2014-10-13T00:04:07Z scymtym_ quit (Ping timeout: 272 seconds)
2014-10-13T00:11:25Z Hache__ joined #sbcl
2014-10-13T00:12:21Z Hache_ quit (Ping timeout: 272 seconds)
2014-10-13T00:51:45Z pkhuong: christoph_debian: http://paste.lisp.org/display/144013 should take care of frlock failures
2014-10-13T00:51:52Z pkhuong: what else fails?
2014-10-13T01:49:05Z psy_ quit (Ping timeout: 255 seconds)
2014-10-13T02:24:09Z Hache__ quit (Remote host closed the connection)
2014-10-13T02:34:57Z psy_ joined #sbcl
2014-10-13T02:38:52Z christoph_debian quit (Ping timeout: 240 seconds)
2014-10-13T02:39:36Z psy_ quit (Ping timeout: 246 seconds)
2014-10-13T02:42:30Z bugrum joined #sbcl
2014-10-13T02:49:36Z psy_ joined #sbcl
2014-10-13T02:52:14Z christoph_debian joined #sbcl
2014-10-13T03:46:27Z nyef quit (Quit: G'night all.)
2014-10-13T03:56:20Z jhao quit (Ping timeout: 260 seconds)
2014-10-13T03:58:14Z bugrum quit (Remote host closed the connection)
2014-10-13T06:08:43Z pranavrc joined #sbcl
2014-10-13T06:08:44Z pranavrc quit (Changing host)
2014-10-13T06:08:44Z pranavrc joined #sbcl
2014-10-13T06:08:49Z edgar-rft joined #sbcl
2014-10-13T06:11:59Z oleo quit (Quit: Verlassend)
2014-10-13T06:47:07Z stassats joined #sbcl
2014-10-13T06:52:33Z jhao joined #sbcl
2014-10-13T06:53:45Z kanru quit (Remote host closed the connection)
2014-10-13T06:54:44Z kanru joined #sbcl
2014-10-13T06:57:01Z jhao quit (Ping timeout: 260 seconds)
2014-10-13T07:02:43Z prxq joined #sbcl
2014-10-13T07:38:24Z prxq quit (Remote host closed the connection)
2014-10-13T07:52:59Z DGASAU quit (Read error: Connection reset by peer)
2014-10-13T07:53:34Z DGASAU joined #sbcl
2014-10-13T07:53:46Z DGASAU quit (Read error: Connection reset by peer)
2014-10-13T07:54:29Z DGASAU joined #sbcl
2014-10-13T08:09:41Z psy_ quit (Quit: Leaving)
2014-10-13T08:39:04Z aldo joined #sbcl
2014-10-13T09:25:35Z scymtym_ joined #sbcl
2014-10-13T09:48:22Z aldo quit (Ping timeout: 240 seconds)
2014-10-13T10:24:42Z karswell` quit (Remote host closed the connection)
2014-10-13T10:41:17Z attila_lendvai joined #sbcl
2014-10-13T10:54:33Z scymtym_ quit (Ping timeout: 272 seconds)
2014-10-13T11:02:33Z aldo joined #sbcl
2014-10-13T11:25:42Z DGASAU quit (Read error: Connection reset by peer)
2014-10-13T11:26:22Z DGASAU joined #sbcl
2014-10-13T11:39:29Z attila_lendvai quit (Disconnected by services)
2014-10-13T11:39:29Z attila_lendvai1 joined #sbcl
2014-10-13T11:39:29Z attila_lendvai1 quit (Changing host)
2014-10-13T11:39:29Z attila_lendvai1 joined #sbcl
2014-10-13T11:44:55Z attila_lendvai1 quit (Quit: Leaving.)
2014-10-13T12:17:48Z eudoxia joined #sbcl
2014-10-13T12:19:52Z specbot quit (Remote host closed the connection)
2014-10-13T12:20:38Z specbot joined #sbcl
2014-10-13T12:20:53Z leoc quit (Ping timeout: 240 seconds)
2014-10-13T12:23:23Z yauz quit (Ping timeout: 240 seconds)
2014-10-13T12:23:37Z ferada quit (Ping timeout: 245 seconds)
2014-10-13T12:24:27Z ferada joined #sbcl
2014-10-13T12:24:30Z yauz joined #sbcl
2014-10-13T12:27:28Z leoc joined #sbcl
2014-10-13T12:30:48Z eudoxia quit (Quit: Lost terminal)
2014-10-13T13:01:48Z psilord quit (Quit: Leaving.)
2014-10-13T13:20:51Z Krystof: - 260:8b 45 b8             mov    -0x48(%ebp),%eax
2014-10-13T13:20:51Z Krystof: + 260:8b 45 b0             mov    -0x50(%ebp),%eax
2014-10-13T13:20:54Z Krystof: argh
2014-10-13T13:21:34Z Krystof: if I were to go hunting for code which decides what should go in which stack location, where would I find it?
2014-10-13T13:24:07Z stassats: from what to what?
2014-10-13T13:24:15Z stassats: i mean, the differences
2014-10-13T13:25:59Z stassats: could it be due to unstable sort differences?
2014-10-13T13:27:15Z pkhuong: Krystof: that's pack's job
2014-10-13T13:27:52Z oleo joined #sbcl
2014-10-13T13:28:52Z Krystof: one weird thing is that this shows up only in a handful of files
2014-10-13T13:29:44Z stassats: so, which target and which host?
2014-10-13T13:29:59Z stassats: ok, target i can guess, x86
2014-10-13T13:30:29Z pkhuong: Or indirectly, via VOP template specs
2014-10-13T13:31:02Z Krystof: target is actually x86-64 but my objdump is stupid
2014-10-13T13:31:08Z Krystof: hosts are sbcl and ccl32
2014-10-13T13:31:16Z Krystof: and almost all files are clear of this problem
2014-10-13T13:31:43Z Krystof: one that isn't is src/compiler/x86-64/subprim.lisp; this is the machine code resulting from compiling define-vop (length/list)
2014-10-13T13:31:47Z Krystof: which I presume is the generator
2014-10-13T13:32:09Z Krystof: nb this is the machine code for the generator function not the generator's output machine code
2014-10-13T13:35:28Z stassats: i can see a call to SORT in pack-iterative.lisp
2014-10-13T13:35:56Z pkhuong: stassats: it's sorting by unique ids
2014-10-13T13:36:04Z stassats: alright then
2014-10-13T13:37:58Z Krystof: in fact the only files that are affected by this problem are a handful in src/compiler/x86-64/
2014-10-13T13:37:59Z pkhuong: do we compile with code coverage?
2014-10-13T13:38:01Z Krystof: n
2014-10-13T13:38:02Z Krystof: o
2014-10-13T13:39:03Z Krystof: not completely anyway.  Might be possible to build without PCL.  But wait, code coverage of what?
2014-10-13T13:39:24Z Krystof: we should be able to build the cross-compiler with coverage information
2014-10-13T13:39:34Z Krystof: if our host is sbcl, anyway
2014-10-13T13:39:42Z pkhuong: we have some support to compile stuff and track code coverage, but it's full of hash tables
2014-10-13T13:39:57Z pkhuong: anyway, not likely.
2014-10-13T13:46:08Z pranavrc quit
2014-10-13T13:50:55Z LiamH joined #sbcl
2014-10-13T13:54:30Z Krystof: I bet I know what it is
2014-10-13T13:55:28Z attila_lendvai joined #sbcl
2014-10-13T13:55:42Z attila_lendvai quit (Changing host)
2014-10-13T13:55:43Z attila_lendvai joined #sbcl
2014-10-13T14:09:37Z Krystof: the irony.  The last three repeatable-build problems I've fixed are to do with quote, backquote and code-walking
2014-10-13T14:27:27Z nyef joined #sbcl
2014-10-13T14:34:34Z stassats: so, do you know what is the problem now?
2014-10-13T14:36:51Z psilord joined #sbcl
2014-10-13T14:37:02Z nyef: ... some former edge case, now triggered on a regular basis?
2014-10-13T14:37:11Z nyef: (That's usually the problem, at least.)
2014-10-13T14:37:44Z stassats: can somebody do automated core comparison tests?
2014-10-13T14:40:21Z fridim_ joined #sbcl
2014-10-13T14:41:02Z scymtym: stassats: between successive builds or between different host compilers?
2014-10-13T14:41:15Z stassats: of course not successive builds, between different hosts
2014-10-13T14:41:37Z scymtym: well, tracking core size changes may be a reason for comparing successive builds
2014-10-13T14:41:57Z scymtym: ok, but reading the previous discussion, i see the intention
2014-10-13T14:42:18Z stassats: we need to keep _sanely_ bootstrappable
2014-10-13T14:42:29Z scymtym: which hosts would be interesting? sbcl, clisp, ccl?
2014-10-13T14:42:46Z stassats: for starters
2014-10-13T14:44:25Z scymtym:  i can try to do this
2014-10-13T14:47:14Z scymtym: are there Debian/Ubuntu packages for ccl?
2014-10-13T14:53:34Z Krystof: that is what I'm doing
2014-10-13T14:53:41Z Krystof: at the moment the cores aren't bitwise identical
2014-10-13T14:53:49Z Krystof: there are several fasls which differ
2014-10-13T14:53:55Z Krystof: at the moment, core comparison will say "they're different"
2014-10-13T14:54:05Z Krystof: need to fix all the problems first
2014-10-13T14:54:08Z stassats: after you are erase the differences
2014-10-13T14:54:29Z scymtym: for now, we will at least know when a host stops working
2014-10-13T14:54:47Z scymtym: comparing produced cores can be added when it makes sense
2014-10-13T15:02:22Z psilord quit (Ping timeout: 255 seconds)
2014-10-13T15:09:47Z psilord joined #sbcl
2014-10-13T15:15:22Z psilord quit (Ping timeout: 240 seconds)
2014-10-13T15:24:39Z slyrus quit (Remote host closed the connection)
2014-10-13T15:32:21Z prxq joined #sbcl
2014-10-13T15:43:51Z krzysz00 quit (Ping timeout: 272 seconds)
2014-10-13T15:49:07Z Krystof: \o/ down to 4 files that differ
2014-10-13T15:49:50Z stassats: what if you build it on  february 29?
2014-10-13T15:51:04Z nyef: As though the circumstances that would cause that happen more than twice a decade...
2014-10-13T15:51:20Z stassats: or one on a plain to the west, another on a taxi to the west?
2014-10-13T15:51:28Z stassats: east
2014-10-13T15:52:50Z Krystof: or moving very fast?  There are actually no date-related things in fasls
2014-10-13T15:53:20Z stassats: will the core consume more bytes?
2014-10-13T15:55:00Z Krystof: om nom nom
2014-10-13T15:55:32Z Krystof: (I fixed the date-related things in cross-compilation five and a half years ago)
2014-10-13T15:58:58Z levenson joined #sbcl
2014-10-13T15:59:33Z levenson: Hi guys, could someone explain me how should I use `with-standard-replacement-character' macro?
2014-10-13T16:00:18Z stassats: you shouldn't?
2014-10-13T16:00:43Z Krystof: if you have to ask, you shouldn't; it's not documented or exported
2014-10-13T16:00:57Z Krystof: what are you actually trying to do?
2014-10-13T16:02:46Z levenson: I ment "How I could.. " use it. I have an octet vetor and sometimes I get SB-IMPL::END-OF-INPUT-IN-CHARACTER trying to convert it to a string
2014-10-13T16:03:21Z Krystof: you still shouldn't use it
2014-10-13T16:03:25Z stassats: use an appropriate external format
2014-10-13T16:03:29Z Krystof: you should use documented functionality to do what you want instead
2014-10-13T16:03:52Z Krystof: for example, look at the end of <http://www.sbcl.org/manual/#External-Formats> this section
2014-10-13T16:04:27Z stassats: '(:utf-8 :replacement #\?) would be a start
2014-10-13T16:04:37Z levenson: Krystof: Yes but in my case I have exactly broken UTF line
2014-10-13T16:04:44Z levenson: Ohh
2014-10-13T16:06:04Z levenson: Thanks!
2014-10-13T16:06:11Z levenson: Work like a charm)
2014-10-13T16:17:20Z nyef: ... we can DO that?
2014-10-13T16:17:39Z nyef: What'll it take to also support a :line-terminator-style option?
2014-10-13T16:17:56Z stassats: it will take somebody
2014-10-13T16:18:17Z Krystof: nyef: documented since November... 2009
2014-10-13T16:20:31Z nyef: Fourteen high-priority bugs in launchpad. Two of them are about line-terminators in external-formats.
2014-10-13T16:21:34Z stassats: nobody cares enough
2014-10-13T16:21:46Z stassats: marking bugs as "high priority" is quite easy
2014-10-13T16:22:32Z nyef: Mmm. I've been considering closing 484143 as a duplicate... of a low-priority (or was it wishlist?) bug.
2014-10-13T16:22:43Z stassats: lp 484143
2014-10-13T16:22:44Z specbot: https://bugs.launchpad.net/bugs/484143
2014-10-13T16:23:16Z stassats: nyef: i have no idea
2014-10-13T16:23:44Z stassats: the only different thing that could happen is to blow the stack guard
2014-10-13T16:24:09Z nyef: Number 309096.
2014-10-13T16:24:14Z stassats: but there lies a security problem, if it's so far away past the guard, it may actually hit something mapped and cause some problems
2014-10-13T16:24:21Z stassats: lp 309096
2014-10-13T16:24:21Z specbot: https://bugs.launchpad.net/bugs/309096
2014-10-13T16:26:31Z stassats: and the backtrace is corrupted
2014-10-13T16:26:47Z stassats: that's actually interesting
2014-10-13T16:28:49Z stassats: i actually have two faults, the second one is arrange_return_to_c_function
2014-10-13T16:35:24Z stassats quit (Ping timeout: 246 seconds)
2014-10-13T16:52:43Z attila_lendvai quit (Quit: Leaving.)
2014-10-13T16:57:07Z slyrus joined #sbcl
2014-10-13T17:01:09Z scymtym: https://ci.cor-lab.org/job/sbcl-master/ will now build the --fancy featureset with host {clisp,ccl} on Ubuntu trusty{32,64} (in addition to the previous variants)
2014-10-13T17:10:14Z ccl-logbot joined #sbcl
2014-10-13T17:10:14Z 
2014-10-13T17:10:14Z names: ccl-logbot slyrus levenson prxq fridim_ nyef LiamH oleo leoc yauz ferada specbot DGASAU aldo kanru edgar-rft christoph_debian carvite heddwch gingerale jackdaniel _8hzp jsnell_ leo2007 Bike loke_ fikusz @Krystof irsol reb Posterdati milosn alchemis7 scymtym ams slyrus_ asedeno jdz njmurphy foom pkhuong nicdev White_Flame thoto flip214 Hydan fe[nl]ix Blkt weissschloss akkad minion sav snafuchs qbit p_l sobel les faheem joshe Intensity antoszka Perlboy
2014-10-13T17:10:14Z names: p_l|backup brucem mood ivan`` |3b| pchrist redline6561 luis
2014-10-13T17:55:54Z stassats joined #sbcl
2014-10-13T17:58:06Z stassats quit (Client Quit)
2014-10-13T18:18:22Z levenson quit (Quit: shutdown)
2014-10-13T18:23:08Z krzysz00 joined #sbcl
2014-10-13T18:32:40Z slyrus_ quit (Quit: Client Quit)
2014-10-13T18:43:13Z krzysz00 quit (Ping timeout: 272 seconds)
2014-10-13T19:00:07Z krzysz00 joined #sbcl
2014-10-13T19:06:39Z psilord joined #sbcl
2014-10-13T19:06:39Z psilord quit (Client Quit)
2014-10-13T19:07:04Z psilord joined #sbcl
2014-10-13T19:10:45Z psilord quit (Client Quit)
2014-10-13T19:10:55Z psilord joined #sbcl
2014-10-13T19:40:42Z scymtym_ joined #sbcl
2014-10-13T19:43:06Z aldo quit (Ping timeout: 240 seconds)
2014-10-13T19:50:10Z krzysz00 quit (Ping timeout: 250 seconds)
2014-10-13T21:00:57Z aldo joined #sbcl
2014-10-13T21:06:21Z prxq quit (Ping timeout: 272 seconds)
2014-10-13T21:07:57Z krzysz00 joined #sbcl
2014-10-13T21:09:46Z Bicyclidine joined #sbcl
2014-10-13T21:27:39Z heddwch quit (Ping timeout: 255 seconds)
2014-10-13T21:32:14Z scymtym_: i would like to fix 1380081 by calling getaddrinfo(3) with AI_IDN on platforms which have that (see http://paste.lisp.org/display/144020 for output). the only(?) problem is that the "current locale" has to be used when encoding the query before passing it to getaddrinfo. any ideas how to ensure that?
2014-10-13T21:32:40Z scymtym_: (ironically, macos, for which the 1380081 was reported, doesn't seem to support AI_IDN)
2014-10-13T21:32:55Z heddwch joined #sbcl
2014-10-13T21:33:54Z psilord quit (Quit: Leaving.)
2014-10-13T21:43:44Z Hache_ joined #sbcl
2014-10-13T21:56:54Z oleo is now known as Guest70443
2014-10-13T21:57:42Z pkhuong: can we defer that complexity to a library?
2014-10-13T21:57:48Z oleo__ joined #sbcl
2014-10-13T21:57:58Z pkhuong: what part of getaddrinfo must/should live in sb-bsd-socket?
2014-10-13T22:00:49Z Guest70443 quit (Ping timeout: 272 seconds)
2014-10-13T22:00:52Z Bicyclidine quit (Ping timeout: 255 seconds)
2014-10-13T22:01:07Z Bicyclidine joined #sbcl
2014-10-13T22:07:47Z Bicyclidine quit (Ping timeout: 272 seconds)
2014-10-13T22:08:08Z Bicyclidine joined #sbcl
2014-10-13T22:10:29Z eudoxia joined #sbcl
2014-10-13T22:23:57Z scymtym_: if our get-host-by-name should accept non-ascii strings, we probably have to extend our getaddrinfo usage like that
2014-10-13T22:24:32Z scymtym_: if it shouldn't, i think it would be better to reject non-ascii hostnames right away
2014-10-13T22:25:41Z karswell` joined #sbcl
2014-10-13T22:44:15Z eudoxia quit (Quit: wow such dinnertime)
2014-10-13T22:48:33Z nyef: Is it possible to do the IDN encoding on the SBCL side rather than play crazy encoding games that might not even work to marshal them over to the system library?
2014-10-13T22:56:52Z scymtym_: nyef: no idea whether that is possible. the indented mode of operation is encoding non-ascii queries with the current locale and setting the AI_IDN flag
2014-10-13T22:57:07Z scymtym_: s/with/according to/
2014-10-13T22:58:08Z nyef: Mmm. But the only reason to involve the current locale is to marshal the non-ascii characters over to C code which uses the locale to convert them to unicode in order to do the IDN mapping, isn't it?
2014-10-13T22:58:26Z scymtym_: i think so
2014-10-13T22:59:18Z nyef: So in one sense, the question is, "do IDN domain names appear as normal domain names with a funky encoding, or is it yet another DNS record type?"
2014-10-13T23:02:08Z scymtym_: iolib seems to operate like that using a system called idna for performing the encoding
2014-10-13T23:04:26Z nyef: So, the approach is workable, and doesn't involve crazy external format games.
2014-10-13T23:04:46Z nyef: And is more likely to work even without explicit support for AI_IDN on the host.
2014-10-13T23:05:09Z nyef: The downside being having to maintain an IDN mapper.
2014-10-13T23:05:13Z Bicyclidine quit (Ping timeout: 255 seconds)
2014-10-13T23:05:41Z scymtym_: yes
2014-10-13T23:06:14Z scymtym_: the idna library seems to be around 200 lines of code
2014-10-13T23:06:14Z nyef: Okay, as long as we know what the tradeoffs are.
2014-10-13T23:08:19Z scymtym_: we may need only half of that since the we do not currently return (thus would not have to decode) any canonical names returned by the resolver
2014-10-13T23:13:01Z aldo quit (Ping timeout: 272 seconds)
2014-10-13T23:13:39Z Bicyclidine joined #sbcl
2014-10-13T23:28:38Z gingerale- joined #sbcl
2014-10-13T23:31:16Z gingerale quit (Ping timeout: 272 seconds)
2014-10-13T23:31:16Z gingerale- is now known as gingerale
2014-10-13T23:36:20Z fridim_ quit (Ping timeout: 272 seconds)
2014-10-13T23:38:46Z DGASAU quit (Read error: Connection reset by peer)
2014-10-13T23:39:43Z DGASAU joined #sbcl
2014-10-13T23:39:53Z pkhuong: iirc, punycode is pretty simple.
2014-10-13T23:47:21Z csziacobus joined #sbcl
2014-10-13T23:50:21Z csziacobus quit (Client Quit)