 seconds) 2017-03-26T06:24:35Z igajsin joined #sbcl 2017-03-26T06:38:12Z angavrilov joined #sbcl 2017-03-26T08:26:24Z attila_lendvai joined #sbcl 2017-03-26T08:26:24Z attila_lendvai quit (Changing host) 2017-03-26T08:26:24Z attila_lendvai joined #sbcl 2017-03-26T08:26:56Z Bike quit (Quit: leaving) 2017-03-26T08:27:08Z dougk quit (Ping timeout: 240 seconds) 2017-03-26T08:39:17Z milanj joined #sbcl 2017-03-26T09:01:37Z shka joined #sbcl 2017-03-26T09:41:36Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-26T09:50:51Z gingerale joined #sbcl 2017-03-26T10:28:29Z BitPuffin|osx joined #sbcl 2017-03-26T11:28:03Z vydd joined #sbcl 2017-03-26T11:28:03Z vydd quit (Changing host) 2017-03-26T11:28:03Z vydd joined #sbcl 2017-03-26T11:37:30Z jack_rabbit joined #sbcl 2017-03-26T12:01:36Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-26T12:01:55Z rpg_ joined #sbcl 2017-03-26T12:22:07Z jack_rabbit joined #sbcl 2017-03-26T12:37:40Z sjl joined #sbcl 2017-03-26T12:39:56Z dougk joined #sbcl 2017-03-26T12:44:40Z dougk quit (Read error: Connection reset by peer) 2017-03-26T12:44:55Z dougk joined #sbcl 2017-03-26T12:48:46Z vydd quit 2017-03-26T13:24:18Z milanj quit (Quit: This computer has gone to sleep) 2017-03-26T14:35:21Z milanj joined #sbcl 2017-03-26T14:42:11Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-26T15:31:11Z sjl quit (Read error: Connection reset by peer) 2017-03-26T15:53:17Z stassats joined #sbcl 2017-03-26T16:07:04Z Bike joined #sbcl 2017-03-26T16:53:18Z milanj quit (Quit: This computer has gone to sleep) 2017-03-26T16:55:36Z cromachina quit (Read error: Connection reset by peer) 2017-03-26T16:56:41Z cromachina joined #sbcl 2017-03-26T17:12:27Z rpg_ quit (Quit: Textual IRC Client: www.textualapp.com) 2017-03-26T17:54:56Z milanj joined #sbcl 2017-03-26T17:58:33Z daviid joined #sbcl 2017-03-26T18:15:44Z stassats: allocating a large array immediately triggers a gc, causing it to get scavenged, even though it's still empty 2017-03-26T18:18:42Z stassats: and scavenge treats vectors as heap space, so they can contain non immediate objects 2017-03-26T18:18:55Z stassats: disabling that could catch corruptions earlier 2017-03-26T18:53:00Z rumbler31 joined #sbcl 2017-03-26T19:12:19Z angavrilov quit (Remote host closed the connection) 2017-03-26T19:29:08Z vydd joined #sbcl 2017-03-26T19:30:22Z dougk quit (Ping timeout: 264 seconds) 2017-03-26T19:34:09Z attila_lendvai joined #sbcl 2017-03-26T19:45:20Z blackwolf joined #sbcl 2017-03-26T20:20:45Z sjl joined #sbcl 2017-03-26T20:44:33Z rumbler31 quit (Remote host closed the connection) 2017-03-26T20:55:32Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-26T20:56:33Z attila_lendvai joined #sbcl 2017-03-26T20:57:47Z whiteline quit (Quit: Leaving) 2017-03-26T21:00:12Z whiteline joined #sbcl 2017-03-26T21:00:12Z whiteline_ joined #sbcl 2017-03-26T21:00:49Z whiteline quit (Disconnected by services) 2017-03-26T21:00:57Z whiteline_ is now known as whiteline 2017-03-26T21:01:17Z whiteline_ joined #sbcl 2017-03-26T21:01:33Z whiteline_ quit (Remote host closed the connection) 2017-03-26T21:07:12Z whiteline quit (Remote host closed the connection) 2017-03-26T21:07:42Z whiteline joined #sbcl 2017-03-26T21:10:08Z igajsin quit (Ping timeout: 240 seconds) 2017-03-26T21:12:26Z whiteline quit (Remote host closed the connection) 2017-03-26T21:12:49Z whiteline joined #sbcl 2017-03-26T21:14:31Z shka quit (Ping timeout: 260 seconds) 2017-03-26T21:15:46Z schjetne quit (Ping timeout: 256 seconds) 2017-03-26T21:20:49Z whiteline quit (Remote host closed the connection) 2017-03-26T21:21:12Z whiteline joined #sbcl 2017-03-26T21:23:59Z whiteline quit (Remote host closed the connection) 2017-03-26T21:33:25Z whiteline joined #sbcl 2017-03-26T21:44:00Z gingerale quit (Remote host closed the connection) 2017-03-26T21:55:35Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-26T22:08:02Z attila_lendvai joined #sbcl 2017-03-26T22:26:56Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-26T22:27:07Z rumbler31 joined #sbcl 2017-03-26T22:32:07Z dougk joined #sbcl 2017-03-26T22:40:08Z dougk: stassats: I already have a change for post-freeze that causes scav_boxed to look within the object for pointers instead of "cleverly" returning 1 so that scavenge() picks up the remainder of the vector. 2017-03-26T22:40:20Z dougk: this is a performance win in fact 2017-03-26T22:41:49Z stassats: i changed fixnump to is_lisp_immediate in scavenge, a big difference for simple-vector filled with floats or characters 2017-03-26T22:42:08Z stassats: i'm trying my hand at using AVX2 for scavenging vectors 2017-03-26T22:45:08Z dougk quit (Ping timeout: 260 seconds) 2017-03-26T22:57:40Z dougk joined #sbcl 2017-03-26T22:59:08Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-26T23:02:18Z sjl quit (Read error: Connection reset by peer) 2017-03-26T23:19:55Z stassats: (sb-ext:gc :full t) with a four gigabyte simple-vector: before 3 seconds, after 1.2 seconds 2017-03-26T23:21:42Z stassats: when there are actually pointers, 10.4 seconds vs 10.8 seconds 2017-03-26T23:21:47Z stassats: so, a slight slow down 2017-03-26T23:25:30Z stassats: using SIMD to speed up the GC is a worthy strategy 2017-03-26T23:32:58Z stassats: https://github.com/stassats/sbcl/commit/cef1357daff602169e63014edc3854b83f8bbc7f 2017-03-26T23:33:14Z stassats: the AVX bits are pretty trivial, though not sure how to actually optionally integrate it 2017-03-26T23:36:46Z stassats: i'm leaning towards a build time feature, AVX2 is quite prevalent these days 2017-03-26T23:41:40Z stassats: 74.24% of Steam users have AVX, it does not report AVX2 2017-03-27T00:28:14Z aeth: That's rather low for something that came out in 2011 (6 years now). 2017-03-27T00:28:57Z aeth: https://en.wikipedia.org/wiki/Advanced_Vector_Extensions 2017-03-27T00:29:05Z stassats: 75% is not low... 2017-03-27T00:29:21Z aeth: AVX 2 started in 2015... I suspect it would be much lower in usage. 2017-03-27T00:29:27Z aeth: e.g. no Haswell. 2017-03-27T00:29:41Z aeth: Oh wait, it is in Haswell. 2017-03-27T00:29:53Z stassats: avx2 started in 2015, yet i'm running it on a computer bought in 2013 2017-03-27T00:29:58Z aeth: No pre-Haswell, I guess. 2017-03-27T00:31:00Z aeth: stassats: The article's wrong, then. 2017-03-27T00:31:14Z aeth: It says "first supported by Intel.... in 2015" 2017-03-27T00:31:32Z stassats: well, it does not say that, if you read carefully 2017-03-27T00:32:23Z aeth: It means to say that the first Xeon Phi support was Knights Landing, I think. But it is poorly written. 2017-03-27T00:32:47Z stassats: it's about AVX512 2017-03-27T00:33:12Z aeth: Ah, I see 2017-03-27T00:33:14Z aeth: It's just not bolded. 2017-03-27T00:33:24Z aeth: And is in the paragraph for AVX2. 2017-03-27T00:34:19Z aeth: Still poorly written, but more of my fault for skimming rather than reading carefully. 2017-03-27T00:34:21Z stassats: i think that's enough about wikipedia articles for #sbcl 2017-03-27T00:34:46Z aeth: Anyway, if it's back in 2013, it probably will still support a decent amount. 2017-03-27T00:35:39Z aeth: It will mostly drop support for i5-2500k (or is it i5-2550k?) that so many gamers think is still good enough today, and so will probably be represented a lot on Steam's hardware survey 2017-03-27T00:37:12Z stassats: i don't care who has avx2 or not 2017-03-27T00:37:31Z stassats: it always will be optional 2017-03-27T00:39:42Z aeth: How hard is it to change SBCL's GC? 2017-03-27T00:40:07Z stassats: it requires a text editor 2017-03-27T00:40:43Z aeth: Afaik, it's written entirely in C, in sbcl/src/runtime/ and... that's as far as I know. 2017-03-27T01:06:04Z blackwolf quit (Ping timeout: 260 seconds) 2017-03-27T01:15:30Z axion joined #sbcl 2017-03-27T01:15:38Z vydd quit 2017-03-27T01:17:23Z axion: Hello. I noticed a possible case where the compiler is doing something odd 2017-03-27T01:17:52Z stassats: you just successfully described sbcl 2017-03-27T01:18:09Z axion: line 86 is a duplicate: https://gist.github.com/mfiano/7682d22ffde51b3ce5eb4c54eafa9446 2017-03-27T01:18:19Z axion: I would have thought COMPLEMENT would produce the same code 2017-03-27T01:18:30Z axion: considering REMOVE-IF-NOT is deprecated 2017-03-27T01:26:01Z stassats: remove-if-not is not deprecated 2017-03-27T01:26:45Z stassats: and line 86 is not a duplicate 2017-03-27T01:31:35Z axion: stassats: What does CLHS mean by "The functions delete-if-not and remove-if-not are deprecated. 2017-03-27T01:31:39Z axion: "? 2017-03-27T01:32:53Z axion: Also, if line 86 is not a duplicate, what is it doing differently and why? 2017-03-27T01:33:40Z stassats: axion: it means "ignore me" 2017-03-27T01:33:49Z stassats: axion: why do you need to know? 2017-03-27T01:34:25Z axion: Not me personally. A few lisp hackers in #lispgames were wondering why sbcl is doing this incorrectly. 2017-03-27T01:34:48Z stassats: incorrectly? 2017-03-27T01:34:59Z aeth: axion: I brought this up in #lisp the other day (a week ago? maybe?). I should have poked you at the time. 2017-03-27T01:36:02Z Bike: it would be nice to knwo what all the breaks in sbcl disassemblies do, tho 2017-03-27T01:36:09Z aeth: stassats: It should be an easy optimization to turn (remove-if (complement #'foo) ...) into (remove-if-not #'foo ...) *if* it's possible. When I brought it up in #lisp recently, the people I was talking with didn't know if it was possible. 2017-03-27T01:36:21Z axion: aeth: I didn't notice. psilord was mostly wondering as he just realized too, and so did I. I guess a lot of people have the same question then. 2017-03-27T01:36:40Z stassats: aeth: it's very easy to describe "just optimize this" 2017-03-27T01:36:53Z stassats: that's what all the people who actually never wrote compilers are telling 2017-03-27T01:37:05Z aeth: stassats: I was told that it would be doable if it were "open coded", but I don't know what that means in the context of SBCL's internals. 2017-03-27T01:37:17Z aeth: When I mentioned it in #lisp a while back. 2017-03-27T01:37:46Z stassats: Bike: just follow the labels 2017-03-27T01:37:51Z aeth: stassats: I didn't say it's esay, I said it was either easy or impossible. :-p 2017-03-27T01:38:04Z Bike: that's not true either. 2017-03-27T01:38:20Z axion: psilord had noticed it because he is very experienced with writing compilers. He does this coupled with reverse engineering for a career. 2017-03-27T01:39:14Z rumbler31 quit (Remote host closed the connection) 2017-03-27T01:40:07Z stassats: axion: yet he come to wrong conclusions 2017-03-27T01:40:11Z stassats: came 2017-03-27T01:40:29Z axion: Which were his conclusions and why were they wrong? 2017-03-27T01:41:06Z stassats: whatever you said 2017-03-27T01:41:22Z axion: and the why? 2017-03-27T01:41:58Z stassats: beats me, lack of knowledge of sbcl internalas, probably 2017-03-27T01:42:33Z axion: No, why do YOU call it wrong? I had asked you what the extra line is doing there 2017-03-27T01:43:29Z stassats: breaking things, do you want to do something with it? 2017-03-27T01:43:55Z stassats: i don't want to explains things just to satisfy some idle curiosity 2017-03-27T01:44:17Z axion: Normally when one calls something wrong, they have the answer. Thanks anyway 2017-03-27T01:44:22Z axion left #sbcl 2017-03-27T01:44:37Z aeth: It looks like nyef said it was plausible, followed by: 2017-03-27T01:44:38Z aeth: 2017-03-20T01:25:48Z nyef: For them to be fully equivalent in terms of code generated, I'd expect remove-if-not and complement to be open-coded, and I don't know that they are. 2017-03-27T01:44:42Z stassats: it wasn't me who labeled it being "incorrect" 2017-03-27T01:44:43Z aeth: Based on the public #lisp logs 2017-03-27T01:45:01Z aeth: When I last talked about it. 2017-03-27T01:45:15Z aeth: oooh, apparently it also requires complement to be open-coded? 2017-03-27T01:45:40Z stassats: aeth: no, just use remove-if-not 2017-03-27T01:45:42Z Bike: can't say that makes sense to me 2017-03-27T01:46:02Z aeth: okay, then nyef was wrong. 2017-03-27T01:46:13Z aeth: nyef: poke 2017-03-27T01:46:42Z stassats: there's very little insensitive to optimize COMPLEMENT 2017-03-27T01:46:57Z stassats: just use -if-not variants, the code will be clearer as well 2017-03-27T01:47:27Z stassats: incentive 2017-03-27T01:47:35Z aeth: There are 12 -if-nots, so it wouldn't just be remove-if-not. That just seems to be the popular one to talk about. 2017-03-27T01:47:39Z aeth: http://l1sp.org/search?q=%2Dif%2Dnot 2017-03-27T01:49:16Z aeth: Using foo-if-not seems to be more common than the foo (complement #'bar) style, but the latter style isn't non-existent. (You'd have to ask Xach or someone else who has all of Quicklisp to grep to see how common the two are relatively, but the former is much more common in the packages I have installed.) 2017-03-27T01:50:53Z aeth: Ime talking about it in various Lisp places, it does seem to be a legitimate style issue that divides people, like setq vs. setf or #'(lambda vs (lambda 2017-03-27T01:51:47Z stassats: i don't think it's comparable, some people just believe that deprecated line, even though it's a double lie 2017-03-27T01:52:37Z aeth: -if-notters like me use the "it's probably faster in SBCL based on the different disassembly" argument (I've brought that argument up before, I'm surprised it took so long to get to here), so I can also see why the complementarians would want it optimized in SBCL. 2017-03-27T01:53:36Z stassats: they should just stop writing ugly code 2017-03-27T01:53:39Z aeth: I agree with you that the deprecated line is nonsense, but I don't think complementarians are going to change their minds or style. 2017-03-27T01:54:31Z aeth: And if I use a library written by a complementarian, I'd actually like to have it disassemble as foo-if-not, because it could hurt my own code's performance, too, if I use their libraries. 2017-03-27T01:56:23Z aeth: It does seem to be a style out there that exists (although, as I said before, only someone grepping all of Quicklisp would know roughly how common it is, by grepping "-if (complement" and "-if-not ") 2017-03-27T02:00:39Z stassats: the effect on remove will be negligible 2017-03-27T02:03:19Z aeth: The place where this argument seems to be the deepest is #lispgames, where people are obviously trying to push SBCL to its limits. 2017-03-27T02:03:53Z karswell joined #sbcl 2017-03-27T02:04:08Z stassats: they shouldn't be using REMOVE-IF-NOT then 2017-03-27T02:04:18Z aeth: Personally, I try to write in a way that gives the best SBCL disassembly, so I agree. 2017-03-27T02:07:28Z aeth: But the other side of the argument is that if the optimization is possible and not that hard, then it should be done. And I can see their point if the style is common enough and the optimization is easy enough, if someone does want to make that optimization. Would the SBCL team reject a patch to optimize foo-if (complement or do you just not want to implement it yourself? 2017-03-27T02:07:42Z aeth: I'm not sure if that last part got cut off. 2017-03-27T02:07:47Z stassats: how do you know whether it's hard or not? 2017-03-27T02:08:04Z aeth: The last sentence (in case it was cut off) was: Would the SBCL team reject a patch to optimize foo-if (complement or do you just not want to implement it yourself? 2017-03-27T02:08:52Z stassats: i don't want to implement it, and if the patch does it in a reasonable fashion then it will be accepted 2017-03-27T02:08:58Z aeth: stassats: The only way to know whether something is hard or not is to implement it, so I can't say that I know whether it is hard or not. 2017-03-27T02:14:17Z aeth: I suppose someone who implemented something similar in SBCL could know the probable difficulty. 2017-03-27T02:21:47Z rumbler31 joined #sbcl 2017-03-27T02:28:05Z rumbler31 quit (Remote host closed the connection) 2017-03-27T02:35:56Z stassats quit (Ping timeout: 260 seconds) 2017-03-27T04:12:35Z daviid quit (Ping timeout: 240 seconds) 2017-03-27T04:23:11Z shka joined #sbcl 2017-03-27T04:28:35Z rumbler31 joined #sbcl 2017-03-27T04:32:40Z rumbler31 quit (Ping timeout: 246 seconds) 2017-03-27T05:29:35Z shka quit (Ping timeout: 240 seconds) 2017-03-27T05:56:08Z dougk quit (Ping timeout: 240 seconds) 2017-03-27T06:06:10Z scymtym quit (Ping timeout: 256 seconds) 2017-03-27T06:19:49Z milanj quit (Quit: This computer has gone to sleep) 2017-03-27T06:32:39Z igajsin joined #sbcl 2017-03-27T07:16:09Z attila_lendvai joined #sbcl 2017-03-27T07:16:09Z attila_lendvai quit (Changing host) 2017-03-27T07:16:09Z attila_lendvai joined #sbcl 2017-03-27T07:51:49Z scymtym joined #sbcl 2017-03-27T08:09:52Z attila_lendvai quit (Disconnected by services) 2017-03-27T08:09:52Z attila_lendvai1 joined #sbcl 2017-03-27T08:09:52Z attila_lendvai1 quit (Changing host) 2017-03-27T08:09:52Z attila_lendvai1 joined #sbcl 2017-03-27T08:12:43Z angavrilov joined #sbcl 2017-03-27T08:23:22Z Bike quit (Quit: leaving) 2017-03-27T08:30:28Z salva quit (Remote host closed the connection) 2017-03-27T08:59:22Z gargaml joined #sbcl 2017-03-27T08:59:46Z irsol quit (Remote host closed the connection) 2017-03-27T09:01:10Z irsol joined #sbcl 2017-03-27T09:32:16Z rtmpdavid joined #sbcl 2017-03-27T09:40:41Z attila_lendvai1 quit (Ping timeout: 252 seconds) 2017-03-27T09:41:57Z Xof: seriously? 2017-03-27T09:42:02Z Xof: 105 commits and not one touches NEWS? 2017-03-27T09:42:15Z Xof: looks like I'll earn my release manager salary this month 2017-03-27T09:42:19Z flip214: Ain't no NEWS here when you're gone... 2017-03-27T09:51:54Z milanj joined #sbcl 2017-03-27T10:07:54Z gingerale joined #sbcl 2017-03-27T10:12:43Z daviid joined #sbcl 2017-03-27T10:21:43Z m00natic joined #sbcl 2017-03-27T11:27:47Z m00natic quit (Read error: Connection reset by peer) 2017-03-27T11:28:01Z m00natic joined #sbcl 2017-03-27T11:33:08Z dougk joined #sbcl 2017-03-27T12:29:16Z sjl joined #sbcl 2017-03-27T13:08:05Z daviid quit (Ping timeout: 240 seconds) 2017-03-27T13:23:24Z sjl quit (Read error: Connection reset by peer) 2017-03-27T13:24:28Z dougk quit (Ping timeout: 240 seconds) 2017-03-27T13:47:19Z dougk joined #sbcl 2017-03-27T13:53:18Z cromachina quit (Read error: Connection reset by peer) 2017-03-27T14:02:21Z stassats joined #sbcl 2017-03-27T14:04:01Z dougk quit (Ping timeout: 260 seconds) 2017-03-27T14:10:45Z sjl joined #sbcl 2017-03-27T14:48:34Z daviid joined #sbcl 2017-03-27T14:58:57Z stassats: valid_lisp_pointer_p doesn't validate undefined-function on riscs 2017-03-27T14:59:04Z stassats: strange 2017-03-27T15:08:07Z rszeno joined #sbcl 2017-03-27T15:15:24Z rszeno quit (Quit: Leaving.) 2017-03-27T15:15:33Z dougk joined #sbcl 2017-03-27T15:18:43Z burtons joined #sbcl 2017-03-27T15:23:07Z stassats: search_spaces succeeds, properly_tagged_descriptor_p does not 2017-03-27T15:28:47Z stassats: n-entries is marked as #!-64-bit 2017-03-27T15:28:50Z stassats: hmm 2017-03-27T15:30:01Z stassats: so all arm64 tramps are broken 2017-03-27T15:31:03Z nyef: Wait, is this why undefined-function backtrace tends to be busted on non-x86oids? 2017-03-27T15:34:21Z stassats: it does tend to be 2017-03-27T15:34:25Z stassats: it just got busted 2017-03-27T15:34:37Z stassats: it's more of "this is why non-x86oids are busted" 2017-03-27T15:35:09Z stassats: eh, i my first should've been "it doesn't tend to be" but it works either way 2017-03-27T15:36:44Z stassats: well, ppc can't validate its tramps either 2017-03-27T15:37:22Z stassats: so, i got to figure out two schemes, one for 64-bit and one for non-64-bit 2017-03-27T15:37:33Z stassats: and how to properly make tagged tramps based on that 2017-03-27T15:39:56Z stassats: and figure out a few magic constants 2017-03-27T15:41:50Z nyef: Might have to cover the boxed tramps vs. unboxed tramps as well? 2017-03-27T15:41:58Z nyef: Or is that an orthogonal issue? 2017-03-27T15:42:06Z stassats: dunno 2017-03-27T15:42:18Z stassats: presumably, they weren't expected to get validated before anyhow 2017-03-27T15:43:06Z nyef: Oh, right. Wasn't there some sort of special case in MAKE-LISP-OBJ for this? 2017-03-27T15:43:45Z Bike joined #sbcl 2017-03-27T15:43:56Z stassats: there is... it's not used anymore 2017-03-27T15:46:24Z stassats: that's because it was in assem.S 2017-03-27T15:46:32Z stassats: couldn't validate anything from there 2017-03-27T15:46:55Z nyef: Right, okay. That makes sense. 2017-03-27T15:47:18Z nyef: And then the new stuff isn't tied into the code-object header, so the validator still can't find it. 2017-03-27T15:48:07Z stassats: i still can't figure out the new CODE layout 2017-03-27T15:57:32Z sjl quit (Ping timeout: 258 seconds) 2017-03-27T16:00:30Z stassats: define-assembly-routine builds its own code object 2017-03-27T16:00:39Z stassats: no idea where yet 2017-03-27T16:03:19Z nyef: The code object is put in the .assem-obj files, and then genesis loads them to the read-only space. 2017-03-27T16:03:33Z stassats: well, who puts it into .assem-obj 2017-03-27T16:04:02Z nyef: dump-assembler-routines. 2017-03-27T16:04:18Z nyef: Called from src/assembly/assemfile.lisp, ASSEMBLE-FILE. 2017-03-27T16:04:35Z nyef: And defined in serc/compiler/dump.lisp. 2017-03-27T16:04:46Z nyef: s/serc/src/ 2017-03-27T16:05:48Z nyef: And, of course, since there's a specific FOP for this, genesis gets involved. 2017-03-27T16:05:53Z dougk_: the new code layout just stores offsets to simple-funs so it's one less collection of pointers to fixup. as a special case, the header word contains the first offset 2017-03-27T16:06:53Z stassats: looks like it's fop-assembler-code 2017-03-27T16:06:56Z stassats: in genesis 2017-03-27T16:06:59Z dougk_: and speaking of unhandled edge cases- a GF with a self-contained trampoline isn't enlivened by the program pointer pointing to its trampoline. 2017-03-27T16:08:01Z stassats: reg_CODE is usually involved in that case 2017-03-27T16:09:10Z stassats: i'm close to giving up and just special casing make-lisp-obj like in the good old days 2017-03-27T16:11:24Z stassats: it looks like it doesn't have to be adjusted for each backend, thankfully, just in genesis 2017-03-27T16:12:32Z shka_ joined #sbcl 2017-03-27T16:14:17Z vydd joined #sbcl 2017-03-27T16:19:55Z stassats: but how does genesis handle it for dynamic-space code objects 2017-03-27T16:25:30Z stassats: cold-load-code does something special 2017-03-27T16:33:19Z milanj quit (Quit: This computer has gone to sleep) 2017-03-27T17:03:01Z m00natic quit (Remote host closed the connection) 2017-03-27T17:45:12Z blackwolf joined #sbcl 2017-03-27T17:45:48Z angavrilov quit (Ping timeout: 240 seconds) 2017-03-27T17:57:49Z rpg joined #sbcl 2017-03-27T18:21:06Z milanj joined #sbcl 2017-03-27T18:56:35Z daviid quit (Ping timeout: 240 seconds) 2017-03-27T19:03:00Z dougk__ joined #sbcl 2017-03-27T19:03:49Z dougk quit (Ping timeout: 246 seconds) 2017-03-27T19:14:56Z daviid joined #sbcl 2017-03-27T19:45:37Z scymtym quit (Ping timeout: 258 seconds) 2017-03-27T20:09:25Z shka_ quit (Ping timeout: 260 seconds) 2017-03-27T20:42:24Z scymtym joined #sbcl 2017-03-27T20:45:00Z prxq joined #sbcl 2017-03-27T20:54:05Z igajsin quit (Ping timeout: 240 seconds) 2017-03-27T21:19:06Z gingerale quit (Remote host closed the connection) 2017-03-27T21:36:06Z prxq quit (Remote host closed the connection) 2017-03-27T21:40:13Z blackwolf quit (Quit: ERC (IRC client for Emacs 24.5.1)) 2017-03-27T21:40:44Z dougk_: i would like to rename symbol-header-widetag to symbol-widetag. same for value-cell-header. Does anyone mind? It's more consistent, though there are still a few more. (code, which is arguably very much a "header" for the function, and then of course functions) 2017-03-27T21:40:56Z stassats: code-header? 2017-03-27T21:41:04Z stassats: ok, didn't read fully 2017-03-27T21:41:41Z dougk_: could do them all, but i'm not as disturbed by code-header. i mean, nobody ever wanted weak-pointer-header-widetag, if we were to go the other direction with it 2017-03-27T21:42:15Z stassats: dougk_: what benchmarks did you use to measure GC performance? 2017-03-27T21:42:38Z dougk_: i use make-host-1 in a loop, and my own wrapper instead of call-with-timing. 2017-03-27T21:42:57Z dougk_: it turns out that we introduce more jitter into the result with the attempt to compensate for the measurement overhead 2017-03-27T21:43:26Z dougk_: then i wait for the baseline versus reference to stabilize to about 200 seconds of GC time and three places after the decimals 2017-03-27T21:43:38Z stassats: using AVX2 is a clear win when a large simple-vector never has any pointers 2017-03-27T21:44:06Z stassats: but i'm not sure about all pointy simple-vector, or small simple-vector 2017-03-27T21:44:48Z dougk_: i want to go one further than just changing fixnump() to is_lisp_immediate() in the one place in scavenge that you pointed out. I think vector scavenging should only ever call is_lisp_pointer() and skip other tests and/or assertions 2017-03-27T21:44:52Z stassats: maybe it could just use AVX2 to find the first pointer and then switch to generic scavenging 2017-03-27T21:44:55Z dougk_: that's what the code does that I have 2017-03-27T21:46:15Z stassats: yes, that's what i did, only checking for is_lisp_pointer, but using SIMD to do four tests at a time 2017-03-27T21:46:51Z stassats: all pointers start with #b11 2017-03-27T21:47:47Z dougk_: very nice 2017-03-27T21:49:32Z stassats: that may also be a win for fully boxed instances 2017-03-27T21:51:33Z dougk_: yes, i have that fixed also 2017-03-27T21:52:22Z stassats: i'll wait for your changes then, before adding simd stuff 2017-03-27T21:52:29Z dougk_: would you like a preview? it's a stack of patches fairly well separated for post-commit audit 2017-03-27T21:52:50Z dougk_: but they're ready to go, so either way is fine with me 2017-03-27T21:54:11Z stassats: and i think i found the source of the copyprop issue... it was caused by a change that avoided another infinite loop 2017-03-27T21:54:30Z dougk_: very nice! 2017-03-27T21:54:59Z stassats: err, constraintprop 2017-03-27T21:55:40Z sjl joined #sbcl 2017-03-27T21:55:59Z stassats: yeah, nice, except that constraint-propagate is probably the worst piece of code to reason about 2017-03-27T21:57:03Z dougk_: all my grandmother's grandchildren were here favorite; all code is the worst code to reason about 2017-03-27T22:09:57Z stassats: i can't possible debug constraint propagation by inserting some print statements 2017-03-27T22:10:15Z stassats: there's just too much information generated 2017-03-27T22:10:26Z stassats: and i'm really out of ideas how to approach it effectively 2017-03-27T22:25:57Z edgar-rft quit (Ping timeout: 240 seconds) 2017-03-27T22:26:51Z burtons quit (Ping timeout: 268 seconds) 2017-03-27T22:29:38Z stassats: i wonder what effect gc-card-size has on performance 2017-03-27T22:32:13Z dougk_: i just broke arm64. will fix in a sec. Missing 1 line in backend-parms 2017-03-27T22:36:03Z edgar-rft joined #sbcl 2017-03-27T22:49:54Z stassats: scavenge in is_lisp_pointer needs to handle four cases, it goes through scavtab, branching to inlined scavengers would probably be faster 2017-03-27T22:50:15Z jack_rabbit joined #sbcl 2017-03-27T22:52:00Z dougk_: my thought exactly 2017-03-27T22:52:40Z dougk_: just pushed most of my changes, 1 or 2 more pending. But take a look at scav1(), heap_scavenge(), and scavenge() 2017-03-27T22:53:05Z daviid quit (Ping timeout: 240 seconds) 2017-03-27T22:55:15Z stassats: there should be a scavenge table for which the scav function is "return 1" 2017-03-27T22:56:35Z stassats: i guess that's only scav_immediate, then it's just is_lisp_immediate 2017-03-27T22:56:46Z dougk_: that's the way i see it. 2017-03-27T23:03:19Z stassats: and all widetags start with 1, so any widetag tables can be twice as compact, provided that all pointers and fixnums are discarded 2017-03-27T23:03:41Z stassats: though pointers start with 1 too, so, just fixnums 2017-03-27T23:08:49Z stassats: dougk_: does scavenge need to return n_words? 2017-03-27T23:08:59Z dougk_: yes 2017-03-27T23:09:16Z dougk_: it's how heap_scavenge() advances to the next object 2017-03-27T23:09:39Z stassats: heap_scavenge doesn't call scavenge though 2017-03-27T23:09:56Z stassats: or is scavenge in the scavtab now? 2017-03-27T23:17:58Z dougk_: heap_scavenge()'s last case is object_ptr += scavtab[...] 2017-03-27T23:18:29Z dougk_: oh. i see 2017-03-27T23:18:44Z dougk_: you're asking whether scavenge() needs to, i thought you meant the generalized scavengers, so maybe no then. 2017-03-27T23:19:28Z rpg_ joined #sbcl 2017-03-27T23:19:46Z stassats: not returning something could be a win 2017-03-27T23:19:56Z stassats: an infinitesimal 2017-03-27T23:20:11Z dougk_: i think i changed it to void, and one thing failed 2017-03-27T23:20:22Z rpg quit (Ping timeout: 246 seconds) 2017-03-27T23:20:32Z stassats: it's returned from scav_boxed_macro 2017-03-27T23:20:34Z dougk_: yah, i interpreted your question to mean does the general concept of scavenging need to return n_words, and i think i wondered the same thing. so i tried it, and got a failure 2017-03-27T23:20:46Z edgar-rft quit (Ping timeout: 264 seconds) 2017-03-27T23:21:06Z dougk_: and without that, the marknsweep scavenger has to call the sizer on each object to know how to stop at holes. that's the win there 2017-03-27T23:21:20Z dougk_: don't want to call the sizer 2017-03-27T23:23:00Z stassats: lldb is seriously brain damaged 2017-03-27T23:23:10Z stassats: (the apple debugger) 2017-03-27T23:25:22Z edgar-rft joined #sbcl 2017-03-27T23:25:56Z stassats: making scavenge() void does make it shorter 2017-03-27T23:26:19Z stassats: 9 bytes 2017-03-27T23:27:02Z dougk_: i re-wonder why it can't be made void. Looking at marknsweep, the uses of it don't use its value 2017-03-27T23:27:08Z cromachina joined #sbcl 2017-03-27T23:27:27Z stassats: DEF_SCAV_BOXED 2017-03-27T23:27:42Z stassats: it just needs to save the size instead of tail calling scavenge 2017-03-27T23:27:42Z dougk_: of course 2017-03-27T23:28:05Z dougk_: so just do that one with its own thing maybe. i'd be surprised if making it void is a huge win though. could measure 2017-03-27T23:28:31Z stassats: one stack push and pop pair 2017-03-27T23:28:35Z stassats: that's clang 2017-03-27T23:33:35Z stassats: gcc is more eagerly inlining stuff 2017-03-27T23:39:26Z stassats: there's only 19KB of simple-vectors that do not contain pointers, in bare ./run-sbcl.sh 2017-03-27T23:40:50Z stassats: so i think using SIMD only until the first pointer is found would be a good idea 2017-03-27T23:42:28Z stassats: or a more complicated rule, if the distance between two pointer regions is too short 2017-03-27T23:49:56Z stassats: it's always heap_base = (void*)DYNAMIC_SPACE_START; 2017-03-27T23:50:10Z stassats: so perhaps it should be somehow made constant and not an extern 2017-03-27T23:54:50Z stassats: gcing a large pointy array, 10.155 => 9.393 2017-03-27T23:54:58Z stassats: that's replace heap_base with DYNAMIC_SPACE_START 2017-03-28T00:00:01Z dougk_: i think people imagine a world in which heap_base is actually changeable at runtime though 2017-03-28T00:00:08Z dougk_: at least once at image startup 2017-03-28T00:00:48Z stassats: well, yeah, relocatable cores and so on, but why we wait another decade for it to materialize might as well save some cycles 2017-03-28T00:00:56Z stassats: s/why/while/ 2017-03-28T00:03:14Z stassats: and it's not a magic number, so it'd be easy to put it back in 2017-03-28T00:04:25Z stassats: and DYNAMIC_SPACE_START is already used throughout 2017-03-28T00:15:13Z daviid joined #sbcl 2017-03-28T00:15:22Z stassats: and how to avoid scanning a newly allocated large array 2017-03-28T00:16:57Z rumbler31 joined #sbcl 2017-03-28T00:21:22Z rumbler31 quit (Ping timeout: 264 seconds) 2017-03-28T00:30:22Z whiteline quit (Ping timeout: 264 seconds) 2017-03-28T01:24:16Z stassats quit (Ping timeout: 246 seconds) 2017-03-28T01:30:27Z milanj quit (Quit: This computer has gone to sleep) 2017-03-28T01:30:36Z dougk_: apparently cheneygc is now broken due to my latest round of optimizations 2017-03-28T02:20:46Z whiteline joined #sbcl 2017-03-28T02:40:55Z BitPuffin|osx quit (Ping timeout: 246 seconds) 2017-03-28T02:51:25Z dougk__ quit (Ping timeout: 246 seconds) 2017-03-28T03:00:12Z daviid quit (Ping timeout: 256 seconds) 2017-03-28T03:36:37Z fiddlerwoaroof quit (Ping timeout: 240 seconds) 2017-03-28T03:42:43Z shka_ joined #sbcl 2017-03-28T03:48:39Z fiddlerwoaroof joined #sbcl 2017-03-28T03:49:57Z myrkraverk quit (Read error: Connection reset by peer) 2017-03-28T03:51:02Z myrkraverk joined #sbcl 2017-03-28T04:02:08Z shka_ quit (Ping timeout: 240 seconds) 2017-03-28T04:29:10Z dougk joined #sbcl 2017-03-28T05:06:26Z fiddlerwoaroof quit (Ping timeout: 258 seconds) 2017-03-28T05:12:33Z fiddlerwoaroof joined #sbcl 2017-03-28T05:14:30Z rpg_ quit (Ping timeout: 256 seconds) 2017-03-28T05:31:57Z fiddlerwoaroof quit (Ping timeout: 240 seconds) 2017-03-28T05:38:41Z fiddlerwoaroof joined #sbcl 2017-03-28T05:53:54Z igajsin joined #sbcl 2017-03-28T06:30:28Z scymtym quit (Ping timeout: 240 seconds) 2017-03-28T06:32:41Z angavrilov joined #sbcl 2017-03-28T06:56:09Z milanj joined #sbcl 2017-03-28T07:08:29Z aeth: I was trying to find a higher order function equivalent to this just to see how easy/hard it would be: (defun bar () `(psetf ,@(loop for results in (mapcar (lambda (x) (list (list 'foo 10 x) (list 30 x))) (list 1 2 3)) append results))) 2017-03-28T07:09:06Z aeth: This seems to do it: `(psetf ,@(mapl (lambda (x) (psetf (car x) (list 'foo 10 (car x)) (cdr x) (cons (list 30 (car x)) (cdr x)))) (list 1 2 3))) 2017-03-28T07:09:15Z aeth: Until I put it in a function foo. 2017-03-28T07:09:48Z jdz quit (Ping timeout: 240 seconds) 2017-03-28T07:10:02Z aeth: Then SBCL gets: "Heap exhasuted during garbage collection: 0 bytes available, 16 requested" in *inferior-lisp* (not even in slime, at first I thought it had hung up) 2017-03-28T07:11:44Z aeth: It works as expected in ccl, and when outside of a function in SBCL, and in ecl 2017-03-28T07:13:41Z jdz joined #sbcl 2017-03-28T07:14:24Z vydd quit (Remote host closed the connection) 2017-03-28T07:18:55Z flip214: aeth: do you have a complete paste, please? 2017-03-28T07:21:46Z aeth: I am going to change it a bit to make a bit more sense 2017-03-28T07:22:12Z aeth: Just realized that the generated code would try to call a number and error (even though obviously the generated code isn't going to be evaluated) 2017-03-28T07:22:25Z aeth: so my test list was a bad one 2017-03-28T07:23:46Z aeth: pasting 2017-03-28T07:24:08Z aeth: still waiting on the paste 2017-03-28T07:24:20Z aeth: flip214: http://paste.lisp.org/+7CDZ 2017-03-28T07:26:31Z aeth: ooooh, it doesn't work in ECL but only if it's compiled instead of run in the REPL 2017-03-28T07:27:10Z aeth: CCL has no problem with it, ever. 2017-03-28T07:27:45Z flip214: looking 2017-03-28T07:27:45Z aeth: I'm guessing it's because SBCL doesn't expect the list to be modified within a mapl while it's iterating? 2017-03-28T07:28:44Z aeth: hmm, interesting, now I tried it again and it didn't even work outside of the defun 2017-03-28T07:29:50Z flip214: aeth: well, modifying the current cons' cell CDR will lead to an endless loop 2017-03-28T07:30:06Z flip214: I guess that outside of DEFUN (in sbcl) it's interpreted 2017-03-28T07:30:15Z flip214: therefore the difference 2017-03-28T07:30:50Z flip214: leaving away setting the CDR makes it just work 2017-03-28T07:30:54Z aeth: Is it undefined whether it's an endless loop or not in mapl? It's strange how there's a split 2017-03-28T07:30:56Z flip214: what are you trying to achieve here? 2017-03-28T07:31:04Z flip214: clhs mapl 2017-03-28T07:31:05Z specbot: http://www.lispworks.com/reference/HyperSpec/Body/f_mapc_.htm 2017-03-28T07:31:33Z flip214: http://www.lispworks.com/documentation/HyperSpec/Body/03_f.htm 2017-03-28T07:31:36Z flip214: For list traversal operations, the cdr chain of the list is not allowed to be destructively modified. 2017-03-28T07:31:39Z flip214: aeth: ^^ 2017-03-28T07:31:53Z aeth: (psetf (foo 10 1) (bar 1) (foo 10 2) (bar 2) (foo 10 3) (bar 3)) is the expected result, on *some* Lisps 2017-03-28T07:32:05Z aeth: ah 2017-03-28T07:32:14Z aeth: so they're the ones not conforming, then 2017-03-28T07:32:19Z flip214: well, why not use (loop ... collect )? 2017-03-28T07:32:36Z aeth: actually, it'd be (loop ... append ...) 2017-03-28T07:32:50Z aeth: I was trying to see if I could beat its performance. Where it works, I think it would slightly do so. 2017-03-28T07:32:54Z flip214: aeth: no, (loop ... collect ... collect ...) 2017-03-28T07:32:59Z flip214: and it's simply _undefined_ 2017-03-28T07:34:12Z aeth: Any chance of adding a warning in maplist/mapl/mapcon for when it detects the cdr is being set in a lambda? Or is that too advanced for SBCL's warning system? 2017-03-28T07:34:23Z aeth: Probably can't do it for all functions, but it might be possible for this case, where it's a lambda 2017-03-28T07:35:59Z vydd joined #sbcl 2017-03-28T07:36:15Z flip214: with a high safety setting that might be a nice idea 2017-03-28T07:36:36Z aeth: Although I doubt anyone will stumble upon it unless they're looking for a dozen different ways to set up a psetf/setf (and stuff like that) just to see which is the fastest. 2017-03-28T07:36:43Z aeth: I don't think I've ever seen mapl used. 2017-03-28T07:37:38Z aeth: (Almost everyone will probably look at the options and settle on loop, this was a microbenchmark of no real use.) 2017-03-28T07:38:42Z aeth: I can see why it would infinite loop on some CLs (and in some places!) and not on others. It depends on which cdr it's using, the one at the start or the end. 2017-03-28T07:40:17Z flip214: aeth: http://paste.lisp.org/display/342647#1 2017-03-28T07:40:30Z aeth: ah 2017-03-28T07:40:39Z flip214: the difference is whether the _next_ cons cell is "cached" before the current one is processed or not 2017-03-28T07:41:12Z flip214: you would be able to kill all these implementations by settings (CDR (CDR current)), I guess 2017-03-28T07:42:19Z scymtym joined #sbcl 2017-03-28T07:42:41Z aeth: flip214: my loop is faster 2017-03-28T07:42:51Z aeth: it's a variant of alexandria's mappend 2017-03-28T07:42:57Z aeth: (i.e. map append) 2017-03-28T07:44:01Z aeth: http://paste.lisp.org/+7CDZ/2 2017-03-28T07:44:58Z aeth: flip214: http://paste.lisp.org/+7CDZ/3 2017-03-28T07:45:55Z flip214: well, a few msec immediately calls "measurement error" to me... 2017-03-28T07:45:55Z aeth: Maybe in 14 or so hours I'll see if I can find faster ways in loop or do. This kind of thinking is how I ran into this undefined behavior with mapl 2017-03-28T07:46:06Z flip214: at least a few seconds of time would be useful. 2017-03-28T07:46:14Z vydd quit 2017-03-28T07:46:17Z flip214: but is speed really important for that one? 2017-03-28T07:47:33Z aeth: longer benchmark: http://paste.lisp.org/+7CDZ/4 2017-03-28T07:47:59Z flip214: aeth: http://paste.lisp.org/display/342647#5 2017-03-28T07:48:03Z aeth: It's still faster, but by less. possibly because it conses more 2017-03-28T07:48:15Z flip214: well, the difference isn't that high. 2017-03-28T07:48:40Z flip214: more real-time but less run-time? 2017-03-28T07:48:54Z flip214: perhaps you'd want a (sb-ext:gc :full t) before the benchmarks 2017-03-28T07:49:37Z flip214: if you used SBCL's timing function (TIME) it would show processor cycles, which I find to be a nice measurement value 2017-03-28T07:49:46Z aeth: flip214: I got about the same: http://paste.lisp.org/+7CDZ/6 2017-03-28T07:50:08Z aeth: lowest consing, though 2017-03-28T07:50:21Z flip214: well, BYTES-CONSED is smaller 2017-03-28T07:50:26Z aeth: running the full GC puts it at 1.708 2017-03-28T07:50:42Z aeth: it's probably all measurement error in timing, so consing might be more relevant 2017-03-28T07:51:24Z Bike quit (Quit: leaving) 2017-03-28T07:52:20Z flip214: fine, then I win! ;P 2017-03-28T07:52:41Z aeth: there's probably a faster way in DO 2017-03-28T07:52:55Z flip214: don't think so. 2017-03-28T07:53:05Z flip214: all the higher-level stuff compiles down to the same primitives, after all 2017-03-28T07:53:47Z aeth: DO is just GO in TAGBODY 2017-03-28T07:54:06Z aeth: it's possible that the higher level stuff than DO add extra things 2017-03-28T07:56:37Z dougk_ quit (Ping timeout: 246 seconds) 2017-03-28T07:56:58Z foom quit (Ping timeout: 246 seconds) 2017-03-28T07:57:05Z flip214: hmmm, for me LOOP is quite a bit faster... that's with safety 3, though 2017-03-28T08:09:15Z dougk_ joined #sbcl 2017-03-28T08:09:16Z foom joined #sbcl 2017-03-28T08:18:20Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-28T08:34:45Z vydd joined #sbcl 2017-03-28T08:35:16Z vydd quit (Client Quit) 2017-03-28T08:54:13Z Xof: suddenly I have an urgent desire to make CL:NIL and :NIL be EQ 2017-03-28T08:55:34Z jack_rabbit joined #sbcl 2017-03-28T08:55:50Z Xof: I wonder if anything breaks 2017-03-28T08:56:52Z Xof: I suppose (defun :nil () 'foo) would probably no longer work 2017-03-28T08:59:25Z Shinmera: Would that even be compliant? 2017-03-28T09:03:08Z jack_rabbit quit (Ping timeout: 260 seconds) 2017-03-28T09:04:29Z cromachina quit (Ping timeout: 268 seconds) 2017-03-28T09:04:49Z cromachina joined #sbcl 2017-03-28T09:19:49Z gingerale joined #sbcl 2017-03-28T09:21:47Z attila_lendvai joined #sbcl 2017-03-28T09:37:41Z DeadTrickster quit (Ping timeout: 260 seconds) 2017-03-28T09:50:12Z DeadTrickster joined #sbcl 2017-03-28T09:52:01Z salva0 joined #sbcl 2017-03-28T10:00:56Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-28T10:10:06Z m00natic joined #sbcl 2017-03-28T10:35:26Z attila_lendvai joined #sbcl 2017-03-28T10:56:29Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-28T11:04:02Z m00natic quit (Remote host closed the connection) 2017-03-28T11:12:44Z m00natic joined #sbcl 2017-03-28T11:13:45Z milanj_ joined #sbcl 2017-03-28T11:15:28Z milanj quit (Ping timeout: 256 seconds) 2017-03-28T11:21:11Z BitPuffin|osx joined #sbcl 2017-03-28T12:01:35Z rpg joined #sbcl 2017-03-28T12:06:11Z scymtym quit (Ping timeout: 258 seconds) 2017-03-28T12:06:53Z scymtym joined #sbcl 2017-03-28T12:11:24Z stassats joined #sbcl 2017-03-28T12:16:53Z stassats: after using scavenge() for normal simple-vectors in scav_vector, using AVX2 is not that much of a win 2017-03-28T12:17:21Z stassats: it is a win, but since it slightly slows down full pointer vectors it's probably not worth it 2017-03-28T12:20:28Z sjl quit (Ping timeout: 240 seconds) 2017-03-28T12:35:00Z rpg quit (Ping timeout: 260 seconds) 2017-03-28T12:35:48Z phoe_ joined #sbcl 2017-03-28T12:36:09Z phoe_: http://www.sbcl.org/porting.html <- the link on this page about cross-compilation is broken 2017-03-28T12:36:36Z stassats: "ok" 2017-03-28T12:39:02Z phoe_: I wonder if I can cross-compile modern SBCL for freebsd on sparc64 - I wonder if that information from this page present in some other places 2017-03-28T12:39:21Z stassats: you can 2017-03-28T12:40:00Z stassats: you'd have to port o sparc64 first 2017-03-28T12:40:02Z stassats: to 2017-03-28T12:53:55Z daviid joined #sbcl 2017-03-28T13:08:57Z scymtym quit (Remote host closed the connection) 2017-03-28T13:09:01Z scymtym_ joined #sbcl 2017-03-28T13:16:35Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-28T13:21:28Z attila_lendvai joined #sbcl 2017-03-28T13:21:28Z attila_lendvai quit (Changing host) 2017-03-28T13:21:28Z attila_lendvai joined #sbcl 2017-03-28T13:33:54Z DeadTrickster joined #sbcl 2017-03-28T13:38:52Z cromachina quit (Read error: Connection reset by peer) 2017-03-28T13:41:07Z attila_lendvai quit (Quit: Leaving.) 2017-03-28T13:41:53Z attila_lendvai joined #sbcl 2017-03-28T13:41:53Z attila_lendvai quit (Changing host) 2017-03-28T13:41:53Z attila_lendvai joined #sbcl 2017-03-28T13:53:23Z dougk quit (Read error: Connection reset by peer) 2017-03-28T13:53:58Z dougk joined #sbcl 2017-03-28T14:00:38Z dougk quit (Read error: Connection reset by peer) 2017-03-28T14:01:16Z dougk joined #sbcl 2017-03-28T14:05:03Z burtons joined #sbcl 2017-03-28T14:10:05Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-28T14:15:08Z rpg joined #sbcl 2017-03-28T14:16:57Z Bike joined #sbcl 2017-03-28T14:18:07Z sjl joined #sbcl 2017-03-28T14:31:51Z scymtym_ quit (Ping timeout: 258 seconds) 2017-03-28T14:41:44Z milanj_ quit (Quit: This computer has gone to sleep) 2017-03-28T15:22:33Z dougk quit (Ping timeout: 260 seconds) 2017-03-28T15:56:38Z phoe_ quit (Quit: Page closed) 2017-03-28T16:00:39Z milanj_ joined #sbcl 2017-03-28T16:07:05Z burtons quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-28T16:12:40Z shka_ joined #sbcl 2017-03-28T16:42:22Z dougk joined #sbcl 2017-03-28T16:45:03Z scymtym joined #sbcl 2017-03-28T16:49:08Z sjl quit (Ping timeout: 240 seconds) 2017-03-28T17:04:50Z m00natic quit (Remote host closed the connection) 2017-03-28T17:06:10Z burtons joined #sbcl 2017-03-28T17:21:46Z sjl joined #sbcl 2017-03-28T17:48:07Z blackwolf joined #sbcl 2017-03-28T18:01:19Z vydd joined #sbcl 2017-03-28T18:36:38Z edgar-rft quit (Quit: edgar-rft) 2017-03-28T19:17:40Z dougk_: gc looks about 17% faster relative to feb 8th, and all-inclusive bottom line performance of iterated make-host-1 test is 9% faster 2017-03-28T19:18:28Z stassats: dougk_: does that include my scav_vector change? 2017-03-28T19:18:35Z dougk_: yes 2017-03-28T19:18:46Z stassats: ok, nice 2017-03-28T19:22:34Z stassats: i wonder if first comparison in find_page_index could be safely removed 2017-03-28T19:22:49Z stassats: probably not without invoking undefined C behavior 2017-03-28T19:23:23Z dougk: wouldn't unsigned arithmetic be fine and not ub ? 2017-03-28T19:23:47Z dougk: but i guess subtraction of pointers not related, so yah. 2017-03-28T19:24:08Z stassats: it does provide a slight speed up 2017-03-28T19:25:25Z stassats: 3.5% on my test case 2017-03-28T19:26:53Z stassats: can gcc avoid loading page_table_pages in a loop? 2017-03-28T19:28:17Z dougk: probably not allowed to elide reload if there's a call 2017-03-28T19:30:29Z stassats: it would have to be spilled anyway, but that would probably give better cache locality 2017-03-28T19:32:43Z stassats: i replaced it with a constant and that only gives 2.5% 2017-03-28T19:33:54Z stassats: coupled with getting rid of the first branch, that gives 7.5% 2017-03-28T19:53:50Z stassats: 7% is also what i get with replacing the scavtab lookup in FIX_POINTER 2017-03-28T19:55:57Z dougk_: looks like scav_boxed/short_boxed/tiny_boxed should all skip over the word at 'where' for 1 fewer loop iteration 2017-03-28T19:56:29Z stassats: right 2017-03-28T19:57:03Z dougk_: i'm still trying to debug cheneygc though. even after replacing the obviously broken scavenge() calls that need to become heap_scavenge(), it still gives me at least 2 different lossage errors 2017-03-28T20:05:32Z milanj__ joined #sbcl 2017-03-28T20:07:46Z milanj_ quit (Ping timeout: 246 seconds) 2017-03-28T20:42:33Z stassats: it's a bit strange that the garbage collector is invoked after allocating 2017-03-28T20:42:57Z stassats: not only might there be no space, but the newly allocated object has to be scavenged yet it has no pointers 2017-03-28T20:44:25Z stassats: and it will get promoted, the fill after that will punch through memory protection, etc. etc. 2017-03-28T20:44:59Z stassats: well, on x86oids it's pinned down, but still 2017-03-28T20:48:22Z igajsin quit (Ping timeout: 264 seconds) 2017-03-28T21:07:20Z dougk_: there's no page protection in gen0 2017-03-28T21:08:12Z dougk_: if there were maybe we could avoid some work 2017-03-28T21:11:13Z gingerale quit (Remote host closed the connection) 2017-03-28T21:13:29Z mood_ joined #sbcl 2017-03-28T21:15:06Z slyrus joined #sbcl 2017-03-28T21:19:31Z slyrus: yay! another MacOS release, another round of SBCL bustage :( 2017-03-28T21:26:51Z angavrilov quit (Remote host closed the connection) 2017-03-28T21:33:38Z stassats: slyrus: just use the latest and greatest macOS 2017-03-28T21:33:53Z slyrus: oh, I am 2017-03-28T21:34:59Z stassats: dougk_: wouldn't the object that caused GC be promoted? 2017-03-28T21:35:13Z stassats: at least on precise backends 2017-03-28T21:36:06Z slyrus: anyway, the symptom(s) are 1: older SBCLs (SBCL 1.3.8-1.3.1 at least. maybe more on both sides) hang at make-target-2 launching SBCL 2017-03-28T21:36:13Z dougk: in theory we only promote after a number of cycles, which is 2, but in practice half the surviving objects that were created on a cycle go along for the ride, since we don't track "survived gc cycles" per page 2017-03-28T21:36:19Z slyrus: and 2: memory faults in the opticl-test suite 2017-03-28T21:36:21Z stassats: slyrus: i use sbcl on macOS for 80% of the time, it works great 2017-03-28T21:36:32Z slyrus: stassats: new MacOS? 2017-03-28T21:36:42Z stassats: ProductVersion: 10.12.4 2017-03-28T21:36:42Z slyrus: 10.12.4? 2017-03-28T21:36:46Z slyrus: hmm... 2017-03-28T21:36:54Z slyrus: have you tried building older SBCL on it? 2017-03-28T21:37:12Z slyrus: and if you get a chance, could you run the opticl asdf test for me? 2017-03-28T21:38:22Z mood_ quit (Quit: Gone.) 2017-03-28T21:38:33Z mood_ joined #sbcl 2017-03-28T21:45:39Z slyrus: 1.3.10 is good... 2017-03-28T21:47:38Z stassats: slyrus: building old sbcls with the current sbcl? 2017-03-28T21:48:18Z slyrus: yes, but it's in make-target-2 where it fails, so presumably that's the old sbcl failing at that point 2017-03-28T21:48:41Z slyrus: 1.3.9 hangs in make-target-2 2017-03-28T21:48:52Z rpg quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-28T21:49:06Z stassats: Did 123 checks. 2017-03-28T21:49:06Z stassats: Pass: 121 (98%) 2017-03-28T21:49:07Z stassats: Skip: 0 ( 0%) 2017-03-28T21:49:08Z stassats: Fail: 2 ( 1%) 2017-03-28T21:49:10Z stassats: that's opticl 2017-03-28T21:49:15Z dougk_: can you try just "src/runtime/sbcl --core output/cold-sbcl.core" and see if it gets a repl? 2017-03-28T21:49:16Z shka_ quit (Ping timeout: 246 seconds) 2017-03-28T21:49:59Z slyrus: stassats: OK. that appears to only fail on MacOS for me, and I don't remember seeing this before. trying to figure out if it was an SBCL change or a MacOS change that caused those two failures. 2017-03-28T21:50:17Z slyrus: dougk_: tried and it hangs 2017-03-28T21:50:29Z dougk_: ok, so it's not actually make-target-2 per se 2017-03-28T21:50:40Z slyrus: ok, fair enough 2017-03-28T21:50:44Z stassats: where does it hang? 2017-03-28T21:50:53Z dougk_: it's just plain dead. can you start it up under lldb and see if you get to call_into_lisp_first_time ? 2017-03-28T21:51:01Z slyrus: ...BSD-style licenses. See the CREDITS and COPYING files in the 2017-03-28T21:51:02Z slyrus: distribution for more information. 2017-03-28T21:52:09Z slyrus: thread #1, queue = 'com.apple.main-thread', stop reason = signal SIGUSR1 2017-03-28T21:52:15Z dougk_: right, just continue 2017-03-28T21:52:39Z slyrus: thread #1, queue = 'com.apple.main-thread', stop reason = signal SIGTTIN 2017-03-28T21:52:42Z slyrus: keep going? 2017-03-28T21:52:52Z dougk_: that seems right, it should have printed a prompt then? 2017-03-28T21:53:03Z slyrus: Process 34896 exited with status = 1 (0x00000001) 2017-03-28T21:53:23Z stassats: i would've interrupted it when it's hanging 2017-03-28T21:53:37Z slyrus: tried that with gdb and got an uninformative stack trace 2017-03-28T21:54:54Z slyrus: I'm not sure that we care why old SBCLs don't build, but it seems important somehow 2017-03-28T21:55:13Z slyrus: and it probably doesn't fix my opticl problem, which is really a retrospectiff problem 2017-03-28T22:00:19Z stassats: so, cl-jpeg has (safety 0) 2017-03-28T22:00:30Z stassats: the actual error is The value # is not of type JPEG::DESCRIPTOR 2017-03-28T22:01:25Z blackwolf quit (Quit: ERC (IRC client for Emacs 24.5.1)) 2017-03-28T22:03:03Z stassats: from read-jpeg-tables 2017-03-28T22:03:12Z stassats: to jpeg::read-marker 2017-03-28T22:03:30Z scymtym: on x86_64, (sleep 7.2993028420866d7) doesn't sleep at all (and would signal a TYPE-ERROR at sufficient safety) because SPLIT-FLOAT in SPLIT-SECONDS-FOR-SLEEP always uses SINGLE-FLOAT literals. i think, i will turn SPLIT-FLOAT into a MACROLET with suitable float literals for each case. any objections? 2017-03-28T22:04:31Z slyrus: stassats: oh, thanks! 2017-03-28T22:05:01Z slyrus: as for the building issue, your changes near 88a92a8129b6559c140e16e8d0e01bb7bba56f6a fix the problem, but it still exists at e34dd0064409fe1c3e9c4360bc05c5e14c404db8 2017-03-28T22:06:22Z attila_lendvai joined #sbcl 2017-03-28T22:06:22Z attila_lendvai quit (Changing host) 2017-03-28T22:06:22Z attila_lendvai joined #sbcl 2017-03-28T22:06:46Z stassats: slyrus: that must be 2fd854b3b20fc95db33012b04d20dc27bdac0f91 2017-03-28T22:07:01Z slyrus: yes! thanks! 2017-03-28T22:07:14Z stassats: i think it was spinning endlessly 2017-03-28T22:08:16Z attila_lendvai quit (Read error: Connection reset by peer) 2017-03-28T22:08:18Z attila_lendvai1 joined #sbcl 2017-03-28T22:08:18Z attila_lendvai1 is now known as attila_lendvai 2017-03-28T22:08:18Z attila_lendvai quit (Changing host) 2017-03-28T22:08:18Z attila_lendvai joined #sbcl 2017-03-28T22:08:19Z mood is now known as mood__ 2017-03-28T22:08:24Z mood_ is now known as mood 2017-03-28T22:08:55Z slyrus: well, that's cute. I turn off safety 0 and I get: failed AVER: (NOT (SB-C::CTRAN-NEXT SB-C::CTRAN)) 2017-03-28T22:08:56Z slyrus: This is probably a bug in SBCL itself. (Alternatively, SBCL 2017-03-28T22:09:33Z slyrus: ... 2017-03-28T22:10:16Z dougk quit (Ping timeout: 246 seconds) 2017-03-28T22:11:25Z burtons quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-28T22:11:29Z stassats: re (sleep 7.2993028420866d7), it gets decoded fine, but the nanoseconds are 33 bit 2017-03-28T22:11:53Z stassats: nanoseconds are supposed to be long, so for 32-bit platforms 33 bits would be a problem 2017-03-28T22:12:06Z stassats: but it sleeps ok on linux-x86, but exits immediately on linux-x86-64 2017-03-28T22:12:53Z stassats: (sb-unix:nanosleep 72993028 (ldb (byte 32 0) 4420865997)) sleeps long enough 2017-03-28T22:13:55Z stassats: but one second in nanoseconds should be 30 bits 2017-03-28T22:18:44Z scymtym: stassats: i think the decoding is wrong. i get (SB-IMPL::SPLIT-SECONDS-FOR-SLEEP 7.2993028420866e7) => 72993028 4420865997 2017-03-28T22:19:45Z scymtym: the second value contradicts (truly-the (integer 0 #.(expt 10 9)) ...) 2017-03-28T22:20:10Z scymtym: and is an illegal argument for nanosleep 2017-03-28T22:21:08Z stassats: it's not about constants 2017-03-28T22:21:20Z scymtym: sorry, should have written (SB-IMPL::SPLIT-SECONDS-FOR-SLEEP 7.2993028420866d7), i.e. DOUBLE-FLOAT value 2017-03-28T22:21:22Z stassats: it's about the truncation not being precise 2017-03-28T22:22:22Z stassats: that actually looks like a miscompilation 2017-03-28T22:23:58Z vydd quit 2017-03-28T22:25:46Z stassats: ok, it's actually because of FLOAT 2017-03-28T22:26:36Z scymtym: compare (<= 0 (sb-impl::%unary-truncate (* (- #1=7.2993028420866d7 (float (sb-impl::%unary-truncate #1#))) 1f9)) 999999999) vs (<= 0 (sb-impl::%unary-truncate (* (- #1=7.2993028420866d7 (float (sb-impl::%unary-truncate #1#) 1d0)) 1d9)) 999999999) 2017-03-28T22:27:21Z stassats: it should be (float whole-seconds seconds) 2017-03-28T22:29:08Z scymtym: yes. but wouldn't it better to make the second argument of float constant and also make 1e9 have the same float format? that's why i suggested the MACROLET solution 2017-03-28T22:29:22Z stassats: no macrolet, it's already inlined 2017-03-28T22:29:37Z stassats: also it's crafted in a very particular way to avoid consing 2017-03-28T22:32:13Z stassats: although things are better now when it comes to boxing 2017-03-28T22:32:52Z stassats: it really should be just done in C 2017-03-28T22:32:56Z scymtym: ok, i will just do (float whole-seconds seconds) and try to write a test, then 2017-03-28T22:33:25Z stassats: nanosleep-float and nanosleep-double, otherwise it can cons anyway 2017-03-28T22:35:27Z stassats: scymtym: hold your changes, i think i should finally do nanosleep-float/double 2017-03-28T22:36:42Z scymtym: stassats: sure 2017-03-28T22:36:51Z scymtym: is that so the %unary-truncate transform can fire? 2017-03-28T22:37:05Z stassats: so that no %unary-truncate is needed 2017-03-28T22:38:37Z stassats: split-seconds-for-sleep would just do ratios 2017-03-28T22:39:01Z stassats: well, it would still do everything, for the SLEEP transform, but for the SLEEP function, just ratios 2017-03-28T22:45:36Z stassats: i see what stopped me the last time i tried 2017-03-28T22:45:41Z stassats: that restarting stuff 2017-03-28T22:45:50Z stassats: that should be translated into C first 2017-03-28T22:55:42Z prxq joined #sbcl 2017-03-28T23:00:24Z prxq quit (Remote host closed the connection) 2017-03-28T23:01:31Z slyrus: OK. opticl problems fixed. one was a cl-jpeg API change opticl didn't fix, the other a cl-jpeg bug. 2017-03-28T23:01:48Z slyrus: i'm not going to worry about the fact that we can't build old SBCL on new macos. 2017-03-28T23:01:58Z stassats: never use (safety 0) 2017-03-28T23:05:37Z salva0 quit (Ping timeout: 268 seconds) 2017-03-28T23:17:53Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-28T23:20:08Z cromachina joined #sbcl 2017-03-28T23:33:45Z aeth: If you set your safety and debug high enough, SBCL should give a warning if the safety is 0. 2017-03-28T23:34:46Z stassats: it shouldn't 2017-03-28T23:35:16Z aeth: so e.g. (if (and (= safety 3) (= debug 3) (= safety 0)) (warn "never use (safety 0)")) 2017-03-28T23:35:48Z stassats: never going to happen 2017-03-28T23:35:57Z aeth: Right, it is never going to happen. 2017-03-28T23:36:30Z stassats: (sb-ext:restrict-compiler-policy 'safety 3) 2017-03-28T23:36:38Z aeth: I think you missed my joke 2017-03-28T23:37:06Z aeth: (and (= foo 3) (= foo 0)) 2017-03-28T23:37:23Z stassats: now you're overplaying your joke 2017-03-28T23:39:10Z aeth: actually, I wonder how sophisticated SBCL is. Would it turn (and (= foo 3) (= foo 0)) to NIL if foo is defined? It doesn't look like it. (defun foo (x) (and (= x 3) (= x 0))) 2017-03-28T23:39:16Z aeth: (disassemble #'foo) 2017-03-28T23:50:21Z stassats: pushed SLEEP, should probably solve a couple of test failures in sleep no-consing 2017-03-29T00:00:07Z dougk joined #sbcl 2017-03-29T00:16:34Z slyrus quit (Ping timeout: 264 seconds) 2017-03-29T00:40:18Z dougk quit (Read error: Connection reset by peer) 2017-03-29T00:40:37Z dougk joined #sbcl 2017-03-29T01:34:34Z stassats quit (Ping timeout: 264 seconds) 2017-03-29T01:39:43Z rpg joined #sbcl 2017-03-29T01:50:16Z sjl quit (Ping timeout: 268 seconds) 2017-03-29T02:09:22Z rpg quit (Ping timeout: 264 seconds) 2017-03-29T02:17:59Z rpg joined #sbcl 2017-03-29T02:32:57Z rpg quit (Read error: Connection reset by peer) 2017-03-29T02:33:07Z rpg_ joined #sbcl 2017-03-29T02:57:06Z BitPuffin|osx quit (Ping timeout: 240 seconds) 2017-03-29T03:10:45Z milanj__ quit (Quit: This computer has gone to sleep) 2017-03-29T03:24:13Z justicefries joined #sbcl 2017-03-29T03:56:44Z shka_ joined #sbcl 2017-03-29T04:05:26Z aeth quit (Quit: ...) 2017-03-29T04:06:32Z aeth joined #sbcl 2017-03-29T04:21:57Z daviid quit (Ping timeout: 240 seconds) 2017-03-29T04:23:24Z burtons joined #sbcl 2017-03-29T04:43:06Z dougk quit (Ping timeout: 240 seconds) 2017-03-29T04:49:53Z shka_ quit (Ping timeout: 260 seconds) 2017-03-29T05:05:12Z fjl_ joined #sbcl 2017-03-29T05:07:36Z fjl quit (Ping timeout: 240 seconds) 2017-03-29T05:14:24Z rpg_ quit (Ping timeout: 256 seconds) 2017-03-29T05:56:05Z luis quit (Ping timeout: 240 seconds) 2017-03-29T05:56:58Z luis joined #sbcl 2017-03-29T06:04:40Z igajsin joined #sbcl 2017-03-29T06:38:35Z scymtym quit (Ping timeout: 240 seconds) 2017-03-29T06:48:21Z angavrilov joined #sbcl 2017-03-29T07:07:57Z milanj__ joined #sbcl 2017-03-29T07:12:30Z mateuszb quit (Read error: Connection reset by peer) 2017-03-29T07:12:58Z mateuszb joined #sbcl 2017-03-29T07:32:00Z salva0 joined #sbcl 2017-03-29T07:42:49Z scymtym joined #sbcl 2017-03-29T07:50:25Z gingerale joined #sbcl 2017-03-29T07:57:35Z salva0 quit (Ping timeout: 258 seconds) 2017-03-29T08:11:37Z salva0 joined #sbcl 2017-03-29T08:12:06Z Bike quit (Quit: sleep tho) 2017-03-29T08:19:49Z edgar-rft joined #sbcl 2017-03-29T08:20:02Z DeadTrickster quit (Remote host closed the connection) 2017-03-29T08:20:17Z attila_lendvai joined #sbcl 2017-03-29T08:20:17Z attila_lendvai quit (Changing host) 2017-03-29T08:20:17Z attila_lendvai joined #sbcl 2017-03-29T09:30:18Z milanj__ quit (Quit: This computer has gone to sleep) 2017-03-29T09:41:47Z daviid joined #sbcl 2017-03-29T10:02:35Z justicefries quit 2017-03-29T10:11:11Z sjl joined #sbcl 2017-03-29T10:17:33Z burtons quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-29T10:38:11Z m00natic joined #sbcl 2017-03-29T10:49:46Z milanj__ joined #sbcl 2017-03-29T10:56:49Z DeadTrickster joined #sbcl 2017-03-29T11:44:07Z m00natic quit (Remote host closed the connection) 2017-03-29T11:45:06Z stassats joined #sbcl 2017-03-29T11:48:36Z dougk joined #sbcl 2017-03-29T11:53:20Z dougk quit (Ping timeout: 256 seconds) 2017-03-29T12:00:24Z rpg joined #sbcl 2017-03-29T12:03:28Z sjl quit (Ping timeout: 240 seconds) 2017-03-29T12:06:15Z BitPuffin|osx joined #sbcl 2017-03-29T12:23:19Z sjl joined #sbcl 2017-03-29T12:29:27Z rpg quit (Ping timeout: 240 seconds) 2017-03-29T12:31:57Z stassats quit (Ping timeout: 240 seconds) 2017-03-29T13:22:08Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-29T13:33:26Z dougk joined #sbcl 2017-03-29T13:45:57Z dougk quit (Read error: Connection reset by peer) 2017-03-29T13:46:14Z dougk joined #sbcl 2017-03-29T13:55:12Z cromachina quit (Read error: Connection reset by peer) 2017-03-29T13:56:00Z dougk quit (Ping timeout: 258 seconds) 2017-03-29T14:02:25Z burtons joined #sbcl 2017-03-29T14:14:36Z rtmpdavid` joined #sbcl 2017-03-29T14:14:36Z rtmpdavid` left #sbcl 2017-03-29T14:18:15Z edgar-rft quit (Quit: edgar-rft) 2017-03-29T14:35:52Z scymtym quit (Ping timeout: 258 seconds) 2017-03-29T14:46:58Z nyef: ... damned lying comments that lie. /-: 2017-03-29T14:49:32Z daviid quit (Ping timeout: 260 seconds) 2017-03-29T14:50:59Z slyrus joined #sbcl 2017-03-29T14:52:38Z nyef: On the upside, this is pure DX/STACK constraint damage, so Eric still hasn't managed to find a hole in the vestigial exit handling. 2017-03-29T14:57:01Z slyrus_ joined #sbcl 2017-03-29T15:04:37Z nyef: Oh, and I'm fairly sure that this is a regression. 2017-03-29T15:13:46Z nyef: Okay, maybe it isn't. 2017-03-29T15:14:14Z nyef takes a step back and prepares to do a proper analysis of WTF is going on. 2017-03-29T15:16:44Z rpg joined #sbcl 2017-03-29T15:16:56Z nyef: Oh, wow. This actually may be a vestigial-exit case after all. 2017-03-29T15:20:31Z stassats joined #sbcl 2017-03-29T15:23:21Z nyef: Ahh... It's the NLX-ENTRY. 2017-03-29T15:23:21Z stassats: what really causes the first gc to happen when starting up? 2017-03-29T15:23:28Z sjl quit (Ping timeout: 246 seconds) 2017-03-29T15:24:45Z nyef: Right, I'm going to need to get the IR1 as it is immediately post-conversion, before the optimizers get at it. /-: 2017-03-29T15:25:10Z stassats: will (with-ir1-namespace (let ((*source-paths* (make-hash-table)) (*undefined-warnings*) (*lexenv* (make-lexenv :default (make-null-lexenv) :policy *policy* :handled-conditions *handled-conditions* :disabled-package-locks *disabled-package-locks*))) (ir1-toplevel '(let ((x 10)) (foo x)) nil nil)))do? 2017-03-29T15:25:39Z nyef: That was quick. 2017-03-29T15:32:10Z stassats: interestingly, grabbing mutexes can uncover write protection 2017-03-29T15:34:58Z stassats: but we can't tell if something is written to an older generation is actually a pointer 2017-03-29T15:35:04Z stassats: a case for software barriers 2017-03-29T15:37:18Z stassats: i need to decipher what all the columns in print_generation_stats mean 2017-03-29T15:42:42Z stassats: do instances ever go to unboxed pages, or is the layout object preventing that? 2017-03-29T15:49:40Z Bike joined #sbcl 2017-03-29T15:50:51Z stassats: upon start up there's a lot of WP triggering, perhaps due to variable sets 2017-03-29T15:51:03Z stassats: it'd be great if all variables were bunched up in a single page 2017-03-29T15:51:21Z stassats: or have unboxed variables 2017-03-29T15:52:03Z stassats: though SYMBOL has name and package 2017-03-29T15:52:32Z stassats: yeah, we need a precise GC with software barriers to avoid that stuff 2017-03-29T15:56:39Z stassats: wait a minute, are large boxed regions not WPed? 2017-03-29T15:58:42Z stassats: ok, they are 2017-03-29T16:00:35Z stassats: uh, but scavenge doesn't check for WP, does it? 2017-03-29T16:03:44Z stassats: i have an 8GB simple-vector, it's in the gen-6, it's WP, I do (setf (aref *x* 0) '(0)), the whole array gets scavenged on gc 2017-03-29T16:04:15Z sjl joined #sbcl 2017-03-29T16:04:21Z milanj__ quit (Quit: This computer has gone to sleep) 2017-03-29T16:16:08Z scymtym joined #sbcl 2017-03-29T16:20:16Z shka_ joined #sbcl 2017-03-29T16:26:22Z stassats: the whole contiguous page region is treated as !WP if at least one page is !WP 2017-03-29T16:29:42Z stassats: so, i do (setf (aref *x* 0) '(0)), then each time i do (loop repeat 10000 do (eval '(make-list 10000))) the whole array gets scavenged, it never gets re-protected 2017-03-29T16:29:56Z stassats: is that the time to conclude that our GC is bonkers? 2017-03-29T16:31:56Z daviid joined #sbcl 2017-03-29T16:33:53Z dougk joined #sbcl 2017-03-29T16:37:28Z slyrus_ quit (Ping timeout: 240 seconds) 2017-03-29T16:39:58Z jdz quit (Ping timeout: 264 seconds) 2017-03-29T16:40:08Z jdz joined #sbcl 2017-03-29T16:48:33Z stassats: but it gets reprotected after (setf (aref *x* 0) 1) 2017-03-29T16:51:35Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-29T16:54:23Z myrkraverk quit (Remote host closed the connection) 2017-03-29T16:56:06Z nyef: stassats: ... because one is a pointer and the other an immediate? 2017-03-29T16:56:27Z nyef: At least as far as the reprotection goes, that is. 2017-03-29T16:56:39Z stassats: nyef: yes, but that pointer never gets promoted 2017-03-29T16:56:49Z stassats: so the large array gets stuck unprotected 2017-03-29T16:57:32Z DeadTrickster joined #sbcl 2017-03-29T16:57:46Z nyef: Sounds like a case for remembered sets? 2017-03-29T16:58:36Z stassats: well, (loop repeat 10000 do (eval '(make-list 10000))) only conses in the 0 gen, which gets cleared enough to not require '(0) promotion 2017-03-29T16:58:53Z stassats: so it sits in gen 1 2017-03-29T16:59:22Z stassats: and *X* is in gen-5, 5>1, no protection, and endless scavenging of all array pages 2017-03-29T17:03:27Z stassats: the number of gen1 pages slowly grows, so it could get reprotected sometime this century 2017-03-29T17:04:47Z stassats: and doing (setf *x* nil) wouldn't help either 2017-03-29T17:06:03Z stassats: scavenging only the unprotected pages will help, but still not ideal 2017-03-29T17:08:26Z myrkraverk joined #sbcl 2017-03-29T17:10:09Z nyef: For the pages with "young pointers", keep a record of where the young pointers are, to within some precision, and re-protect the page. Use the record to determine where to scavenge if the page is still protected. 2017-03-29T17:11:14Z nyef: If the page has been deprotected, rebuild the record. Or notice that there are no young pointers anymore, discard the record, and reprotect the page. 2017-03-29T17:11:32Z stassats: why do we need seven generations? 2017-03-29T17:11:50Z nyef: No idea. 2017-03-29T17:12:06Z stassats: i feel like four could do 2017-03-29T17:12:55Z stassats: but i'm afraid gencgc can't be patched up 2017-03-29T17:13:41Z nyef: Too large, to fragile, and to WTF, huh? 2017-03-29T17:13:47Z nyef: s/to/too/g 2017-03-29T17:14:35Z stassats: i think it's only ever reasonable if you never write into old generations and don't get much garbage promoted 2017-03-29T17:16:15Z stassats: or if you get a lot of garbage promoted so that it can do full gc 2017-03-29T17:16:36Z stassats: but i think it's hopeless 2017-03-29T17:17:22Z stassats: and really what's holding SBCL back 2017-03-29T17:17:35Z gargaml quit (Quit: WeeChat 1.7) 2017-03-29T17:17:57Z nyef: Certainly ONE of the things holding SBCL back. 2017-03-29T17:18:10Z stassats: especially now that you can have lots of RAM, and i just paralyzed it with a 8GB vector 2017-03-29T17:18:14Z flip214: the LAST thing before SBCL-on-the-desktop becomes true! 2017-03-29T17:18:41Z stassats: flip214: it's actually ok the desktop, but not so much on servers with lots of RAM 2017-03-29T17:18:42Z flip214: yeah, I've seen similar things.... using SBCL with big objects is a pain, at least in some circumstances 2017-03-29T17:18:53Z flip214: stassats: well, depends on the desktop, right? 2017-03-29T17:19:10Z stassats: not really 2017-03-29T17:19:14Z flip214: I wanted to have 10GB SBCL for an 8GB vector or so, some time ago.... and failed. 2017-03-29T17:19:42Z nyef: Had an issue in production a few years back. Not even big objects, just huge numbers of threads. The system would take so long stopping and restarting the world that it couldn't handle HTTP requests in a timely fashion once it hit some (low!) number of concurrent connections. 2017-03-29T17:20:31Z flip214: well, for that usecase the apache model (multiple processes, multiple threads each) might be a good idea 2017-03-29T17:20:49Z stassats: i think it's time to hit the literature on the state of the art, what does JVM do? 2017-03-29T17:21:25Z flip214: "the wrong thing" ;) I read about similar problems with that one a few months ago, too 2017-03-29T17:22:36Z stassats: i'd be ok with latency spikes, but the situation i demonstrated basically poisons the heap indefinitely 2017-03-29T17:22:52Z stassats: so, it's realtime, since it's guaranteed to always be slow 2017-03-29T17:23:08Z nyef: I suspect that, no matter what we come up with for GC, there's always going to be some pathological case that brings it to its knees. 2017-03-29T17:24:02Z stassats: right now every other case manages to do that 2017-03-29T17:25:58Z flip214: will a (GC :full t) promote everything that's still needed to gen7? 2017-03-29T17:26:10Z flip214: just thinking about my RPi doing PWM via SBCL... 2017-03-29T17:26:17Z stassats: flip214: yes, gc :full resolves that issue 2017-03-29T17:26:23Z stassats: but it's not useful 2017-03-29T17:26:35Z flip214: as long as the main loop isn't consing, will a (GC) every now and then clean up gen0 and so on? 2017-03-29T17:27:03Z stassats: what for? 2017-03-29T17:27:04Z flip214: well, it's useful for me, to define a boundary between long-lived data and the (nearly) dynamic extent one 2017-03-29T17:27:04Z jackdaniel quit (Remote host closed the connection) 2017-03-29T17:27:57Z flip214: well, processing a new input pattern is not always non-consing. but that happens only so often, whereas the main loop runs with IIRC 50Hz or so... 2017-03-29T17:28:22Z flip214: and having that interrupted every now and then would be visible 2017-03-29T17:30:28Z stassats: so you don't actually know what's going on and are already trying to solve it with a hammer? 2017-03-29T17:32:25Z jackdaniel joined #sbcl 2017-03-29T17:33:40Z dougk quit (Ping timeout: 246 seconds) 2017-03-29T17:34:30Z nyef: "When all you have is a hammer, every problem looks like a thumb." 2017-03-29T17:42:21Z milanj joined #sbcl 2017-03-29T18:07:27Z sjl quit (Ping timeout: 240 seconds) 2017-03-29T18:15:24Z slyrus_ joined #sbcl 2017-03-29T18:27:16Z sjl joined #sbcl 2017-03-29T18:43:27Z sjl quit (Ping timeout: 240 seconds) 2017-03-29T18:43:41Z sjl__ joined #sbcl 2017-03-29T19:10:14Z slyrus_ quit (Ping timeout: 256 seconds) 2017-03-29T19:12:35Z stassats: huh, does update_page_write_prot scan the whole vector for pointers again? 2017-03-29T19:13:06Z stassats: so for each gc it gets scavenged then scanned for pointers to younger generations 2017-03-29T19:13:42Z stassats: /* Skip if it's already write-protected, pinned, or unboxed */ 2017-03-29T19:13:47Z stassats: ok, that's better 2017-03-29T19:32:16Z sjl__ quit (Read error: Connection reset by peer) 2017-03-29T19:34:31Z nyef: ... The (GO TAG6) is getting snapped out, somehow, causing the ENTRY for TAG6 to not have any associated EXITs. But this is without going through the M-D-V-E path. 2017-03-29T19:35:20Z nyef: The GO IR1-TRANSLATOR definitely creates an EXIT node here. 2017-03-29T19:35:42Z nyef: ... No *value semantics*. 2017-03-29T19:36:30Z nyef: Well played, Eric, well played. 2017-03-29T19:37:03Z nyef: Now I "just" need to figure out how to make this work anyway. 2017-03-29T19:57:31Z daviid quit (Ping timeout: 246 seconds) 2017-03-29T20:06:25Z nyef: So, can I make the vestigial exit for TAGBODY/GO be required? (-:< 2017-03-29T20:06:52Z stassats: i'm afraid only you can answer that question 2017-03-29T20:07:35Z slyrus_ joined #sbcl 2017-03-29T20:08:30Z stassats: what realistically can go into large objects? 2017-03-29T20:08:36Z stassats: vectors and bignums? 2017-03-29T20:09:03Z stassats: and bignums are unboxed 2017-03-29T20:09:19Z nyef: Possibly sufficiently-large CODE-OBJECTs? 2017-03-29T20:10:45Z stassats: nyef: unlikely to be modified 2017-03-29T20:10:59Z stassats: but yeah 2017-03-29T20:11:17Z stassats: i'm trying to come up with a way to avoid scavenging WP parts of large objects 2017-03-29T20:11:42Z stassats: it'd be easier if a large object region contained only a single large object 2017-03-29T20:11:44Z nyef: Unlikely to be modified only because breakpoints are broken and we don't yet have anything even more clever to replace them with. 2017-03-29T20:12:10Z nyef: Umm... I thought that they *did* only contain a single large object? 2017-03-29T20:12:25Z stassats: i need to prove that allegation 2017-03-29T20:12:35Z nyef: ... And possibly some dead conses that get excised via m_a_l_o()? 2017-03-29T20:13:08Z nyef: It's pretty baked into how large-object regions get allocated in the first place. 2017-03-29T20:13:54Z stassats: so if that's the case, i can visit at most 2 pages 2017-03-29T20:14:04Z stassats: the header page and the unprotected page 2017-03-29T20:14:13Z edgar-rft joined #sbcl 2017-03-29T20:14:17Z stassats: well, or how many unprotected pages there are 2017-03-29T20:14:26Z nyef: Why do you need to visit the header? 2017-03-29T20:14:48Z stassats: if it's a humongous instance, i need to know its layout 2017-03-29T20:14:52Z nyef: Ah. 2017-03-29T20:14:59Z nyef: Fair enough. 2017-03-29T20:15:22Z stassats: or just ignore that case and scan all pages for humongous instances or code objects 2017-03-29T20:15:31Z stassats: so, really, only deal with vectors 2017-03-29T20:15:54Z stassats: and, there are hash-table vectors 2017-03-29T20:17:07Z slyrus_ quit (Ping timeout: 246 seconds) 2017-03-29T20:19:18Z stassats: complicated 2017-03-29T20:19:45Z stassats: i'll start with normal simple vectors and see where that leads 2017-03-29T20:27:29Z aeth: stassats: what do you mean by large objects when you say go into large objects? 2017-03-29T20:27:52Z stassats: aeth: it's an sbcl gc term 2017-03-29T20:28:03Z stassats: large objects are not... small objects 2017-03-29T20:29:16Z nyef: aeth: There's a "large object threshold". Anything above that size is considered a "large object", and handled specially. 2017-03-29T20:29:26Z nyef: Mostly so that it doesn't have to keep getting copied around. 2017-03-29T20:29:27Z aeth: What about a hash-table of structs? I could imagine that easily getting into the thousands in a straightforward way to hold game objects in some #lispgames games. Or components (e.g. a physics struct) if someone's going more for the ECS route. 2017-03-29T20:33:31Z aeth: Games often use struct objects instead of CLOS objects because the objects are simple and the slots are usually things like numbers or numerical vectors (e.g. single-float vecs of size 2 or 3 or 4) 2017-03-29T20:33:52Z aeth: And they'd be repeatedly modified 2017-03-29T20:35:27Z aeth_ joined #sbcl 2017-03-29T20:36:02Z aeth_: My connection to Freenode imploded again. If it was like last time, aeth is about to quit via "netsplit" at some point. 2017-03-29T20:36:05Z aeth quit (Quit: Reconnecting) 2017-03-29T20:39:24Z nyef: aeth_: A "large object" is a single object that takes up some amount of space. Looks to be at least four times the largest of the backend page size, the GC "card" size, and the GC allocation granularity. 2017-03-29T20:39:52Z nyef: So, on x86oids, that's going to be at least 16k for a single object. 2017-03-29T20:39:57Z aeth_: ah 2017-03-29T20:40:04Z pkhuong: aeth_: go full ECS and index into specialised arrays. 2017-03-29T20:40:45Z aeth_: pkhuong: there is no advantage to index into specialized arrays in CL because the arrays are just going to be pointers to those structs 2017-03-29T20:40:45Z aeth_ is now known as aeth 2017-03-29T20:40:45Z pkhuong: what? 2017-03-29T20:40:50Z aeth: The reason ECSes in C++ use arrays is for cache coherency. 2017-03-29T20:40:53Z pkhuong: if you want to SoA, you can SoA. 2017-03-29T20:42:05Z pkhuong: you can also use C arrays like the QPX people. 2017-03-29T20:42:17Z aeth: Afaik, if I make an array that can hold physics component structs, where physics component structs are more than just some type of vector (e.g. if they have more than one type of number in them), it has to be a T array. 2017-03-29T20:42:47Z stassats: nyef: backend page size is actually 32K, not 4K 2017-03-29T20:43:02Z pkhuong: aeth: or you can have a struct of arrays instead of an array of structs. 2017-03-29T20:43:49Z nyef: stassats: I was going by the hardware page size, but you're right. Doesn't invalidate my point, just raises the value to 128k for a single object. 2017-03-29T20:44:08Z nyef: ... I remember having less *address space* than that. /-: 2017-03-29T20:44:36Z aeth: pkhuong: so you're saying there shouldn't be fphysics components at all, but instead one physics component, which turns all of its slots into arrays? 2017-03-29T20:44:42Z aeth: s/fphysics/physics 2017-03-29T20:46:19Z aeth: pkhuong: well, then, I guess we've just constructed a large object for nyef :-) 2017-03-29T20:46:26Z pkhuong: there's no big difference for the coder between arrays of structs and structs of arrays when you always pass indices around. SBCL does much better at SoA instead of AoS. 2017-03-29T20:47:12Z pkhuong: vectorisation also tends to work better with SoA. 2017-03-29T20:50:02Z aeth: I knew there was a simple solution, I just didn't know what 2017-03-29T20:50:07Z aeth: Thanks 2017-03-29T20:50:35Z aeth: Will structs efficiently handle holding vectors of vectors? 2017-03-29T20:50:48Z pkhuong: why would you need vectors of vectors? 2017-03-29T20:51:42Z pkhuong: instead of doing (struct-slot (aref component entity-id)), you (aref (struct-slot component) entity-id). 2017-03-29T20:51:49Z aeth: An entity's velocity is a (simple-array single-float (3)) 2017-03-29T20:53:08Z nyef: Do you need it in that form, or could you use a 2D array with one of the indices being the entity-id? 2017-03-29T20:55:12Z pkhuong: you could go for a 2d array (or do the index computation yourself), or you can, again, permute the indices and represent the entities' velocities with a sequence of 3 (simple-array single-float 1). 2017-03-29T20:56:03Z aeth: nyef: a part of a 2D array is a different type 2017-03-29T20:56:14Z aeth: nyef: and the math libraries hardcode (simple-array single-float (3)) 2017-03-29T20:57:03Z Shinmera: You could keep large matrices and index those instead of vectors. 2017-03-29T20:57:55Z pkhuong: if you're using arrays of (simple-array single-float (3)), I hope it's for convenience, because it's really not the way to get performance. 2017-03-29T20:58:04Z nyef: Hrm... You could mmap() a non-heap page for these, and then do a dead-reckoning calculation to find a given velocity vector... 2017-03-29T20:58:28Z nyef: And this is only plausible because they're unboxed arrays. 2017-03-29T20:59:06Z aeth: pkhuong: Actually, I'm using hash-tables because once performance is killed that way, nothing beats a hash-table anyway. 2017-03-29T20:59:34Z aeth: pkhuong: I have rewritten most of my external math dependencies myself, so I could change the type of the vectors in the vector math, though 2017-03-29T20:59:42Z pkhuong: if you're using libraries that force you to cons up a bunch of small arrays, I hope it's for convenience 2017-03-29T20:59:44Z stassats: (simple-array single-float (3)) has more overhead than data 2017-03-29T21:00:13Z gingerale quit (Remote host closed the connection) 2017-03-29T21:00:50Z aeth: pkhuong: Actually, I've gotten rid of consing in most cases, by emulating the difference between map and map-into. 2017-03-29T21:01:06Z aeth: pkhuong: for any vec+, I can also have a vec+-into! and then consing goes away 2017-03-29T21:01:17Z pkhuong: just working with small arrays is a bad plan. 2017-03-29T21:01:28Z aeth: This is what like 10 different vector math libraries do 2017-03-29T21:02:02Z Shinmera: (mine doesn't) 2017-03-29T21:02:27Z pkhuong: that's what happens when you ask the wrong question. When you do "X" in a loop, make the whole loop fast, not just individual iterations. 2017-03-29T21:02:41Z aeth: Oh, now I see what's going on. The type changes from simple-array once it's displaced, which most 2D arrays will be. 2017-03-29T21:04:07Z stassats: displaced arrays, well, you're clearly not concerned with performance 2017-03-29T21:04:41Z aeth: subseq copies. How does one work on a range of a 2D arrays? 2017-03-29T21:05:02Z aeth: Just pass in the individual values and then replace them? 2017-03-29T21:05:19Z slyrus_ joined #sbcl 2017-03-29T21:06:37Z aeth: Making it 1-D, then, might make more sense, because then it would just be replacing subseqs 2017-03-29T21:07:03Z shka_ quit (Ping timeout: 260 seconds) 2017-03-29T21:09:15Z aeth: A subseq of size 3 can be taken, declared dynamic extent, the vector math operation(s) can be done, and then the original subseq can be overriden with (replace original-mega-array result-vec-3 :start1 offest-index :start2 (+ offset-index 3)) (or would it be + 2? depends on the API of replace) 2017-03-29T21:09:48Z aeth: I consider this the worst case, but possibly necessary, way of handling things. I guess it's necessary? 2017-03-29T21:09:54Z stassats: it can't be declared dynamic extent... 2017-03-29T21:10:16Z aeth: If the math is all inlined... 2017-03-29T21:10:45Z stassats: are you just guessing things? 2017-03-29T21:11:14Z aeth: no, I disassemble and check for no allocations. 2017-03-29T21:11:38Z aeth: I just haven't done this subseq and replace scheme yet because I wasn't sure if it was necessary or if there was a simpler way. I have done similar stuff, though. 2017-03-29T21:11:52Z pkhuong: sure, you could have code to materialise a single struct / array at a time, but that'll never be as efficient as working with structs of arrays natively. 2017-03-29T21:12:33Z stassats: aeth: you wouldn't use subseq, but indexes into an array 2017-03-29T21:12:56Z aeth: but if the subseq is declared dynamic-extent, there's no issue with allocation, is there? 2017-03-29T21:13:29Z nyef: Hah! You wish. 2017-03-29T21:14:28Z nyef: DX unboxed arrays only works on x86oids right now, IIRC. 2017-03-29T21:14:41Z stassats: aeth: subseq wouldn't dx, but even if you dx stuff, that's still copying 2017-03-29T21:15:13Z aeth: nyef: My engine only works for x86-64, with SBCL and CCL. 2017-03-29T21:15:27Z nyef: aeth: Not even aarch64? 2017-03-29T21:15:35Z pkhuong: nyef: too slow ;) 2017-03-29T21:16:15Z aeth: This is a game engine. Good luck deploying it on the anti-GC Apple's iOS store. That would leave Android as the only major thing I'm leaving out. 2017-03-29T21:16:57Z aeth: I don't think you'd get acceptable performance for this on mobile, though. 2017-03-29T21:17:20Z stassats: well, not if you're using subseq or displaced arrays 2017-03-29T21:18:19Z aeth: Mobile gaming for the type of games this engine is designed for is basically irrelevant, though. People were complaining about paying a mere $10 for Super Mario Run. 2017-03-29T21:18:20Z pjstirling joined #sbcl 2017-03-29T21:18:26Z stassats: and my phone is faster than my old computers, you're telling me there was no games back then? 2017-03-29T21:19:09Z aeth: You cannot directly compare things between architectures. Also, this would kill battery life. Also, there is no viable business model for games that would be charging around $30 on Android. 2017-03-29T21:19:16Z stassats: i can 2017-03-29T21:20:02Z nyef: Absolutely you can directly compare things between architectures. This is the *heart* of architecture selection. 2017-03-29T21:20:30Z pkhuong: my phone can emulate some of my old computers faster than realtime. I think we can compare some things. 2017-03-29T21:21:02Z aeth: Afaik, the only Lisp game that has been deployed on Android uses ECL, which I cannot support because it already is too slow for what I'm doing. Also, my engine hardcodes OpenGL assumptions, so a large part of the renderer might need to be rewritten to do a mobile port. 2017-03-29T21:21:18Z aeth: It is explicitly designed for x86-64 desktop games. 2017-03-29T21:23:07Z aeth: The only Lisps that are a match for what I'm doing are SBCL and CCL, and SBCL is twice as fast (probably because it tends to produce functions that disassemble to half the size) 2017-03-29T21:23:57Z stassats: that's a strange assertion 2017-03-29T21:24:08Z aeth: which one? 2017-03-29T21:24:20Z stassats: the last one 2017-03-29T21:24:30Z aeth: probably. 2017-03-29T21:25:13Z aeth: that would be assuming x86-64 is understandable. :-p 2017-03-29T21:25:39Z aeth: Anyway, what I'm writing can only work on 64 bit because it makes assumptions about single-float being fast and about the size of the fixnum 2017-03-29T21:25:55Z aeth: So that already rules out a lot of ARM afaik 2017-03-29T21:27:44Z aeth: Imo, Lisps need at least 36 bits. 2017-03-29T21:28:20Z stassats: do they really? 2017-03-29T21:29:04Z aeth: Historical 32 bit Lisps boxed single-float and had a tiny fixnum, right? 2017-03-29T21:29:27Z stassats: "historical" 2017-03-29T21:29:29Z stassats: right... 2017-03-29T21:29:34Z pkhuong: historically, single floats weren't necessarily 32 bits. 2017-03-29T21:29:34Z aeth: I'm not sure how SBCL behaves at 32 bit. 2017-03-29T21:31:00Z aeth: I do make an assumption of a fixnum of at least 60, though. I use logical operations on integers to check to see if an entity has a component. 2017-03-29T21:31:41Z aeth: A component limit of 60 should make sure that the membership sets stay fixnum. 2017-03-29T21:32:04Z aeth: And that ensures that the iteration remains fast. 2017-03-29T21:35:09Z pjstirling: hi, it looks like sbcl is re-binding *print-level* which is making debugging my macro a little tricky, is there a way around that? 2017-03-29T21:36:06Z stassats: pjstirling: be more specific 2017-03-29T21:36:56Z pjstirling: I get a diagnostic which starts printing # for forms before it provides anything useful 2017-03-29T21:37:35Z stassats: just macroexpand your macro 2017-03-29T21:37:58Z stassats: don't wait until compilation 2017-03-29T21:38:40Z igajsin quit (Ping timeout: 246 seconds) 2017-03-29T21:39:40Z aeth: I still don't understand what's wrong with making a temporary vector that's dynamic-extent, doing a bunch of things to it, and then copying its result into the ECS. It doesn't create garbage, and it doesn't look like it adds that much of a performance penalty. 2017-03-29T21:40:12Z edgar-rft quit (Quit: edgar-rft) 2017-03-29T21:41:37Z aeth: I'm not sure if there is an alternative, because if there's a function that's x1 y1 z1 x2 y2 z2 it has to return something like (values x y z) (and not everything is independent on each axis like + or -, e.g. rotation isn't) 2017-03-29T21:42:19Z aeth: And I'm not sure how returning values is necessarily more efficient than replacing the three parts of a temporary dynamic-extent vector within the math function itself. 2017-03-29T21:42:25Z aeth: (the inlined math function) 2017-03-29T21:42:55Z stassats: the suggested way was using indexes 2017-03-29T21:42:57Z stassats: not values 2017-03-29T21:43:54Z aeth: If indices are used, how can complicated operations be done? And it seems like it would become a nightmare to debug if the math functions themselves are directly setting the final result. 2017-03-29T21:44:34Z aeth: By complicated I mean e.g. when the same vector appears more than once. 2017-03-29T21:46:03Z aeth: oh, nevermind, I see what can be done 2017-03-29T21:46:35Z BitPuffin|osx quit (Ping timeout: 240 seconds) 2017-03-29T21:47:24Z aeth: if there's a pure vec+ that allocates a new vec, and a vec+-into! that replaces a vec's contents, there can also be a vec+-into-range! that replaces a size-3 range in any (simple-array single-float (*)) 2017-03-29T21:47:32Z aeth: And the design doesn't need to be significantly changed 2017-03-29T21:48:12Z aeth: and each of the three arguments of the third function could have a range through keyword arguments. 2017-03-29T21:50:25Z aeth: The three functions would be generated together via a macro so the logic would be almost identical, and a macro could transform what appears to be calls to the simple functional one into what are actually calls to one of the latter two, depending on the macro. 2017-03-29T21:50:52Z aeth: So with that macro it can look like it's using vec+, but actually be using vec+-into-range!, which is almost identical, except faster. 2017-03-29T21:51:07Z aeth: stassats, nyef, pkhuong: would that work? 2017-03-29T21:52:28Z stassats: i lost interest 2017-03-29T21:55:53Z aeth: A macro could turn composition of functions like (vec+ a b (vec+ c d)) into a properly ordered list of instructions like (vec+-into! temp-dynamic-extent-vec c d) (vec+-into! temp-dynamic-extent-vec temp-dynamic-extent b) (vec+-into-range! a a temp-dynamic-extent-vec :start1 i :start2 i :start3 0) 2017-03-29T21:59:18Z aeth: (In this case, working left to right would have produced a much more efficient result, though.) 2017-03-29T22:00:16Z stassats: made scavenge_generations more granular for large-simple-vector: after touching the first element of an 8GB vector (time (sb-ext:gc)) was 0.5 seconds => 0.010 seconds 2017-03-29T22:14:01Z angavrilov quit (Remote host closed the connection) 2017-03-29T22:16:42Z stassats: pushed, there's probably not many tests that will exercise this, but it seems simple enough to get it right 2017-03-29T22:27:08Z aeth: pkhuong: by the way, thank you, I've finally figured out how to organize things... your comments were very helpful. Soon, I'll be able to move onto the physics engine because of your help in structuring the data properly. 2017-03-29T22:32:36Z pjstirling: would it be difficult to give sbcl functions that took and/or returned raw alien values? I would be willing to live with redefinition restrictions similar to those for structures 2017-03-29T22:33:52Z stassats: pjstirling: you mean without boxing? 2017-03-29T22:33:56Z pjstirling: yes 2017-03-29T22:34:07Z stassats: it would be difficult 2017-03-29T22:44:08Z daviid joined #sbcl 2017-03-29T22:46:12Z stassats: no transport function for object 0x10027b241f (widetag 0xc3) 2017-03-29T22:46:24Z stassats: good, something does exercise it 2017-03-29T22:49:21Z aeth: What's better in SBCL? (= foo 1) (not (zerop foo)) (plusp foo) or something else, where foo is an element of a bit array? 2017-03-29T22:49:38Z aeth: zerop is pretty easy for the 0 one, I'm just not sure what SBCL is best at for the 1 one. 2017-03-29T22:51:02Z stassats: measure it 2017-03-29T22:51:40Z dougk joined #sbcl 2017-03-29T22:56:34Z dougk quit (Read error: Connection reset by peer) 2017-03-29T22:57:13Z dougk joined #sbcl 2017-03-29T22:59:36Z aeth: The differences are too small to measure properly, but based on the disassembly, probably one of the first two. 2017-03-29T23:00:01Z stassats: if the differences is too small then why are you asking? 2017-03-29T23:01:40Z aeth: Because it will probably be run over a trillion times. 2017-03-29T23:02:10Z stassats: then run it a trillion times and measure the difference 2017-03-29T23:12:35Z aeth: heap exhausted :p 2017-03-29T23:12:59Z stassats: then you're measuring something else 2017-03-29T23:13:39Z aeth: yes 2017-03-29T23:14:18Z aeth: ooooh, "Heap exhausted, game over." 2017-03-29T23:14:27Z aeth: In the *inferior-lisp*, that's a new one 2017-03-29T23:14:43Z aeth: I guess trivial-benchmark itself conses or something. 2017-03-29T23:15:39Z Bike: it accumulates garbage just from testing some equalities? maybe it's not trivial after all 2017-03-29T23:16:37Z stassats: why would you use trivial-benchmark for testing that? 2017-03-29T23:16:52Z aeth: it averages the time 2017-03-29T23:18:12Z attila_lendvai joined #sbcl 2017-03-29T23:18:12Z attila_lendvai quit (Changing host) 2017-03-29T23:18:12Z attila_lendvai joined #sbcl 2017-03-29T23:18:21Z aeth: and it does track consing, too 2017-03-29T23:26:32Z aeth: I see what you're saying though. When I'm just comparing (onep-one 1) vs (onep-zero 1) and (onep-one 0) vs (onep-zero 0) (the two versions) there's no consing either so a simple dotimes would work 2017-03-29T23:26:41Z aeth: It looks like they're basically equivalent 2017-03-29T23:31:27Z pkhuong: or maybe the comparison is optimised out 2017-03-29T23:34:00Z aeth: It's not, I disassembled them first. 2017-03-29T23:35:48Z aeth: and it adds or removes time based on how long the iteration is 2017-03-29T23:36:03Z aeth: which suggests it doesn't just optimize away the whole iteration that has no side effect and returns nothing (even though it could!) 2017-03-29T23:37:43Z stassats: can you paste your function? now i'm convinced it's optimized away 2017-03-29T23:38:25Z aeth: (defun onep-zero (x) (declare (bit x)) (not (zerop x))) (defun onep-one (x) (declare (bit x)) (= x 1)) 2017-03-29T23:39:22Z stassats: no dotimes here, ok then 2017-03-29T23:39:46Z aeth: right, it can't be optimized away because they're not inlined and they're just function calls 2017-03-29T23:39:58Z aeth: So SBCL can't know if one is being redefined to have side effects... probably 2017-03-29T23:40:31Z aeth: (well, I mean, when placed in dotimes they're just function calls) 2017-03-29T23:40:43Z stassats: congratulations, you've wasted on this more time than you could have possibly saved if one of them were faster 2017-03-29T23:40:46Z aeth: e.g. (defun foo () (dotimes (i 10000000) (onep-zero 1))) 2017-03-29T23:42:20Z pkhuong: you're measuring the function call overhead. 2017-03-29T23:44:18Z aeth: there is exactly one meaningful difference 2017-03-29T23:45:20Z aeth: not zerop's disassembly has TEST RBX, RBX and the other has CMP RBX, 2 2017-03-29T23:51:20Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-29T23:52:10Z clintm quit (Remote host closed the connection) 2017-03-30T00:00:14Z cromachina joined #sbcl 2017-03-30T00:16:20Z stassats: so, me hitting a problem actually demonstrates that the gc improvement is not theoretical 2017-03-30T00:16:54Z stassats: but it's just much more visible for multi-gigabyte vectors 2017-03-30T00:18:05Z stassats: next up: sbcl should release memory to the OS more often 2017-03-30T00:41:43Z slyrus_ quit (Ping timeout: 246 seconds) 2017-03-30T00:54:05Z slyrus quit (Ping timeout: 240 seconds) 2017-03-30T01:08:19Z nyef: Hrm. Five expected failures in the test suite, plus one failure, on x86_64/linux. That sounds about right, doesn't it? 2017-03-30T01:08:48Z stassats: nyef: not really 2017-03-30T01:09:05Z stassats: nothing should be red 2017-03-30T01:09:40Z nyef: Failure: compiler.pure.lisp / (COMPILE EXIT-DELETION 2017-03-29) 2017-03-30T01:09:51Z stassats: that's new 2017-03-30T01:10:09Z nyef: I haven't commit it yet, but yes. It's new. 2017-03-30T01:10:31Z nyef: Eric reported it earlier today, in fact. d-: 2017-03-30T01:10:36Z stassats: that would be expected as well then 2017-03-30T01:11:12Z stassats: whack-a-mole 2017-03-30T01:12:02Z slyrus joined #sbcl 2017-03-30T01:12:20Z nyef: Indeed. And I got tired of not knowing if I was re-breaking stuff, so I chucked everything into the test suite two months ago, and am now adding to the test suite before I make a serious go at fixing the bug. 2017-03-30T01:12:55Z stassats: and i have a constraint propagation problem ahead of me 2017-03-30T01:13:12Z stassats: which is caused by an earlier fix of mine 2017-03-30T01:13:14Z nyef: At this point I'm thinking that the appropriate action is for me to start writing down what all this exit stuff actually does and why. 2017-03-30T01:15:16Z stassats: and constraint propagation is probably flawed somewhere, i remember it flipping one constraint back and forth never terminating 2017-03-30T01:15:38Z stassats: really hard to diagnose due to a lot of noise 2017-03-30T01:18:08Z nyef: Hrm. I could probably add a read-barrier to SBCL, but it'd basically require a full audit of the compiler backend to install. 2017-03-30T01:18:49Z nyef: But I'm far more confident of being able to do it than I would have been if you asked me ten minutes ago. 2017-03-30T01:19:07Z nyef: Now I "just" need a good read-barrier instruction sequence to use. (-: 2017-03-30T01:20:05Z stassats: just test and branch? 2017-03-30T01:20:10Z nyef: Maybe. 2017-03-30T01:26:03Z burtons quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-30T01:28:06Z nyef: ... I should implement some awesome, fancy, parallel, incremental, no-stop-the-world GC for SBCL... on some platform that nobody else has hardware for. 2017-03-30T01:31:11Z nyef: All these fancy features, and then "only works on System 6.0.7 with MultiFinder". 2017-03-30T02:11:16Z stassats quit (Ping timeout: 256 seconds) 2017-03-30T02:11:27Z edgar-rft joined #sbcl 2017-03-30T02:30:45Z paroneayea quit (Read error: Connection reset by peer) 2017-03-30T02:31:20Z paroneayea joined #sbcl 2017-03-30T02:59:52Z aeth: nyef: do it for RISC-V 2017-03-30T03:01:49Z nyef: aeth: I was thinking PARISC. Too much chance people would actually get RISC-V hardware. 2017-03-30T03:02:49Z aeth: you said no one has, not no one wants 2017-03-30T03:02:59Z nyef: (Besides, I already have PARISC hardware, although I suspect Linux kernel bugs where overall support is involved.) 2017-03-30T03:03:33Z nyef: Now, see, the easiest way to make sure that nobody *gets* the hardware is to arrange for it to be hardware that nobody wants in the first place. d-: 2017-03-30T03:04:54Z aeth: port SBCL to Symbolics Lisp Machines, and put the fancy GC there 2017-03-30T03:05:45Z nyef: Oh god no. That would be *way* out of character. 2017-03-30T03:06:00Z nyef: TI Lisp Machines would be far more appropriate. 2017-03-30T03:09:50Z aeth: but if you do it for discontinued hardware that is a collector's item, then no one can use it except for a few people, and the increase in demand will drive the price up 2017-03-30T03:10:03Z aeth: making it even harder to use! 2017-03-30T03:12:27Z daviid quit (Ping timeout: 240 seconds) 2017-03-30T03:14:10Z nyef: If anything, it's *easier* to get Symbolics LispM hardware than TI LispM hardware. 2017-03-30T03:14:31Z nyef: I have a MacIvory. I don't yet have a microExplorer. 2017-03-30T03:54:37Z milanj quit (Quit: This computer has gone to sleep) 2017-03-30T04:07:20Z Blkt quit (Read error: Connection reset by peer) 2017-03-30T04:07:20Z fe[nl]ix quit (Read error: Connection reset by peer) 2017-03-30T04:08:43Z Blkt joined #sbcl 2017-03-30T04:08:57Z fe[nl]ix joined #sbcl 2017-03-30T04:36:58Z dustinm` quit (Quit: Leaving) 2017-03-30T04:39:56Z shka_ joined #sbcl 2017-03-30T04:41:02Z dustinm` joined #sbcl 2017-03-30T04:59:55Z slyrus_ joined #sbcl 2017-03-30T05:15:56Z karswell quit (Remote host closed the connection) 2017-03-30T05:15:57Z karswell joined #sbcl 2017-03-30T05:15:57Z rpg quit (Ping timeout: 258 seconds) 2017-03-30T05:20:59Z shka_ quit (Ping timeout: 240 seconds) 2017-03-30T05:27:08Z reinuseslisp joined #sbcl 2017-03-30T05:40:23Z igajsin1 joined #sbcl 2017-03-30T05:43:23Z igajsin1 quit (Read error: Connection reset by peer) 2017-03-30T05:44:28Z rtmpdavid quit (Read error: Connection reset by peer) 2017-03-30T05:51:08Z scymtym quit (Ping timeout: 256 seconds) 2017-03-30T06:01:18Z igajsin1 joined #sbcl 2017-03-30T06:18:14Z attila_lendvai joined #sbcl 2017-03-30T06:18:14Z attila_lendvai quit (Changing host) 2017-03-30T06:18:14Z attila_lendvai joined #sbcl 2017-03-30T06:18:35Z angavrilov joined #sbcl 2017-03-30T06:25:43Z reinuseslisp quit (Quit: Leaving) 2017-03-30T07:26:58Z Bike quit (Quit: sleep) 2017-03-30T07:28:14Z scymtym joined #sbcl 2017-03-30T07:30:44Z gargaml joined #sbcl 2017-03-30T07:40:44Z milanj joined #sbcl 2017-03-30T08:51:54Z gargaml quit (Quit: WeeChat 1.7) 2017-03-30T08:53:33Z gargaml joined #sbcl 2017-03-30T08:53:41Z gargaml quit (Client Quit) 2017-03-30T08:53:55Z gargaml joined #sbcl 2017-03-30T08:53:57Z gargaml quit (Client Quit) 2017-03-30T08:54:41Z gargaml joined #sbcl 2017-03-30T08:56:09Z gingerale joined #sbcl 2017-03-30T08:56:24Z dougk quit (Ping timeout: 260 seconds) 2017-03-30T09:02:48Z scymtym: minion: memo for stassats: "Reduce scavenging of old large-object simple-vector." seems to have broken x86 builds (e.g. https://ci.cor-lab.org/view/rsx-trunk/job/sbcl-master/2116/featureset=1,label=ubuntu_trusty_32bit/console search for "unexpected forwarding pointer in scavenge") 2017-03-30T09:02:49Z minion: Remembered. I'll tell stassats when he/she/it next speaks. 2017-03-30T09:05:22Z foom quit (Ping timeout: 246 seconds) 2017-03-30T09:05:22Z dougk_ quit (Ping timeout: 246 seconds) 2017-03-30T09:07:34Z mateuszb quit (Ping timeout: 264 seconds) 2017-03-30T09:09:37Z foom joined #sbcl 2017-03-30T09:09:41Z dougk_ joined #sbcl 2017-03-30T09:14:59Z mateuszb joined #sbcl 2017-03-30T10:05:45Z sjl__ joined #sbcl 2017-03-30T10:20:15Z m00natic joined #sbcl 2017-03-30T11:18:05Z stassats joined #sbcl 2017-03-30T11:51:25Z sjl__ quit (Ping timeout: 260 seconds) 2017-03-30T12:01:35Z rpg joined #sbcl 2017-03-30T12:18:18Z sjl__ joined #sbcl 2017-03-30T12:18:45Z scymtym: minion: memo for stassats: maybe already be fixed. builds are still running. nvm and sorry in that case. 2017-03-30T12:18:48Z minion: Remembered. I'll tell stassats when he/she/it next speaks. 2017-03-30T12:25:52Z rpg quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-30T12:29:23Z sjl__ quit (Quit: WeeChat 1.3) 2017-03-30T12:29:45Z sjl joined #sbcl 2017-03-30T13:06:40Z White_Flame quit (Ping timeout: 268 seconds) 2017-03-30T13:08:26Z BitPuffin|osx joined #sbcl 2017-03-30T13:42:49Z burtons joined #sbcl 2017-03-30T14:00:02Z stassats: minion: forget my memos 2017-03-30T14:00:03Z minion: OK, I threw it out. 2017-03-30T14:01:53Z dougk joined #sbcl 2017-03-30T14:04:48Z cromachina quit (Read error: Connection reset by peer) 2017-03-30T14:09:56Z rpg joined #sbcl 2017-03-30T14:14:05Z rpg quit (Ping timeout: 240 seconds) 2017-03-30T14:36:03Z scymtym quit (Ping timeout: 258 seconds) 2017-03-30T14:38:27Z rpg joined #sbcl 2017-03-30T14:48:43Z sjl quit (Read error: Connection reset by peer) 2017-03-30T15:11:33Z Bike joined #sbcl 2017-03-30T15:27:37Z rpg quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-30T15:32:21Z rpg joined #sbcl 2017-03-30T15:37:02Z rpg_ joined #sbcl 2017-03-30T15:37:04Z rpg quit (Ping timeout: 256 seconds) 2017-03-30T15:45:57Z angavrilov quit (Ping timeout: 240 seconds) 2017-03-30T15:47:28Z angavrilov joined #sbcl 2017-03-30T16:09:43Z sjl joined #sbcl 2017-03-30T16:13:26Z m00natic quit (Remote host closed the connection) 2017-03-30T16:27:13Z rpg_ quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-30T16:28:50Z rpg joined #sbcl 2017-03-30T16:46:49Z dougk quit (Ping timeout: 268 seconds) 2017-03-30T16:47:29Z gargaml quit (Quit: WeeChat 1.7) 2017-03-30T16:55:39Z dougk joined #sbcl 2017-03-30T16:58:32Z shka_ joined #sbcl 2017-03-30T17:01:32Z pjstirling quit (Quit: Ex-Chat) 2017-03-30T17:10:38Z dougk quit (Read error: Connection reset by peer) 2017-03-30T17:10:41Z dougk__ joined #sbcl 2017-03-30T17:12:03Z daviid joined #sbcl 2017-03-30T17:47:08Z milanj quit (Quit: This computer has gone to sleep) 2017-03-30T17:55:27Z dougk__ quit (Ping timeout: 240 seconds) 2017-03-30T17:59:35Z brucem quit (Ping timeout: 276 seconds) 2017-03-30T17:59:36Z rpg quit (Ping timeout: 269 seconds) 2017-03-30T18:00:46Z brucem joined #sbcl 2017-03-30T18:01:00Z sjl quit (Ping timeout: 260 seconds) 2017-03-30T18:03:39Z vydd joined #sbcl 2017-03-30T18:05:23Z dougk joined #sbcl 2017-03-30T18:17:10Z dougk quit (Ping timeout: 264 seconds) 2017-03-30T18:20:08Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-30T18:24:27Z rpg joined #sbcl 2017-03-30T18:33:29Z dougk joined #sbcl 2017-03-30T18:51:59Z stassats: i never see anything in generations 2-4 2017-03-30T19:06:26Z sjl joined #sbcl 2017-03-30T19:12:45Z stassats: (setf (extern-alien "small_generation_limit" int) 0) releases memory more willingly 2017-03-30T19:20:56Z scymtym joined #sbcl 2017-03-30T19:23:21Z stassats: building with 1.2.13.72 2:23, with HEAD 2:15 2017-03-30T19:29:48Z stassats: with small_generation_limit = 0, 2:20 2017-03-30T19:30:21Z igajsin1 quit (Read error: No route to host) 2017-03-30T19:30:32Z nyef: How about with enough RAM and address space that you don't need to GC in the first place? d-: 2017-03-30T19:35:27Z stassats: trying without any freeing and with 16GB DS 2017-03-30T19:35:33Z dougk_: apropos of nothing, i think it's a bug that scav_instance returns 1 when there is no layout. It should return 1+slots, because there can't be anything in them. 2017-03-30T19:36:01Z dougk_: it's a rare edge case, but fixing this makes the contract of what a scavenge function returns more consistent 2017-03-30T19:37:39Z stassats: that goes to 2:12 2017-03-30T20:16:26Z milanj joined #sbcl 2017-03-30T20:28:00Z attila_lendvai joined #sbcl 2017-03-30T20:28:00Z attila_lendvai quit (Changing host) 2017-03-30T20:28:00Z attila_lendvai joined #sbcl 2017-03-30T20:47:26Z edgar-rft quit (Quit: edgar-rft) 2017-03-30T20:50:24Z gingerale quit (Remote host closed the connection) 2017-03-30T21:11:33Z angavrilov quit (Remote host closed the connection) 2017-03-30T21:11:53Z vydd quit 2017-03-30T21:15:29Z prxq joined #sbcl 2017-03-30T21:48:29Z White_Flame joined #sbcl 2017-03-30T22:14:01Z prxq quit (Remote host closed the connection) 2017-03-30T22:39:24Z rpg quit (Quit: Textual IRC Client: www.textualapp.com) 2017-03-30T22:54:31Z attila_lendvai quit (Quit: Leaving.) 2017-03-30T22:54:37Z attila_lendvai joined #sbcl 2017-03-30T22:55:40Z DeadTrickster quit (Ping timeout: 256 seconds) 2017-03-30T23:03:56Z attila_lendvai quit (Ping timeout: 252 seconds) 2017-03-30T23:07:17Z DeadTrickster joined #sbcl 2017-03-30T23:22:16Z cromachina joined #sbcl 2017-03-30T23:34:55Z burtons quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-30T23:52:43Z rumbler3_ joined #sbcl 2017-03-30T23:55:05Z shka_ quit (Ping timeout: 240 seconds) 2017-03-30T23:58:11Z slyrus_ quit (Ping timeout: 260 seconds) 2017-03-30T23:59:42Z slyrus_ joined #sbcl 2017-03-31T00:07:23Z rumbler3_ quit (Remote host closed the connection) 2017-03-31T00:20:08Z sjl quit (Ping timeout: 240 seconds) 2017-03-31T00:44:46Z slyrus_ quit (Ping timeout: 264 seconds) 2017-03-31T01:06:35Z karswell quit (Ping timeout: 240 seconds) 2017-03-31T01:09:49Z karswell joined #sbcl 2017-03-31T01:25:49Z burtons joined #sbcl 2017-03-31T01:26:03Z stassats quit (Ping timeout: 240 seconds) 2017-03-31T01:31:50Z brucem quit (Changing host) 2017-03-31T01:31:50Z brucem joined #sbcl 2017-03-31T02:07:54Z rumbler3_ joined #sbcl 2017-03-31T02:09:05Z burtons quit (Ping timeout: 240 seconds) 2017-03-31T02:12:05Z rumbler3_ quit (Ping timeout: 240 seconds) 2017-03-31T02:59:10Z dougk quit (Ping timeout: 246 seconds) 2017-03-31T03:13:57Z milanj quit (Quit: This computer has gone to sleep) 2017-03-31T03:25:50Z aeth: I have a good April Fool's idea for you SBCL maintainers. 2017-03-31T03:26:05Z aeth: Announce that you're adding units to SBCL's numeric tower. 2017-03-31T03:26:07Z aeth: These units. https://upload.wikimedia.org/wikipedia/commons/c/c8/English_length_units_graph.png 2017-03-31T03:27:12Z nyef: aeth: Or we could just announce that we've merged a windows backend. d-: 2017-03-31T03:27:13Z aeth: (yard (* (shaftment 4) (foot 3))) 2017-03-31T03:27:59Z aeth: (skein (+ (gunters-chain 37) (ramsdens-chain 32) (* 2 (cubit 8)))) 2017-03-31T03:28:24Z nyef: Really, though, just lengths and not the whole furlong-firkin-fortnight setup? 2017-03-31T03:28:35Z aeth: oh, add them all. 2017-03-31T03:28:50Z aeth: As long as they're not metric. 2017-03-31T03:29:19Z nyef: The windows backend one would be the better joke. I mean, who seriously uses windows enough to port SBCL to it? (-; 2017-03-31T03:29:44Z aeth: SBCL doesn't run on Windows? 2017-03-31T03:30:58Z nyef: I have no idea if it does or not these days. I did the original windows port, but I don't use windows these days. 2017-03-31T03:31:12Z aeth: People use it on Windows. 2017-03-31T03:31:19Z dougk joined #sbcl 2017-03-31T03:31:45Z nyef: Besides, SBCL's charter, like cliki's charter, is for unix systems, not for Windows. 2017-03-31T03:45:31Z nyef: Maybe for April Fool's 2018, we can announce that the HPPA/Linux backend is finally passing the test suite and generally expected to work? 2017-03-31T03:45:53Z nyef: (Yes, that gives us a year to do the work involved.) 2017-03-31T03:48:43Z aeth: Not only do people use SBCL on Windows, I bet some distribute Windows executables. 2017-03-31T03:48:57Z aeth: I wonder which Lisp dto uses for his games on Windows. 2017-03-31T04:16:46Z rumbler3_ joined #sbcl 2017-03-31T04:22:38Z dougk quit (Ping timeout: 256 seconds) 2017-03-31T04:32:58Z dougk joined #sbcl 2017-03-31T05:00:03Z flip214: you know the "units" command line tool, right? 2017-03-31T05:01:39Z nyef: Know, and occasionally use. 2017-03-31T05:03:15Z shka_ joined #sbcl 2017-03-31T05:07:42Z flip214: Well, I'd just call that via two pipes for such things. (Yeah, I understand April Fools' Day.) (( "Fools'" because it is plural, isn't it? )) 2017-03-31T05:10:43Z nyef: Now, see, the really appropriate thing to do is something impressive-yet-implausible on the day, like merging a new garbage collector that nobody saw coming, or a new compiler backend... 2017-03-31T05:11:07Z nyef: So everybody gets to think "nah, that's a prank", except that the prank is that it isn't a prank. 2017-03-31T05:15:09Z flip214: oh, a double-back. 2017-03-31T05:16:21Z flip214: well, how about announcing that sbcl on amd64 now uses the CPU microcode instead of assembler for even more speed? (implicitly, it does so already anyway ;) 2017-03-31T05:41:06Z shka_ quit (Ping timeout: 258 seconds) 2017-03-31T05:48:22Z akkad joined #sbcl 2017-03-31T05:48:41Z akkad: Symbol "*INTEXP-MAXIMUM-EXPONENT*" not found in the SB-EXT package. Did it get removed? 2017-03-31T05:49:47Z nyef: akkad: Mid last October, yes. 2017-03-31T05:51:02Z akkad: is there another constant? or can I just use some big num 2017-03-31T05:51:30Z Bike: do you want it for something? 2017-03-31T05:51:54Z nyef: Umm... It was a tuning parameter, and defaulted to NIL. 2017-03-31T05:52:34Z nyef: It certainly wasn't a constant. 2017-03-31T05:52:46Z Bike: i thought all it did was cause an error if you passed numbers bigger than it to expt 2017-03-31T05:53:00Z akkad: nope. just going to remove it. cl-bench was setting it 2017-03-31T05:53:04Z akkad: thanks 2017-03-31T05:53:18Z nyef: Yeah, that's basically what it did. 2017-03-31T05:55:03Z Bike quit (Quit: slug gourmet) 2017-03-31T06:04:40Z scymtym quit (Ping timeout: 246 seconds) 2017-03-31T06:16:53Z angavrilov joined #sbcl 2017-03-31T06:35:22Z rumbler3_ quit (Remote host closed the connection) 2017-03-31T06:35:28Z salv0 joined #sbcl 2017-03-31T06:47:03Z Reinisch quit (Ping timeout: 240 seconds) 2017-03-31T06:50:45Z Reinisch joined #sbcl 2017-03-31T06:52:50Z igajsin1 joined #sbcl 2017-03-31T07:20:44Z scymtym joined #sbcl 2017-03-31T07:28:30Z milanj joined #sbcl 2017-03-31T07:30:41Z daviid quit (Ping timeout: 260 seconds) 2017-03-31T07:36:05Z rumbler3_ joined #sbcl 2017-03-31T07:41:08Z DeadTrickster quit (Ping timeout: 240 seconds) 2017-03-31T07:41:33Z rumbler3_ quit (Ping timeout: 240 seconds) 2017-03-31T07:54:03Z DeadTrickster joined #sbcl 2017-03-31T07:58:17Z gargaml joined #sbcl 2017-03-31T08:02:23Z gingerale joined #sbcl 2017-03-31T08:09:27Z milanj quit (Ping timeout: 240 seconds) 2017-03-31T08:48:04Z specbot quit (Remote host closed the connection) 2017-03-31T08:48:04Z minion quit (Remote host closed the connection) 2017-03-31T08:51:31Z specbot joined #sbcl 2017-03-31T08:51:31Z minion joined #sbcl 2017-03-31T09:00:26Z gingerale quit (Ping timeout: 258 seconds) 2017-03-31T09:09:34Z milanj joined #sbcl 2017-03-31T09:37:59Z milanj quit (Quit: This computer has gone to sleep) 2017-03-31T10:14:30Z gingerale joined #sbcl 2017-03-31T10:14:40Z m00natic joined #sbcl 2017-03-31T10:19:04Z DeadTrickster quit (Ping timeout: 256 seconds) 2017-03-31T10:34:00Z DeadTrickster joined #sbcl 2017-03-31T11:15:22Z attila_lendvai joined #sbcl 2017-03-31T11:24:30Z milanj joined #sbcl 2017-03-31T11:46:38Z attila_lendvai quit (Quit: Leaving.) 2017-03-31T12:04:25Z attila_lendvai joined #sbcl 2017-03-31T12:05:18Z attila_lendvai quit (Client Quit) 2017-03-31T12:25:45Z daviid joined #sbcl 2017-03-31T13:15:03Z cromachina quit (Read error: Connection reset by peer) 2017-03-31T13:29:05Z specbot quit (Ping timeout: 260 seconds) 2017-03-31T13:29:10Z minion quit (Ping timeout: 264 seconds) 2017-03-31T13:31:30Z specbot joined #sbcl 2017-03-31T13:32:08Z minion joined #sbcl 2017-03-31T14:00:19Z burtons joined #sbcl 2017-03-31T14:17:56Z attila_lendvai joined #sbcl 2017-03-31T15:09:54Z gargaml quit (Quit: WeeChat 1.7) 2017-03-31T15:22:37Z dougk quit (Ping timeout: 258 seconds) 2017-03-31T15:24:34Z dougk joined #sbcl 2017-03-31T15:42:07Z shka_ joined #sbcl 2017-03-31T15:45:05Z minion quit (Remote host closed the connection) 2017-03-31T15:45:05Z specbot quit (Remote host closed the connection) 2017-03-31T15:52:26Z milanj quit (Quit: This computer has gone to sleep) 2017-03-31T16:01:06Z minion joined #sbcl 2017-03-31T16:01:45Z specbot joined #sbcl 2017-03-31T16:10:03Z Shinmera: I'm having a strange problem where, after a recent SBCL update, loading a particular FASL errors with "The name NIL does not designate any package." 2017-03-31T16:10:21Z Shinmera: Were there any changes in relation to FASLs or dumping recently? 2017-03-31T16:10:44Z Shinmera: I'll try to figure out a bit more but it's pretty cryptic. 2017-03-31T16:12:05Z Shinmera: It still worked with 1.3.12 2017-03-31T16:15:40Z nyef: That's a window of about four months. 2017-03-31T16:18:37Z Shinmera: I'll see that I can narrow it down some more 2017-03-31T16:19:39Z scymtym: Shinmera: i think, i remember something. let me see whether i can find it 2017-03-31T16:19:55Z nyef: From the 1.3.15 - 1.3.16 range, there's some changes to stream machinery, but that doesn't feel like a good match to the given error. 2017-03-31T16:20:16Z Shinmera: It already happens in 1.3.15 2017-03-31T16:20:38Z Shinmera: Building a 1.3.13 now 2017-03-31T16:22:12Z gingerale quit (Remote host closed the connection) 2017-03-31T16:23:05Z gingerale joined #sbcl 2017-03-31T16:27:02Z scymtym: i was thinking of 21c806c54e2aa05f97c277f9ec4d0fb2a404b9d8 but that was a long time ago 2017-03-31T16:28:25Z nyef: That looks like it's still within our candidate range. 2017-03-31T16:29:31Z scymtym: but that commit was supposed to fix the problem, so it shouldn't happen in 1.3.13 and newer 2017-03-31T16:30:01Z Shinmera: Aha! 2017-03-31T16:30:10Z Shinmera: It was introduced between 1.3.13 and 1.3.14 2017-03-31T16:31:20Z Shinmera: If I abort once that error happens, restart, and then load again it goes through without issue. 2017-03-31T16:31:55Z nyef: An 82 commit range. 2017-03-31T16:34:49Z Shinmera: I have a workaround for my specific issue, but I can't seem to use that information to narrow down to a minimal test case. 2017-03-31T16:36:40Z nyef: What do you have, or is it too large/private/dependency-ridden to share? 2017-03-31T16:36:53Z Shinmera: It's huge. 2017-03-31T16:37:14Z Shinmera: The error is something rather involved too-- emitting a literal package into a quoted form that gets funcall/compiled on LOAD 2017-03-31T16:37:31Z Shinmera: *the error cause 2017-03-31T16:37:36Z pkhuong: Shinmera: but scriptable? maybe git bisect with a script and wait until next week? :\ 2017-03-31T16:38:12Z nyef: Wait, what? You're trying to externalize package objects? 2017-03-31T16:38:55Z Shinmera: The specific form that triggers the error is (export 'some-symbol NIL) where NIL should be a literal package object as produced by a macro. 2017-03-31T16:39:39Z nyef: Hrm. 2017-03-31T16:39:52Z Shinmera: If instead of a package object I emit the name, it all works just fine. 2017-03-31T16:39:54Z nyef: I'm suspecting that that's nonconformant code. 2017-03-31T16:40:14Z Shinmera: Packages are FASL dumpable, if I remember correctly. 2017-03-31T16:41:33Z nyef: clhs 3.2.4.2.2 2017-03-31T16:41:33Z specbot: Definition of Similarity: http://www.lispworks.com/reference/HyperSpec/Body/03_bdbb.htm 2017-03-31T16:41:57Z nyef: Good call, they are, but... Hrm. 2017-03-31T16:44:35Z nyef: And there's a run of package-system work in this range. 2017-03-31T16:44:50Z Shinmera: Oh dear. 2017-03-31T16:45:05Z dougk quit (Ping timeout: 240 seconds) 2017-03-31T16:46:36Z nyef: Is your source available to look at somewhere? 2017-03-31T16:47:42Z Shinmera: https://github.com/Shirakumo/radiance/ is the system that triggers the error on load 2017-03-31T16:48:02Z Shinmera: Specific issue is this expansion https://github.com/Shirakumo/radiance/blob/master/interface-components.lisp#L18 2017-03-31T16:50:19Z nyef: ... How are "interface"s defined? 2017-03-31T16:50:55Z Shinmera: Here's an invocation https://github.com/Shirakumo/radiance/blob/master/standard-interfaces.lisp#L9 2017-03-31T16:51:02Z Shinmera: And the system that provides the base facilities is https://github.com/Shinmera/modularize-interfaces 2017-03-31T16:51:11Z Shinmera: It's been a long time since I wrote most of this, so I unfortunately don't have a good oversight anymore either. It's possible that my emitting the package object here is not legal due to other assorted dumb things I potentially did. 2017-03-31T16:54:36Z daviid quit (Ping timeout: 256 seconds) 2017-03-31T16:56:03Z m00natic quit (Remote host closed the connection) 2017-03-31T16:56:21Z nyef: So, you have an (e-w (:c :l :e) ...) around... 2017-03-31T16:56:28Z nyef: ... where's DEFINE-MODULE defined? 2017-03-31T16:56:51Z Shinmera: https://github.com/Shinmera/modularize/blob/master/module.lisp#L115 2017-03-31T16:58:47Z Shinmera: The use of (funcall (compile NIL '(lambda () ...))) around each interface component expansion is also strange. I can't recall why I had to add that. 2017-03-31T17:00:10Z milanj joined #sbcl 2017-03-31T17:00:51Z dougk joined #sbcl 2017-03-31T17:04:30Z nyef: This works the second time you load it, but blows up the first? 2017-03-31T17:04:57Z nyef: Trying to put together a test case now. 2017-03-31T17:05:34Z Shinmera: Yes. 2017-03-31T17:05:44Z Shinmera: Well, actually 2017-03-31T17:06:04Z Shinmera: When I load the entire system the first time, it errors. When I then abort, restart lisp, and load again it passes. 2017-03-31T17:11:29Z dougk quit (Ping timeout: 258 seconds) 2017-03-31T17:12:53Z nyef: Hrm. 2017-03-31T17:13:35Z nyef: Oh. I see why this test case doesn't break... 2017-03-31T17:13:42Z nyef: ... sortof. 2017-03-31T17:13:58Z nyef: Maybe? 2017-03-31T17:15:01Z nyef: Yeah, wow. I'm not sure how to go about boiling this down into an easier test case. /-: 2017-03-31T17:17:34Z dougk joined #sbcl 2017-03-31T17:18:44Z Shinmera: Me neither, and I wrote the mess. 2017-03-31T17:19:56Z nyef: How on earth do the symbols dump and recreate properly, if the package doesn't exist? 2017-03-31T17:20:14Z nyef: Answer: The package exists at compile-time, but is NIL at load time. 2017-03-31T17:20:51Z stassats joined #sbcl 2017-03-31T17:21:11Z Shinmera: Hm. 2017-03-31T17:21:28Z rumbler3_ joined #sbcl 2017-03-31T17:21:39Z nyef: No, that doesn't quite work either. 2017-03-31T17:22:06Z Shinmera: Since the package object should be dumpable it should exist at load time, too. 2017-03-31T17:22:07Z nyef: The package-name is valid enough for the symbols to get interned, but not enough to resolve the package object itself? 2017-03-31T17:22:34Z Shinmera: What is the supposed effect of compile-file on something like (macrolet ((foo () (make-package "foo"))) (foo))? 2017-03-31T17:22:38Z nyef: But I'd've expected FIND-PACKAGE-OR-LOSE or something like that to be involved... 2017-03-31T17:22:41Z stassats: Shinmera: what's the backtrace? 2017-03-31T17:22:57Z Shinmera: stassats: No, but I'm just trying to figure out what's even legal. 2017-03-31T17:23:04Z Shinmera: Oh, sorry 2017-03-31T17:23:08Z Shinmera: misread 2017-03-31T17:23:22Z nyef: Oh, true. Does this fail as a load-file failure, or does it drop to the debugger in the middle of loading the file? 2017-03-31T17:24:18Z stassats: i don't think it's actually related to dumping packages 2017-03-31T17:24:28Z stassats: but i didn't even look at the code 2017-03-31T17:24:46Z Shinmera: http://paste.lisp.org/display/343015 2017-03-31T17:25:35Z rumbler3_ quit (Ping timeout: 240 seconds) 2017-03-31T17:25:53Z nyef: That's... neat. 2017-03-31T17:26:34Z Shinmera: If I print the package that is emitted into the EXPORT at expansion time, it's never NIL. So it gets aliased to NIL somewhere, somehow. 2017-03-31T17:27:13Z nyef: Yeah, I figured as much. Because at expansion time, you're INTERNing into it. 2017-03-31T17:27:59Z nyef: And when it breaks, you have an external symbol in the package in question in hand. 2017-03-31T17:28:19Z scymtym: Shinmera: try (trace delete-package), maybe? that could cause the name to be NIL 2017-03-31T17:29:25Z nyef: Seems weak. 2017-03-31T17:29:59Z Shinmera: scymtym: No traces outputted. 2017-03-31T17:30:14Z scymtym: Shinmera: thanks 2017-03-31T17:31:02Z |3b|: does the package exist before you load the .fasl? 2017-03-31T17:31:18Z nyef: How about at https://github.com/Shinmera/modularize-interfaces/blob/master/interface.lisp#L124 you print the value of ,interface ? 2017-03-31T17:31:55Z |3b| thought loading compiled code with different package state was undefined behavior 2017-03-31T17:32:04Z nyef: Hrm. Not quite sure what it would imply. 2017-03-31T17:32:13Z nyef: |3b|: Right, but this is code that *sets up* a package state. 2017-03-31T17:32:25Z |3b|: ah 2017-03-31T17:32:45Z |3b|: so package doesn't exist when file is compiled? 2017-03-31T17:32:51Z Shinmera: nyef: Prints the package just fine. 2017-03-31T17:32:59Z nyef: Shinmera: At compile time, load time, or both? 2017-03-31T17:33:20Z Shinmera: nyef: It's in the log twice, so both times. 2017-03-31T17:34:14Z dougk__ joined #sbcl 2017-03-31T17:34:27Z Shinmera: http://paste.lisp.org/display/343015#1 2017-03-31T17:35:19Z dougk quit (Remote host closed the connection) 2017-03-31T17:35:19Z stassats: how do i run this code? 2017-03-31T17:35:52Z Shinmera: stassats: The quick way would be via this https://github.com/Shirakumo/radiance/#getting-it 2017-03-31T17:36:10Z Shinmera: Though I already distributed my workaround in the dist 2017-03-31T17:36:10Z stassats: that has the failing version? 2017-03-31T17:36:38Z stassats: which system to load to get the failure? 2017-03-31T17:36:42Z Shinmera: radiance 2017-03-31T17:37:01Z nyef: And this succeeds if there's already an up-to-date fasl for this source file, and fails if the compiler has to get involved? 2017-03-31T17:37:07Z Shinmera: I think it doesn't have any dependencies outside of what's on quicklisp nowadays, so just cloning it and loading that should reproduce it too 2017-03-31T17:38:07Z nyef: Wait... BAN:UNIMPLEMENTED ? 2017-03-31T17:38:16Z Shinmera: nyef: If I pick the "try-recompiling" restart it just continuously fails the same way. If I "abort", restart the lisp and retry, it works fine. 2017-03-31T17:38:25Z Shinmera: nyef: the symbol should be called UNIMPLEMENTED. 2017-03-31T17:38:50Z nyef: Ah, okay, I see. 2017-03-31T17:39:47Z nyef: Shinmera: Yeah, so I think that part of what's happening is that the process of compiling the file is creating the package, and the file is failing to load if the package exists. 2017-03-31T17:40:17Z Shinmera: stassats: cloning the system alone should work too, if you clone to 1412e745bd36cd08ab67b06644dfe5acb06ede78 2017-03-31T17:42:43Z Shinmera hopes he isn't wasting everyone's time 2017-03-31T17:45:33Z dougk__ quit (Ping timeout: 240 seconds) 2017-03-31T17:48:56Z stassats: i think i get it 2017-03-31T17:51:39Z edgar-rft joined #sbcl 2017-03-31T17:52:12Z stassats: (progn (when (eval nil) (defun foo () #.(find-package "CL"))) (print #.(find-package "CL"))) 2017-03-31T17:52:17Z stassats: compile-file it 2017-03-31T17:54:11Z Shinmera: Huh. 2017-03-31T17:55:04Z stassats: and it's old 2017-03-31T17:56:12Z Shinmera: Can't say I'd know how to explain that behaviour. 2017-03-31T17:56:24Z stassats: you're not expected to 2017-03-31T17:57:54Z nyef: That's... not quite the same bug, is it? 2017-03-31T17:58:25Z nyef: Unless the form that Shinmera is seeing is recently triggered by other changes? 2017-03-31T17:59:00Z stassats: it is the same thing i'm seeing from the original code 2017-03-31T18:01:19Z |3b|: Shinmera: in your backtrace, is the NIL supposed to be the package object for BAN? 2017-03-31T18:02:43Z Shinmera: Yes 2017-03-31T18:04:36Z |3b|: seems odd that it has a symbol BAN:UNIMPLEMENTED exported from a package that didn't exist when the .fasl was loaded 2017-03-31T18:05:16Z Shinmera: The package does exist, it just gets aliased somehow. 2017-03-31T18:05:18Z |3b|: also, where does that UNIMPLEMENTED come from? 2017-03-31T18:05:26Z |3b|: why does it exist? 2017-03-31T18:05:30Z stassats: |3b|: well, i already reduced it 2017-03-31T18:05:43Z |3b| thought that was in same form as the definition of the package 2017-03-31T18:05:47Z Shinmera: because I can print it from another point in the code before the offending place. 2017-03-31T18:06:11Z |3b|: stassats: yeah, yours sounds more like the cause of the NIL, i'm just trying to figure out if the code is conformant aside from that :) 2017-03-31T18:06:31Z |3b|: Shinmera: so it exists, then you do a defpackage and overwrite it? 2017-03-31T18:06:43Z |3b| is confused by the code 2017-03-31T18:07:00Z Shinmera: The code only creates the package if find-package doesn't already return it. 2017-03-31T18:07:05Z dougk joined #sbcl 2017-03-31T18:07:15Z |3b|: ah, ok 2017-03-31T18:08:36Z |3b|: when you were printing, were you loading it into a fresh lisp image? 2017-03-31T18:08:46Z |3b|: (as in not the one where compilation defined the package) 2017-03-31T18:09:10Z Shinmera: If I load the FASL in a fresh image, the problem doesn't occur. 2017-03-31T18:09:20Z burtons quit (Quit: My MacBook has gone to sleep. ZZZzzz…) 2017-03-31T18:13:56Z Shinmera has to scurry off for a while 2017-03-31T18:19:37Z |3b|: stassats: seems to do same thing for hash tables as well 2017-03-31T18:20:55Z specbot quit (Remote host closed the connection) 2017-03-31T18:20:55Z minion quit (Remote host closed the connection) 2017-03-31T18:24:20Z stassats: |3b|: anything with a load-form 2017-03-31T18:24:23Z specbot joined #sbcl 2017-03-31T18:24:26Z |3b|: ah 2017-03-31T18:25:02Z minion joined #sbcl 2017-03-31T18:40:11Z BitPuffin|osx quit (Remote host closed the connection) 2017-03-31T18:46:27Z stassats: load-time-value is guaranteed to be evaluated only once, does that include zero times? 2017-03-31T18:47:36Z minion quit (Remote host closed the connection) 2017-03-31T18:47:36Z specbot quit (Remote host closed the connection) 2017-03-31T18:49:39Z stassats: the only solution i can come up with: get rid of fopcompilation 2017-03-31T18:53:36Z stassats: non-load-form constants aren't skipped by fopcompile-if, so maybe load-time-value shouldn't either 2017-03-31T18:53:40Z specbot joined #sbcl 2017-03-31T18:53:41Z minion joined #sbcl 2017-03-31T19:00:00Z stassats: short of ditching fopcompilation (which i'm in favor for a long time) skip-until thingy should not skip load-time-value forms 2017-03-31T19:04:47Z nyef: Heh. Do you remember when fopcompilation was merged in the first place? 2017-03-31T19:05:28Z stassats: in fasl-dump-load-time-value-lambda, a new fop fop-funcall-no-skip does the job 2017-03-31T19:21:10Z stassats: not skipping any load-time-value doesn't survive cold load 2017-03-31T19:21:22Z stassats: i'll just do that for make-load-form then 2017-03-31T19:31:48Z stassats: Shinmera: fixed 2017-03-31T19:32:04Z stassats: thanks for "wasting everyone's time" 2017-03-31T19:34:28Z scymtym quit (Ping timeout: 258 seconds) 2017-03-31T20:20:23Z scymtym joined #sbcl 2017-03-31T20:22:30Z Shinmera: stassats: Well, I was worried I might be doing something stupid and it wasn't actually anything to the fault of SBCL. 2017-03-31T20:23:08Z Shinmera: Glad to hear it was fixed, though. 2017-03-31T20:23:11Z igajsin1 quit (Ping timeout: 268 seconds) 2017-03-31T20:42:31Z daviid joined #sbcl 2017-03-31T20:57:22Z gingerale quit (Remote host closed the connection) 2017-03-31T21:13:19Z vydd joined #sbcl 2017-03-31T21:13:19Z vydd quit (Changing host) 2017-03-31T21:13:19Z vydd joined #sbcl 2017-03-31T21:20:04Z angavrilov quit (Remote host closed the connection) 2017-03-31T21:42:34Z specbot quit (Remote host closed the connection) 2017-03-31T21:42:34Z minion quit (Remote host closed the connection) 2017-03-31T21:50:10Z vydd quit 2017-03-31T21:59:53Z minion joined #sbcl 2017-03-31T21:59:53Z specbot joined #sbcl 2017-03-31T22:12:25Z dougk quit (Ping timeout: 246 seconds) 2017-03-31T22:27:09Z shka_ quit (Ping timeout: 260 seconds) 2017-03-31T22:42:20Z stassats: dougk_: with immobile layouts, can we have instances in unboxed regions? 2017-03-31T22:42:52Z daviid quit (Ping timeout: 246 seconds) 2017-03-31T22:57:39Z dougk joined #sbcl 2017-03-31T23:09:07Z aeth: oh this would be a good April Fools: SBCL is moving to immutable conses. 2017-03-31T23:09:29Z stassats: this an april fools safe zone 2017-03-31T23:09:43Z aeth: all rplaca, rplacd, setf (car foo) and setf (cdr foo) will warn 2017-03-31T23:32:30Z dougk quit (Ping timeout: 268 seconds) 2017-03-31T23:44:16Z vtomole joined #sbcl 2017-03-31T23:46:42Z dougk joined #sbcl