Index

Show enters and exits. Hide enters and exits.

00:01:23seydar leaves the room.
00:13:16seydar enters the room.
00:13:21seydarwhoa i'm still here
00:13:57seydarDefiler: sure, i'd love a walk through in using format-patch. but rue and I decided it's broken
00:14:53seydarnoooo everyone's idling
00:15:03hoopylurks
00:15:32boyscout1 commit by Philipp Bruschweiler
00:15:33boyscout * added specs for SHA256/384/512; 01baf00
00:16:48seydarDefiler: is it possible, instead of using defined(__APPLE__) to use defined(_STRUCT_MCONTEXT) ?
00:17:20therealadam leaves the room.
00:17:29dbussinkseydar: yeah, but it could be that that is not correct
00:17:38dbussinkseydar: because it's a work around for leopard
00:17:49dbussinkseydar: so that could break systems that do not have that bug
00:17:58seydark. thanks
00:18:33seydardid Defiler leave
00:18:35seydar?
00:19:58dbussinkbrixen: you there?
00:20:13brixendbussink: yeah
00:20:34dbussinkbrixen: doing some lighthouse clean up, looks like 302 could be closed?
00:21:42brixenone sec, I'll check
00:21:52brixenalso, you might coordinate with zenspider on that
00:21:58brixenthe lh cleanup
00:23:46seydarcan someone walk me through using `git format-patch`? i swear i'll write it up in a doc
00:24:01dbussinkbrixen: well, i'm closing some stuff that i know for sure is fixed
00:24:03drbrainam I supposed to be able to include in core?
00:24:31brixendbussink: yep, I think #302 is fine to close
00:25:42cremeswow, lot's of specs in spec/ruby/1.8/core/string/unpack_spec.rb fail against mri... need to fix that before i can patch it up on ppc
00:26:05brixencremes: yep, at least 4, they need to use the endian guards
00:26:24brixencremes: I fixed the pack ones, haven't gotten back to it
00:26:36brixenseydar: what is your question about git format-patch?
00:26:55brixenseydar: and do you have the man pages installed? because the examples part of the man page is pretty clear
00:27:20seydarbrixen: because i found that i was unable to use it, and my man pages were missing
00:27:21cremesbrixen: question then... should formats i, q, s and l use the native format or always default to little endian? it's forced to little now
00:27:41cremesbrixen: but that seems wrong to me; i think they should use native
00:27:54dbussinkseydar: you know how to create your own git branch>
00:28:05brixencremes: I'd have to check, perhaps use ri or pickaxe, etc
00:28:14seydardbussink: kinda, but I'm just trying to do a patch
00:28:19dbussinkseydar: the man pages are also available online
00:28:34dbussinkseydar: yeah, but for git-format-patch you need to branch and then commit
00:28:44rubuildius_ppcPhilipp Bruschweiler: 01baf002a; 1616 files, 5540 examples, 18962 expectations, 0 failures, 93 errors; http://pastie.caboo.se/paste/156159
00:28:45dbussinkseydar: and after that, you create the patch using git-format-patch
00:29:11seydarah! then i should use `git send-email`, right?
00:29:24dbussinkseydar: no, just upload the file it creates to lighthouse
00:29:38seydarok
00:29:46rubuildiusPhilipp Bruschweiler: 01baf002a; 1616 files, 5540 examples, 19081 expectations, 0 failures, 0 errors; http://rafb.net/p/9WFgw994.html
00:29:47dbussinkcremes: hmm, looks like those sha specs create endianness problems too
00:30:05seydari got everything working (almost) - my patch showed in `git show HEAD`
00:30:38cremesdbussink: yep; i'm working on a massive patch right now but it won't be done until tomorrow at the earliest
00:31:10dbussinkseydar: you can use git-format-patch revision where revision is the one where the patch starts, so the patch will include all commits after that revision (non inclusive)
00:31:13cremesdbussink: i'm targeting the pack/unpack problems for the first patch; i'll tackle the sha specs in a second patch
00:31:21dbussinkcremes: nice :)
00:31:36dbussinkcremes: you already have commit access?
00:31:55cremesdbussink: nope, but these patches will get me some!
00:32:58seydarugh. i'll check the man pages now. ltr allgtrz
00:33:03seydar leaves the room.
00:35:57twbray leaves the room.
00:35:59dbussinkit getting too late again, really need to get some sleep
00:37:17lstoll enters the room.
00:40:37sfaxon enters the room.
00:46:24brixendbussink: sleep, that's an order! :D
00:46:30rueevan: Wait, what is the problem that the #pack is solving?
00:46:43dc_ leaves the room.
00:47:03brixenrue: you mean the pastie?
00:47:53dbussinkbrixen: yes boss
00:48:24rueevan: There are like eight different size detection things in there already. Rubinius::WORDSIZE, and Rubinius::L64 at least
00:49:03dbussinkit's about endianness, not size stuff
00:49:43dbussinkbut now really off to bed, night all
00:49:46langenberg leaves the room.
00:49:47brixenyeah, we should add one for ENDIAN rather than rely on the MemoryPointer thing
00:49:53brixennight dbussink
00:50:43rueArgh, there already IS a Rubinius::ENDIAN :D
00:50:55evanthere is?
00:51:07evanha!
00:51:08evani rock.
00:51:16brixenyay
00:51:27brixenthen there's really no excuse for our broken #pack
00:51:43cremeswhere is Rubinius::ENDIAN defined?
00:51:55evanit's created in machine.c
00:52:05evanit's a bootstrapped constant.
00:52:27brixenheh, and it's correct
00:52:31cremesi see it... :big and :little
00:52:36evanyep
00:52:43evanthats what it should be
00:52:49cremesi'll use that facility for my patch instead
00:52:53evangood plan
00:53:48brixenspec'ing mspec to use instance_eval when rspec uses instance_eval sucks
00:54:46drbrainevan: am I allowed to include or attr_reader in kernel/core/?
00:54:58evansure
00:55:05drbrainhrm
00:55:09evanattr_reader can only take one arg though
00:55:19drbrainyeah
00:56:04hassox leaves the room.
00:56:19drbrainunintuitive
00:56:28evanwhat is?
00:56:46drbrainif I list the dependencies that Ar has in # depends on: everything blows up in my face
00:56:59evanhm.
00:57:01drbrainif I remove them, it works
00:57:10ezmobiuswhat's Ar?
00:57:26drbrainezmobius: it reads ar(5) formatted files
00:57:44evandrbrain: you're ok not having depends on lines for Ar
00:57:57drbrainI added a bunch of dependencies all over the place, but eventually died over in breakpoint.rb
00:57:59wifelette leaves the room.
00:58:02evansince nothing uses it while core is loaded
00:58:12evaner. loading.
00:58:15hassox enters the room.
00:58:15drbrainyeah
00:58:51drbrainstill, it's weird to have kernel/core/ar.rbc at the very tip-top of everything
00:58:58fizx enters the room.
00:58:59evanwhy?
00:59:14evanseems fine to me
00:59:28drbrainit just feels odd, that's all
01:00:06wifelette enters the room.
01:00:08wifelette leaves the room.
01:01:29nemerle_afk leaves the room.
01:01:53fizx leaves the room.
01:10:19crossblaim leaves the room.
01:19:13boyscout2 commits by Evan Phoenix
01:19:14boyscout * Properly cast the conversion funcitons; 4dfc3e7
01:19:15boyscout * Include ncurses to link with our readline extension; 0d7a3a8
01:21:01srbaker leaves the room.
01:21:18srbaker enters the room.
01:26:45imajes leaves the room.
01:27:24imajes enters the room.
01:28:23rubuildius_ppcEvan Phoenix: 4dfc3e7fc; 1616 files, 5540 examples, 19005 expectations, 0 failures, 67 errors; http://pastie.caboo.se/paste/156174
01:29:31evandude
01:29:33evanwtf is with ppc.
01:30:00rubuildiusEvan Phoenix: 4dfc3e7fc; 1616 files, 5540 examples, 19081 expectations, 0 failures, 0 errors; http://rafb.net/p/mkc5Ez67.html
01:30:47evanfires up his powerbook
01:30:59drbrainevan: errors may have gone up due to sha256/384/512 specs
01:31:59evanlooks like Digest::* and File.truncate mainly
01:33:06imajes leaves the room.
01:33:17imajes enters the room.
01:33:53KirinDave leaves the room.
01:44:19cored enters the room.
01:45:07twbray enters the room.
01:50:56evanDefiler: you around?
01:51:10evani've solved the MCONTEXT problem
01:51:17evan#if __DARWIN_UNIX03 && __ENVIRONMENT_MAC_OS_X_VERSION_MIN_REQUIRED__ >= 1050
01:52:33twbray leaves the room.
01:57:30brixenevan: you around?
01:57:52evanyeppers
01:58:08evanfixin' ppc
01:58:30brixencool, so it can roll around in my head a little, what was the reason you were wanting to explore the C stack stuff? you said something about only have control of the C stack in subtend
01:59:10tarcieriwhat happens with subtend if two VMs load the same native extension?
01:59:13evanmainly because if you look at every other VM like ours
01:59:30evanthe next level they get to is generating native code for a method, and using the C stack to control flow
01:59:46brixenok, so it's the native code/jit realm, yes?
01:59:49evanbecause machine's have instructions to call functions
01:59:54brixenyep
01:59:56evanit's related, yeah
02:00:18evantarcieri: not sure
02:00:20brixenso, remember karel driesen of the pic paper? have you seen this: http://www.usenix.org/event/jvm02/full_papers/zendra/zendra_html/index.html
02:00:25evantarcieri: i suspect the first one wins
02:00:47brixenevan: also, a bunch of stuff here on profiling and tracing I'm going to start going through: http://www.cs.arizona.edu/~collberg/Teaching/620/2005/Resources/index.html
02:00:55evani could implement subtend_battle_royale() to resolve it
02:00:59tarcierihaha
02:01:24brixenevan: finally, grep this page for driesen, some interesting papers: http://wotan.liu.edu/docis/dbl/oopsla/index.html
02:01:26evanbrixen: ok, i'll take a look shortly
02:01:42evani was working on PIC sendsite code last night
02:01:49evanmaybe i'll finish it tonight
02:01:57lopex leaves the room.
02:02:01brixendriesen seems to be right there with piumarta, ungar, holzle in the dynamic lang stuff
02:02:04evanyeah
02:02:05tarcierihow do you implement userspace execution contexts in a VM that's not "stackless"
02:02:10evani recognize his name
02:02:10tarcierii.e. Tasks
02:02:29tarcierilike uhh, ucontext comes to mind, but I thought you were getting rid of that
02:02:35evana couple of ways
02:02:43evanyeah, ucontext/setjmp/longjmp is one
02:02:57evananother is that the C stack only contains 'managed' frames
02:03:01tarcierisetjmp/longjmp are great for suspending... how do you resume?
02:03:13evanie, the runtime can figure out the contents of every frame, none are opaque
02:03:21imajes leaves the room.
02:03:23evanso you can do stuff like 'unwind' the C stack into the heap
02:03:41evanthen replay it back onto a C stack at a later time
02:03:49imajes enters the room.
02:03:51evanthats what Visual Works does
02:03:53tarcierioh, crazy
02:04:05brixenoh, interesting
02:04:06tarcieriso same thing with closures then?
02:04:15evanyeah
02:04:20brixenso you have something like the spaghetti stack buy unwinding into the heap
02:04:26evanyep
02:04:30tarcierithat warps my fragile little brain
02:04:34brixenok, I'm not so worried then :P
02:04:39evanthen you get the spagetti wet and pull it into a line
02:04:42evanand run it on the C stack
02:04:47brixenok
02:04:55evanlater on, you pick it up again
02:05:12evanbut you don't have to the entire spagetti chain at once
02:05:17brixenvery intersesting
02:05:17evanVW does it in chunks
02:05:27evanso you might unwind into the heap
02:05:42evanand when you go to call again, you only paste the last few contexts into frames, and keep going
02:05:51brixenok, this makes a lot more sense
02:06:04evanif you end up having to return back to the ones that weren't on the C stack, you paste them down
02:06:12tbmcmullen leaves the room.
02:06:18brixenI mentioned the gcc issue with fusion to bart yesterday, he rolled his eyes at gcc
02:06:34evanhah
02:06:44evansounds like a fun guy
02:06:48brixenheh
02:06:48ezmobiusbrixen: did you get to talk to matz about the rubyspec stuff?
02:06:52evanand no, i'm not calling him a mushroom
02:06:58brixenezmobius: yes, a bit
02:07:16brixenezmobius: matz basically said, "contact me any time by email, I don't do IRC much"
02:07:30ezmobiusheh cool
02:07:31brixenI think we really need to coordinate with ko1 on them running the specs
02:07:47ezmobiusyes mri and yarv really need to be running specs.
02:07:50brixenhe also said ko1 can skype him if we really need to get ahold of him
02:07:53ezmobiuskinda crazy they dont test at all
02:08:14brixenwell, yes, we'll do some marketing and outreach soon
02:08:20brixenthis week I'll get rubyspec.org up
02:10:42brixenevan: so, it's fair to say (my question) that we really don't have much C code if we decided on a code gen strategy. i.e. we could code the primitives and instructions and interface to the GC still written in C
02:10:49brixendoes that seem reasonable to say?
02:11:12technomancy leaves the room.
02:11:13evanmm
02:11:16evani don't follow
02:11:57brixenwell, all these chunks of code that make up the primitives, we could pretty easily replace that with something else to gen machine code, as long as calling into that, and calling from that was sane
02:14:21evansure
02:14:28evannot sure why we'd generate machine code directly for the primitives though
02:14:50brixenwell, just thinking of how we might divide this up, so we can tackle a part at a time
02:14:56brixenwithout rewriting the gc for instance
02:14:57evanthe primitives are all leaves in the call tree
02:15:13evanthey don't call back, and contain no control flow
02:15:39evanwell, we'd be doing code generation by looking at the bytecode for a method
02:15:42brixenright, not call back to Ruby, but the other C functions they call
02:15:47evanit doesn't really impact the primitives
02:16:01evanthis would be runtime machine code, held in memory, not saved to disk.
02:16:59brixenI'm thinking the cola stuff, writing it in something besides C, so no C compiler ;)
02:18:31evanhm
02:18:44evanwell, i don't see that as a first step
02:18:58evanthe first step would be to link the VM with the cola runtime
02:19:12evanso that the VM could hand strings to cola and get a function pointer back
02:19:16evanthen we'd build on that
02:19:23brixenok
02:19:26brixenmakes sense
02:20:35evanmy thinking on this is that we go the direct of expury
02:20:37evanthe smalltalk compiler
02:20:50evanwhere the native code compiler is actually all written in smalltalk
02:20:57brixenahh, yeah I need to read up on that, been meaning to
02:21:09evanand the VM has enough hooks to call back into to to generate code for a method
02:21:35evanthere are some minor brain fucks in there, like the compiler trying to compile itself, etc.
02:21:51brixenheh
02:25:59evanok, so, 64bit args are busted on ppc
02:26:03evanthats the issue with truncate
02:26:07evanbecause off_t is a long long
02:32:36brixenhmm, I checked that FFI.type_size(:off_t) == sizeof(off_t) in C
02:33:12brixenso, this is the arg passing in FFI stub?
02:34:22evanyeah
02:34:27evanit's lightning's fault
02:34:30evanit doesn't have a long long type
02:34:41brixenahh
02:37:45evanok
02:37:58evanlooks like on ppc, it splits a long long like it were 2 longs
02:38:04evanusing 2 registers
02:38:22evanoh ha!
02:38:28evanno wonder it's a totally bogus number.
02:38:39evanthe lightning code doesn't even have a case for TYPE_ULL
02:39:04evanso there is just garbage.
02:41:56evancrap on a stick.
02:47:04VVSiz_ enters the room.
02:52:02_mutle enters the room.
02:57:56hornbeck leaves the room.
03:05:01brixencraptastic, we are so broken using instance_eval in mspec, 102 errors
03:05:16VVSiz leaves the room.
03:05:41brixenwell, dinner time
03:06:56mutle leaves the room.
03:08:50Fullmoon enters the room.
03:09:29Fullmoon leaves the room.
03:11:27defunkt enters the room.
03:20:08fork_ leaves the room.
03:25:41d2dchat leaves the room.
03:29:49MenTaLguY enters the room.
03:30:01MenTaLguYhowdy folks!
03:30:19coredhi
03:34:21manverualoha
03:36:11rueHey
03:36:31d2dchat enters the room.
03:36:33manverudrbrain: ping :)
03:46:12jtoy enters the room.
03:49:09d2dchat leaves the room.
03:54:19_ADS_ enters the room.
03:54:39seydar enters the room.
03:56:04seydarDefiler: are you on a coffee break or here?
04:01:19hornbeck enters the room.
04:05:50AndrewO leaves the room.
04:06:02seydarVagabond? you here?
04:08:37seydarok, why are my additions to ipaddr.rb not being recognized by shotgun/rubinius?
04:09:59Vagabondseydar: technically
04:10:19seydarVagabond: get your terminal open. we've got some debugging to do
04:11:17seydarVagabond: can you run something for me?
04:12:11seydaractually.... i'll just open up some free shell accounts and test on them as well
04:12:15Vagabondheh
04:12:27seydardo you want the logins to them so you can test?
04:12:28VagabondI don't really have time tonight
04:13:04seydarbut do you want the logins now anyways?
04:13:22headius leaves the room.
04:14:36Vagabondfor what?
04:14:53seydarto some free shell accounts
04:14:57VagabondI have logins to ~70 machines of various OS/architecture combinations
04:15:14Vagabondso, not so much
04:15:22seydarorly? howd you get those?
04:15:23ttmrichter leaves the room.
04:15:25seydarhaxxor?
04:15:35VagabondI sysadmin
04:16:01seydarh4xXøR
04:17:11seydarwanna share some logins?
04:18:26seydarugh ipaddr is giving me errors
04:18:34Vagabondno, I prefer to not compromise the security of the machines I am responsible for ;)
04:18:44seydarhaha
04:18:54wycats leaves the room.
04:19:02seydari wish i could sysadmin. but first i need to get older. much
04:19:17drbrainseydar: how old are you?
04:19:32seydar15
04:19:45drbrainthat's plenty old
04:19:52seydarwhere could i syadmin?
04:20:19RyanTM leaves the room.
04:20:24drbrainlook for an internship over the summer
04:20:43seydari tried last summer, but no one called me back :-(
04:21:00RyanTM enters the room.
04:21:00drbrainwell, try harder :)
04:21:17seydarmaybe i should be more agressive...
04:21:22seydaraggressive*
04:21:35drbrainyes!
04:21:39seydarEY should move to new hampshire
04:21:59VagabondI suspect not too many people want a 15 year old administrating their servers, no offence
04:22:19VagabondI'd let you admin the windows IIS server that I want to die, but that's about it ;)
04:22:39AndrewO enters the room.
04:22:44drbrainVagabond: I was sysadmin for my high school when I was 16
04:22:56drbrainI created all the student accounts
04:23:01Vagabondah, i didn't think of a school
04:23:05Vagabondthat's much more likely
04:24:03drbrainor less
04:24:06seydarmaaaaaaybe. but i hate my school. i take courses at dartmouth, and that's where i want to get training. sysadmin's in my class, too
04:24:10Vagabondplaying admin on your own hardware is a good way to get experience too
04:24:23AndrewO leaves the room.
04:24:27drbrainconsidering that I had access to things that should have been Top Secret
04:24:32Vagabondhas no training
04:24:41seydarthis is slightly OT, but the errors I'm getting on Tiger I'm getting on my Ubuntu machine
04:24:46Vagabondat least not any formal training ;)
04:24:47drbrainlike, the attendance/grading system
04:24:51seydarthe errors from IPAddr
04:25:04Vagabonddrbrain: schools are always excitingly lax about such things, in my experience
04:25:10drbrainseydar: manveru had problems with IPAddr too
04:25:10seydarmy school is not so
04:25:26seydarwe had major cheating scandal. made NPR in US
04:25:33Vagabonddoes IPAddr work for anyone? ;)
04:26:09drbrainVagabond: it works for Intel OS X people
04:26:20manveruVagabond: works for me
04:26:28Vagabondmanveru: what changed?
04:26:33manverujust 192.168.0.0 doesn't work
04:26:40Vagabondoh
04:27:11seydaranyways, for the cheating scandal, if you look for a person on NPR transcripts who gave their name as Mike Rotch, that was me
04:27:29manveruand that only on linux/amd64, linux/intel32 is fine
04:28:19seydarmanveru: which linux OS? ubuntu/intel32 gives 11 errors for IPAddr
04:28:32manveruseydar: archlinux
04:28:42seydarwell hmph
04:29:10seydarcan we set up communal ssh servers for testing on multiple platforms?
04:30:03manverui can provide access on this machine
04:30:03drbrainwe were talking about that today
04:30:20wycats enters the room.
04:30:36seydarmanveru: nice. drbrain: what was the decision?
04:30:51Vagabondwhat about a way to run automated CI and report back failures
04:31:08drbrainwe were thinking about getting an x-serve and running parallels on it with various OSs
04:31:20VagabondI could set up a cronjob on some of machines, to get more platforms/architectures regularly tested
04:31:43seydarVagabond: use god - pure ruby
04:32:03drbrainVagabond: the real problem is that when things are breaking, nobody is fixing them
04:32:18wycats leaves the room.
04:32:34manveruwell, i would fix things as much as i can...
04:32:51manverubut can't fix anything on the C side
04:33:01defunkt leaves the room.
04:33:01manveruwhich is the one thing most likely to break
04:33:06Vagabonddrbrain: partly because the build system is so byzantine I can't figure out how it all goes together
04:33:43drbrainVagabond: rake git:pull build spec
04:34:04drbrainif you want a clean build all the time, you'll have to do the git clone in a separate step
04:34:20manverui usually do rake git:pull rebuild spec
04:34:23twbray enters the room.
04:34:32Vagabonddrbrain: sure, but it's only meaningful if the results are actually paid attention to
04:34:42drbrainyeah
04:34:56Vagabondmy previous comment was in regard to fixing issues, not reporting failures
04:37:37manveruso, uhm
04:38:01manverudrbrain: wanna help me debug today? :)
04:38:52drbrainmanveru: I think I'm > < close to getting compiler bootstrapping with the now .rbas
04:40:02manveruwasn't there a law that states that you are > < close when your are > < close?
04:40:08drbrainmanveru: give me an hour or so
04:40:17manveru-r
04:40:21defunkt enters the room.
04:40:22manveruheh, sure
04:42:49seydarVagabond: your now on debugging duty. it's simple stuff, dealing with regexs
04:43:13Vagabondseydar: how am I supposed to help?
04:43:37manveruyay, regex!
04:43:44seydari give you a value for addr and some base regex, and you help them work!
04:43:55seydarmaybe, i'll write them in TextualRegexp
04:44:02seydarawesome gem by awesome dudes
04:44:33Vagabondseydar: okay, shoot I guess
04:44:56seydark. hold on
04:48:14seydarugh. parental unit is ruining my life. but basically, IPAddr.new("0123:4567:89ab:cdef:0ABC:DEF0:1234:5678") is returning an error, from line 467
04:48:34seydarlater alligators
04:48:40seydar leaves the room.
04:48:51manveruoO
04:50:05MenTaLguYI think you need to have colons between every octet in IPv6, don't you?
04:52:38VagabondI think that's a valid ipv6 address
04:53:41AndrewO enters the room.
04:55:10nicksieger enters the room.
04:57:16ezmobius leaves the room.
04:57:23cored leaves the room.
04:59:56d2dchat enters the room.
05:00:46AndrewO leaves the room.
05:01:43seydar enters the room.
05:01:59seydarguys i figured out waht's wrong
05:01:59binary42 leaves the room.
05:02:07seydarin IPAddr, the SPEC IS BAD
05:02:19seydarso, i'm gonna fix it. nobody else fix it
05:03:40Vagabondneat, rbx won't even build right now
05:04:51drbrainVagabond: why?
05:06:33Vagabond** Execute runtime/core/kernel/core/archive.rbc
05:06:33Vagabondshotgun/rubinius -Iruntime/stable/compiler.rba -rcompiler/init compile -frbx-safe-math -frbx-kernel kernel/core/archive.rb runtime/core/kernel/core/archive.rbc
05:06:36Vagabondrake aborted!
05:06:43VagabondI'm not sure even what to make of the errors
05:07:56drbrainmanveru: yay!
05:08:47seydarDefiler: you're here, right?
05:09:00seydardamn.
05:09:02twbray leaves the room.
05:09:20seydarok, can someone walk through the EXACT commands for doiing a patch?
05:09:29drbrainseydar: it is midnight on a Friday in Defilerville
05:10:35manverudrbrain: yay!?
05:11:14drbrainmanveru: I successfully navigated the byzantine corridors of bootstrapping a new compiler.rba
05:12:41seydarso what exactly is a ruby archive?
05:12:44manveruwell... i have no idea what compiler.rba does, but congratulates nonetheless :)
05:15:12ezmobius enters the room.
05:15:58manveru'Some people want love - Others want money - Me... well... I just want this code to compile'
05:17:24seydarhow'd you get that?
05:17:42seydarare you and vagabond haxxoring again?
05:17:59seydari wrote that after a long night of haskellll
05:18:00VagabondI gotta run
05:18:01Vagabondlater
05:18:02boyscout4 commits by Eric Hodel
05:18:03boyscout * Bootstrap new ar(5) compiler.rba.; 9ba068b
05:18:04boyscout * New core stable with Ar.; d1c7ad2
05:18:05boyscout * Call Ar#after_loaded.; 7ca509a
05:18:06boyscout * Move Ar to kernel/core.; b2baf09
05:18:21drbrainthere, now you can all share in my joy while I enjoy a sandwich
05:19:49manveruseydar: from the ruby-talk ML
05:20:18seydarmanveru: haha. can you help me submit a patch?
05:20:25manveruno
05:20:33seydardrbrain: i'll give you a sandwich if you help me
05:20:58seydarapple butter, ham, chedder cheese... toasty warm... mmmmmmmm......
05:21:06manveruhttp://rubini.us/contribute
05:21:58manveruand don't ask me for specifics... all i do is filing bugs and annoying devs
05:24:05seydari got you covered on the latter
05:24:19seydarfiling bugs? not so much. but I JUST SUMBIT
05:24:23seydarTHIS MEANS I"M COOL
05:26:59seydarand i found a bug on ubuntu
05:28:06seydardoing a `rake clean build` just to be sure...
05:29:26rubuildius_ppcEric Hodel: 9ba068bf2; 1616 files, 5540 examples, 19005 expectations, 0 failures, 67 errors; http://pastie.caboo.se/paste/156229
05:29:50manveruseydar: i usually do `rake rebuild`
05:30:02seydarbtw drbrain, your awesome bootstrap.rba thing threw up on ubuntu/intel32
05:30:07seydarmanveru: k. trying that
05:30:18drbrainseydar: how so?
05:31:03rubuildiusEric Hodel: 9ba068bf2; 1616 files, 5540 examples, 19081 expectations, 0 failures, 0 errors; http://rafb.net/p/GgzIo498.html
05:31:32seydardrbrain: i just ran `rake rebuild` right after. lemme pastie it for you.
05:32:22_ADS leaves the room.
05:32:36seydardrbrain: http://pastie.caboo.se/156232
05:33:48drbrainseydar: what does file runtime/stable/* say?
05:34:15seydarbootstrap.rba compiler.rba core.rba loader.rbc platform.rba
05:34:24drbrainthe especially strange part is that it finishes
05:34:33drbraindoes it say they are ar archives?
05:34:40seydaryea, that really caught me offguard
05:34:46seydarar archives?
05:35:02_ADS_ enters the room.
05:35:09drbrainfile(1) should print out what kind of file it is
05:35:40seydaryup. current ar archive
05:36:24manveruouch
05:36:31manverudrbrain: likewise, it totally broke
05:36:51manverubut later...
05:36:52seydarubuntu and archlinux, right?
05:37:00drbrainhrm
05:37:08drbrainbut, how did it get to the end?
05:37:19manveruhttp://p.ramaze.net/624
05:37:41manveruwell, wait until you've seen mine ^^;
05:38:39manverumaybe you forgot adding some file?
05:38:43seydarmy guess is that it failed, but the controller of load_order.txt went to pick up the rest
05:38:54jeremydurham enters the room.
05:39:51drbrainmanveru: ar -t runtime/stable/core.rba | grep ar
05:39:56drbrainwhat does that say?
05:40:02drbrainI see an ar.rbc as the first thing
05:40:31manverukernel/core/ar.rbc
05:40:38MenTaLguY leaves the room.
05:40:48manverubut there is only kernel/core/ar.rb
05:41:37drbrainwhen you `rake` it should have built runtime/core/runtime/core/ar.rbc
05:41:52drbrainerr, runtime/core/kernel/core
05:41:56drbrainhrm
05:42:03manveruhmm
05:42:08manveruyeah, running `rake` does that
05:42:35ezmobius leaves the room.
05:42:44manveruyet there is no file...
05:42:59drbrainwhich file is missing?
05:43:23ezmobius enters the room.
05:43:28drbrainruntime/core/kernel/core/ar.rbc?
05:43:55seydari have 7 files in my kernel/core/. how many should i have?
05:44:03manveruno, that's there
05:44:22drbrainand it's in runtime/stable/core.rba?
05:44:33manveruyet `rake spec` is missing Ar
05:45:22manveruwell
05:45:27manveruno
05:46:02manverukernel/core/ar.rbc is referenced from runtime/stable/core.rba
05:46:14manveruls **/ar.rbc
05:46:15manveruruntime/core/kernel/core/ar.rbc
05:46:21manveruis the only one around
05:46:25drbrainmanveru: you should have: kernel/core/ar.rb, runtime/core/kernel/core/ar.rbc
05:46:35manverukernel/core/ only has ar.rb
05:46:54drbrainthere should only be .rb files in kernel/core
05:47:06manveruok... then i guess it's correct
05:47:10Swistak leaves the room.
05:47:13manveruand it just doesn't work
05:47:25drbrainok, next is ar.rb in runtime/core/.load_order.txt ?
05:47:49drbrainsorry, kernel/core/ar.rbc should be listed there
05:47:53seydari need to get some sleep
05:48:03seydardrbrain, you be on tomorrow?
05:48:07manveruthe load_order doesn't mention it at all
05:48:14drbrainseydar: probably
05:48:26drbrainmanveru: ok, try another `rake build`
05:48:36seydargood stuff. night everybody. dibs on fixing the ipaddr specs
05:48:39seydar leaves the room.
05:48:59manverudrbrain: that doesn't change anything
05:49:11drbrainnothing got built?
05:49:16manveruexactly
05:49:39drbrainok, try touch kernel/core/ar.rb; rake build
05:49:49_ADS leaves the room.
05:50:06manveruok, that did it
05:50:08Ingmar leaves the room.
05:50:32drbrainit must have had a timestamp too new for `rake` to DTRT
05:50:34manverunow `rake spec` runs
05:50:34Ingmar enters the room.
05:50:48drbrainyay!
05:51:55seydar enters the room.
05:52:15drbrainok, let's see what I can do to ffi_decode_sockaddr
05:52:30seydarbtw, someone seems to have gotten rid of loader.rbc in my code
05:52:35seydarcan someone do a check?
05:52:57seydarafter a `git pull`?
05:53:01drbrainseydar: loader.rbc should only be checked in in runtime/stable/
05:53:11seydarmspec can't find it
05:53:38seydarwtf it's there!
05:53:45drbrainwhat's the full path it's looking for?
05:53:58drbrainseydar: try touch kernel/loader.rb; rake
05:54:04manverumeh
05:54:14manverui hate that `rake spec` just fails and doesn't tell you _what_ failed
05:55:05seydardrbrain: bootstrap.rba got in the way and failed everything
05:55:16seydari'm tired now. i'll check back tomorrow
05:55:18seydar leaves the room.
05:55:21drbrainok
05:57:23manverurake aborted!
05:57:24manveruCommand failed with status (): [bin/mspec ci -t rbx...]
05:57:48manverudo you guys use another method to run specs?
05:58:55drbrainI used to use bin/ci
05:59:01drbrainbut it's going away now
05:59:12drbraindoes bin/mspec ci give you any more data?
05:59:38manveruthat's what rake is running... no?
05:59:54drbrainthat's exactly what rake is running
06:00:10manveruso how would that give me anything else?
06:00:54drbrainsometimes rake hides things, I wasn't sure if that's what you were seeing or not
06:01:29manverulet's see if zsh is more intelligent...
06:01:36benburkert enters the room.
06:02:03manveruzsh: illegal hardware instruction ./bin/mspec ci -t rbx
06:02:16manveruwell... but i still don't know which spec
06:04:14jeremydurham leaves the room.
06:07:12drbrainwtf?
06:07:21drbrainI hate you mspec
06:07:29drbrainto include AF_INET6::1AF_INET60localhost30217:
06:08:48manveruand a simple change later it shows me which files it runs
06:09:59manveruha
06:10:08manveruhow surprising, it fails on getsockopt
06:10:28manverudrbrain: http://p.ramaze.net/625
06:10:53drbrainmanveru: yeah
06:14:16benburkert leaves the room.
06:16:08manverubbl
06:16:14headius enters the room.
06:16:33GMFlash leaves the room.
06:16:41GMFlash enters the room.
06:17:20benburkert enters the room.
06:27:09defunkt leaves the room.
06:27:21defunkt enters the room.
06:32:34imajes_office leaves the room.
06:42:02sudoer enters the room.
06:42:21jtoy leaves the room.
06:53:40boyscout4 commits by Eric Hodel
06:53:41boyscout * Replace ffi_pack_sockaddr_in and ffi_bind with ruby code.; 43b8516
06:53:42boyscout * Remove ffi_decode_sockaddr, replace with existing ruby code.; a0fe2f7
06:53:43boyscout * Restore ability to mspec ci with ruby to make command-line editing easy.; 1cc32e0
06:53:44boyscout * Include matcher should inspect; 5fdf2d6
06:54:25drbrainwooo! take that ffi_util.c!
06:54:31drbrain96 ++--------------------
06:54:42drbrainmanveru: try it now, I've removed a ton of C code from ffi_util.c
06:54:58benburkert leaves the room.
06:56:11benburkert enters the room.
07:01:26voldor enters the room.
07:01:34voldorpant pant
07:02:10benburkert leaves the room.
07:04:20benburkert enters the room.
07:04:28rubuildius_ppcEric Hodel: 43b8516cd; 1616 files, 5540 examples, 19005 expectations, 0 failures, 67 errors; http://pastie.caboo.se/paste/156243
07:05:57rubuildiusEric Hodel: 43b8516cd; 1616 files, 5540 examples, 19081 expectations, 0 failures, 0 errors; http://rafb.net/p/AClXna93.html
07:09:13d2dchat leaves the room.
07:15:03defunkt_ enters the room.
07:15:14defunkt leaves the room.
07:18:33dewd leaves the room.
07:19:41ezmobius leaves the room.
07:20:48ezmobius enters the room.
07:26:41wycats enters the room.
07:34:55wifelette enters the room.
07:41:52headius leaves the room.
07:45:04rueHm hm
07:51:35aotearoa enters the room.
07:51:39aotearoa leaves the room.
07:59:02rubyconsumer enters the room.
07:59:09rubyconsumer leaves the room.
08:01:44geekounet leaves the room.
08:04:07geekounet enters the room.
08:15:35graza enters the room.
08:18:58headius enters the room.
08:18:58boyscout2 commits by Brian Ford
08:18:59boyscout * Preliminary changes to use instance_eval in MSpec.; 10a2dc0
08:19:00boyscout * Fixed Digest specs to pass with RSpec.; 714efa8
08:27:21pietia enters the room.
08:29:00rubuildius_ppcBrian Ford: 10a2dc058; 1616 files, 5540 examples, 19005 expectations, 0 failures, 67 errors; http://pastie.caboo.se/paste/156253
08:31:25rubuildiusBrian Ford: 10a2dc058; 1616 files, 5540 examples, 19081 expectations, 0 failures, 0 errors; http://rafb.net/p/Bfxu8K77.html
08:34:45graza_ enters the room.
08:34:46graza leaves the room.
08:35:29RyanTM leaves the room.
08:36:13hornbeck leaves the room.
08:36:30hornbeck enters the room.
08:36:32pergesu enters the room.
08:39:09brixenI've pushed an instance_eval branch that changes MSpec to use instance_eval instead of calling the block
08:39:45brixenthat causes 108 spec failures, mostly related to constants, but also method defs
08:39:52brixenif anyone wants to have a look
08:41:14graza_ leaves the room.
08:41:38mae leaves the room.
08:46:03manverudrbrain: well, thanks for the try, but it doesn't work yet
08:51:08perdix enters the room.
08:56:24manverudoesn't get it...
09:00:01manveruhmm
09:00:05manveruinteresting...
09:06:08manveru192.168.0.(1|24|129) work from 192.168.0.(0..255)
09:06:28manveru24 and 129 are in /etc/hosts
09:07:18manveru3 would be localhost
09:07:20graza enters the room.
09:08:04manverumy 32bit machine differs, the network is based on 192.168.1.*
09:08:37manverubut i'm not sure if that has any influence...
09:17:05pietia leaves the room.
09:17:48pietia enters the room.
09:20:15srbaker leaves the room.
09:20:57srbaker enters the room.
09:24:26dbussinkmorning
09:36:58graza_ enters the room.
09:36:59graza leaves the room.
09:38:33perdix leaves the room.
09:46:04dctanner enters the room.
09:46:19ezmobius leaves the room.
09:46:37perdix enters the room.
09:48:31wycats leaves the room.
09:48:51manveruoi dbussink
09:50:33dbussinkhi
09:51:23manverujust started getaddress for every IPv4 address...
09:51:40dbussinkthat's gonna take a while i guess :)
09:51:50manveru^^
09:52:24manverui'm collecting facts
09:52:32manveruto see what kind of bug that is...
09:52:48headius leaves the room.
10:00:23manveruseems like for one lookup /etc/hosts is read 4 times...
10:05:08manveruscratch that, 5 times
10:05:39perdiy enters the room.
10:21:23perdix leaves the room.
10:28:45dbussinkmanveru: well, that's probably glibc's doing
10:36:36pietia leaves the room.
10:37:50graza_ leaves the room.
10:44:34silex leaves the room.
10:46:45rudebwoy enters the room.
10:46:48dctanner leaves the room.
10:49:28chris2 enters the room.
10:54:37be9 enters the room.
10:54:40jtoy enters the room.
11:02:21sudoer leaves the room.
11:06:34langenberg enters the room.
11:19:13headius enters the room.
11:22:10headius leaves the room.
11:29:06headius enters the room.
11:34:06headius leaves the room.
11:34:40headius enters the room.
11:35:02perdiy leaves the room.
11:42:24sudoer enters the room.
11:42:28headius leaves the room.
11:43:07headius enters the room.
11:45:35hornbeck leaves the room.
11:49:10perdix enters the room.
11:52:25lopex enters the room.
12:00:37jtoy leaves the room.
12:16:36crossblaim enters the room.
12:17:40headius leaves the room.
12:19:29crossblaimhi
12:21:52dbussinkafternoon
12:30:03Zl0at5a enters the room.
12:33:33boyscout2 commits by Gianluigi Spagnuolo
12:33:34boyscout * Fix the Regexp.escape behaviour with tab and white space; 6b67a6c
12:33:35boyscout * Added some test to Regexp.quote to manage tab and white space; f8146d2
12:36:00_mutle leaves the room.
12:36:40mutle enters the room.
12:44:19rubuildius_ppcGianluigi Spagnuolo: 6b67a6cc4; 1616 files, 5540 examples, 19011 expectations, 0 failures, 67 errors; http://pastie.caboo.se/paste/156266
12:46:02rubuildiusGianluigi Spagnuolo: 6b67a6cc4; 1616 files, 5540 examples, 19087 expectations, 0 failures, 0 errors; http://rafb.net/p/P21wRa92.html
12:48:35lstoll leaves the room.
12:58:55headius enters the room.
13:19:31jeremydurham enters the room.
13:22:38jeremydurham leaves the room.
13:33:22langenberg leaves the room.
13:40:22pietia enters the room.
13:43:27cremesdbussink: any chance you can review ticket 349?
13:43:40dbussinki was just doing that :)
13:43:53cremesexcellent...
13:44:06dbussinkhmm, it results in failures on leopard intel
13:44:16cremesuh oh
13:44:38cremescan you pastie them?
13:45:04dbussinkcremes: http://pastie.caboo.se/156280
13:48:59cremesi see the problem... for some reason my changes to spec/ruby/1.8/core/string/unpack_spec.rb did not get saved
13:49:10cremesi must have goofed up something with git
13:49:19cremesthe endian guards i added are missing
13:49:30dbussinkyeah, well, i removed the configure file by hand for now
13:50:11cremesi'll fix and attach another patch i guess
13:52:34dbussinkhow much failures are left on ppc with this patch?
13:52:38dbussinkonly the truncate issues?
13:53:01cremestruncate and Digest::SHA
13:53:16dbussinkah ok, i think evan was working on fixing truncate
13:53:35dbussinkproblem is that the ffi code generation is broken for long long's
13:53:41cremesafter i knock out this pack/unpack stuff i'll tackle the others
13:54:00cremesi saw your conversation yesterday... ffi is a bit too deep into the system for me right now
13:54:27cremesi need to get more familiar with the codebase before i can tackle that
13:54:30dbussinkit's close to my edge too ;)
13:54:38cremesheh
13:54:50dbussinkproblem is not in our code, but in the library that is used for generating the code on ppc
13:55:04cremesi need some pointers about using git... should i branch again to make this unpack_spec patch?
13:55:25dbussinkcremes: perhaps the easiest way is to fix it all locally
13:55:30dbussinkthen create a diff
13:55:35cremesi did not delete my earlier branch
13:55:37dbussinkand apply that to a nice and clean branch
13:56:43cremesokay, so follow the git workflow again as shown here: http://rubinius.lighthouseapp.com/projects/5089/using-git
13:57:06dbussinkbasically make it work in your current branch
13:57:15cremesdoing it now...
13:57:16dbussinkcreate a patch file using git diff
13:57:29cremeswhy not git format-patch?
13:57:49dbussinkwell, it would probably include the configure changes too then
13:57:56dbussinkor did you clean that out?
13:57:57cremesi see
13:58:13cremesi did not... i didn't know if it was required or not so i left it in
13:58:30dbussinkthose basically never need to be updated
13:58:44cremesok, in the future i will strip that out
13:59:08dbussinkyeah can use git checkout to clean it up before committing
13:59:22dbussinkmy git knowledge is also quite some trail and error
13:59:52cremesok, so i'm on master right now... i need to create a new 'topic'/branch to fix these guards, right?
14:00:30cremesor are you suggesting i fix it in master and just generate a 'git diff'?
14:01:15dbussinkwell, my idea was to get it all correct in the branch you already have
14:01:23dbussinkto revert the configure change there etc.
14:01:36cremesoh... how do i revert the configure changes?
14:01:43dbussinkbut i don't really know what the easiest / best way is
14:02:50dbussinki think you can do it with git-revert revision path/
14:02:54dbussinkor something like that
14:03:01cremesi'll try that
14:08:36dbussinkcremes: do you have an intel system available to test with?
14:09:02cremesnot right now... i do have an osx intel laptop but i left it at work
14:11:09cremesdbussink: before i attach this patch to the ticket, check it out: http://pastie.caboo.se/156282
14:11:24cremesdbussink: let me know if it fixes all the intel issues
14:11:58dbussinkcremes: checking
14:13:56dbussinkhmm, still some failures
14:14:09cremesargh... please pastie
14:14:23dbussinkhttp://pastie.caboo.se/156283
14:23:07cremeshmmm... looks like i'm going to work to pick up my laptop!
14:23:33dbussinkno access to any other little endian system?
14:23:45dbussinki could give you access to a box
14:23:45cremesno
14:23:54cremesok
14:26:17headius_ enters the room.
14:26:17headius leaves the room.
14:36:53cremesdbussink: the ruby on that box is patchlevel 36
14:37:22dbussinkah, even forgot that
14:37:28dbussinkit usually runs all the specs ok :)
14:37:32dodecaphonic enters the room.
14:37:36cremesok!
14:40:35crossblaim leaves the room.
14:43:38dbussinkcremes: did you see the nice ascii art when you logged in? :P
14:43:38headius leaves the room.
14:44:16cremesdbussink: no, i didn't... the shell prompt is in color... that's all
14:44:37dbussinkah, too bad
14:55:46dodecaphonic leaves the room.
14:58:53dbussinkcremes: any luck with it?
14:59:12cremesdbussink: rsync is still running...
14:59:40cremesit's about 30% done
15:01:10nicksieger leaves the room.
15:05:38thehcdreamer enters the room.
15:12:38mutle leaves the room.
15:13:14mutle enters the room.
15:21:00hornbeck leaves the room.
15:21:17hornbeck enters the room.
15:26:59dbussinkcremes: making progress?
15:27:14cremesdbussink: yes, just built & ran the tests
15:27:26cremesi have mri passing cleanly but rbx has 3 errors
15:27:56dbussinkhmm, so a problem in rbx then
15:27:57dbussink?
15:28:09cremesyes, looking at it now
15:34:56cremesdbussink: i have it passing cleanly now... ONE line change! i hate bugs...
15:35:41dbussinknice
15:35:50dbussinkstill works in ppc? :P
15:37:10gdagley leaves the room.
15:37:21cremesyes!
15:38:23seydar enters the room.
15:49:37cremesdbussink: just went through the git checkout/git rebase/git merge shuffle... can i pastie my new patches for you to check?
15:50:08dbussinkyeah, i'll test to apply them against current head
15:51:18cremesdbussink: http://pastie.caboo.se/paste/156296 and http://pastie.caboo.se/paste/156297
15:51:51seydarhas anyone else had trouble building the latest version?
15:53:00dbussinkcremes: hmm, first one doesn't apply clean for me
15:53:55dbussinkcremes: this is against head? or should i apply the other patches first?
15:53:55cremesdbussink: git prob? i ran the commands as specified on lighthouse...
15:54:10cremesdbussink: apply others first
15:54:24cremesthese are cumulative, i guess
15:54:34dbussinkah ok
15:55:02cremesi tried to make one big patch, but the 'git pull' pulled in a bunch of ffi changes that were then included...
15:55:48mernen enters the room.
15:56:12seydardrbrain: did you fix that bootstrap.rba issue?
15:58:11dbussinkcremes: ok, i now have 4 commits, right?
15:58:37cremesdbussink: yes, would you like me to attach the last two to the ticket?
15:59:34dbussinkcremes: http://pastie.caboo.se/156298
15:59:54dbussinkthat's the failure i'm getting now
16:00:04dbussinki have all four commits applied
16:00:11cremesweird... hold...
16:02:38cremesrebuilding on spongebob... i probably forgot to copy something back to my ppc box
16:06:01cremesdbussink: yep, forgot to copy back one spec correction... should i do the whole branch/merge thing again to produce a 5th patch?
16:06:16dbussinkcremes: you can apply it to your branch
16:06:22dbussinkand create another patch for it
16:06:38crossblaim enters the room.
16:06:51dbussinki've got your patch in without the configure stuff
16:06:58dbussinkso it all works ok, i can push that stuff
16:08:40seydarcremes: when you rebuilt, did it go over a bootstrap.rba file?
16:08:56cremesseydar: let me look...
16:09:25seydardrbrain wrote that, and it was killing me last night.
16:10:10cremesseydar: i did a complete rebuild and it didn't complain about that file
16:10:31dbussinkseydar: i'm not having any problems either
16:10:33seydarok.... whew. didn't kill me.
16:10:39dbussinkseydar: with various platforms
16:10:43seydarthen drbrain fixed it. grats to him
16:10:54seydarit was last night when the bugs were out to play
16:11:02cremesworks on osx ppc too which usually sees *all* the breaks
16:11:18seydarcremes: 10.5 or 10.4?
16:17:00seydari'm running 10.4.
16:22:10seydaranyone having any problems with ipaddr and to_string?
16:22:26cremesseydar: 10.5.2
16:23:39seydarah. the (("%.32x" % addr).gsub!(/.{4}(?!$)/, '\&:')) is giving me a few odd things. like instead of 0s, it produces fs
16:24:10crossblaim leaves the room.
16:24:20binary42 enters the room.
16:25:00cremesdbussink: http://pastie.caboo.se/paste/156302 for the 5th patch... i really appreciate your help here
16:25:34d2dchat enters the room.
16:36:26dbussinkcremes: assume you verified with mri?
16:36:57twbray enters the room.
16:37:11cremesdbussink: i ran it through mri and rbx on both spongebob and my ppc box
16:37:30dbussinkcremes: ok, running mspec now
16:38:08dbussinkcremes: looks ok here
16:38:17cremeswhew!
16:40:27twbray leaves the room.
16:41:15dbussinkcremes: you could attach these patches for the record on lighthouse
16:41:20antares enters the room.
16:41:29cremesdbussink: working on that now...
16:41:49cremesdbussink: is it okay to just attach the links to pastie or should i upload the actual files?
16:42:33dbussinkpasties is ok, just reference to our discussion here or something like that
16:42:42cremesgot it
16:42:56boyscout5 commits by Chuck Remes
16:42:57boyscout * Fix unpack_spec expectation for little-endian byte ordering; 60bbc85
16:42:58boyscout * Fix unpack to use native host byte order for formats /ILQS/; 17e45ce
16:42:59boyscout * Removed some commented out / dead code from String#unpack; 42f5764
16:43:01boyscout * Add some missing endian guards to the unpack_spec; 1540cb7
16:43:01boyscout * Fix several Array#pack and String#unpack bugs related to byte ordering (endiannes); 8488676
16:43:36cremesmy first commits!
16:44:13be9congrats, cremes!
16:44:25cremesbe9: thanks!
16:44:29seydarcongrats!
16:45:07seydarhow can I set a bug to 'resolved' on lighthouse?
16:45:37seydarDigest is completely failing on TIger/PPC
16:46:12cremesseydar: i will be working on that next
16:46:35seydark.
16:46:50seydaranyone have anything they want me to test?
16:51:06cremesevan: 8488676, 1540cb7, 42f5764, 17e45ce, 60bbc85... ssh key == http://pastie.caboo.se/156284
16:55:28rubuildius_ppcChuck Remes: 60bbc8506; 1616 files, 5540 examples, 19007 expectations, 0 failures, 66 errors; http://pastie.caboo.se/paste/156314
16:56:03rubuildiusChuck Remes: 60bbc8506; 1616 files, 5540 examples, 19087 expectations, 0 failures, 0 errors; http://rafb.net/p/ahPzjm74.html
16:56:43seydarwhere is YAML::tagurize defined?
16:56:54cremesand now to start fixing the Digest problems on ppc...
16:58:30dbussinkseydar: you need permissions for resolving tickets
16:58:37sudoer leaves the room.
16:58:38seydarridiculous. YAML::tagurize is reportedly undefined, but I can see it being used
16:58:51seydardbussink: i'll just suck it up and submit it as a patch
16:59:02seydarspeaking of which....
17:00:28d2dchat leaves the room.
17:00:36d2dchat enters the room.
17:02:48seydarthat makes 3 patches from me... yet none have been accepted :-(
17:06:08seydaranyone work with the yaml code here?
17:06:40cremesseydar: keep plugging away... don't give up
17:07:07seydarcremes: i'm trying! but it's so annoying! where the crappzor is _why?
17:12:05dbussinkseydar: where are your patches then?
17:12:25seydardbussink: i sent them to the dev mailing list, like it said on the site
17:12:32seydarrubinius-dev@googlegroups.com
17:12:43dbussinkseydar: ah, put them up on lighthouse
17:12:50dbussinkseydar: that's where it gets read
17:12:56seydardbussink: i totally just deleted them. you're evil
17:13:26dbussinkand i haven't seen any activity on the mailing list either
17:13:53seydaryea, just noticed now they aren't being sent because i'm not a member of it
17:14:34dbussinkah well, anyway, best place to put patches is on lighthouse
17:15:09cremesdbussink: pretty sure i figured out why md5 is failing on ppc
17:15:46seydardbussink: did that
17:15:48dbussinkseydar: btw, stuff like #351 is already known, because there are specs for yaml that fail with the exact same error message
17:16:02seydar!@?>dsf damn you
17:16:09dbussinkhehe
17:16:13cremesdbussink: where in rubinius do we create #defines for use by external libs during compilation?
17:16:29dbussinkwhat is needed?
17:16:33cremesdbussink: i think md5 is failing because WORDS_BIGENDIAN is not defined
17:16:51cremesthere is an #ifdef WORDS_BIGENDIAN, #else branch in the md5 c code
17:17:12dbussinkcremes: hmm, isn't there a general platform guard for that stuf?
17:17:16dbussinkcremes: that we could use?
17:17:45twbray enters the room.
17:17:47cremesdbussink: i don't know what that means in this context; how does a guard get picked up by C code in a lib?
17:18:15dbussinkcremes: well, if there is something defined in POSIX that says whether a system is big or little endian
17:18:21dbussinkcremes: if there is, we should use that
17:18:26dbussinkbut i have to go
17:18:35dbussinkprobably back somewhere tomorrow
17:18:37cremessee ya... i'll look through POSIX stuff
17:19:17dbussinklater
17:19:33seydarcremes: try putting the code between the #ifdef and #else underneath the else
17:19:37seydarlater dbussink
17:19:59cremesseydar: trying that now
17:20:02seydark
17:20:10dctanner enters the room.
17:21:20cremesi made if #if 1 and ran the md5 specs; clean run on ppc but now it will be borked on x86
17:21:33cremesat least now i know where the problem is
17:27:05seydarhuzzah! ppc got shortchanged (again)
17:27:13seydarman i am so close to nailing this yaml problem
17:28:03twbray leaves the room.
17:28:11dctanner leaves the room.
17:29:12seydarwhere is the YAML code, besides in stdlib?
17:33:29seydarholy crap i did! cremes, i found the bug
17:34:03cremesseydar: cool!
17:34:26seydarthanks!
17:37:41RyanTM enters the room.
17:40:45seydarman, rubinius is the easiest thing to work with
17:43:23nicksieger enters the room.
17:50:47cremesi have a nice 5 line patch which fixes all the Digest errors on ppc; testing on x86 now
17:52:56seydarhoooly crap you are crazy
17:53:16cremesit helps to know C
17:53:27cremes:)
17:54:29cremesplus, i made all these bugs in my own co