Index

Show enters and exits. Hide enters and exits.

00:02:00zenspiderwho did the regexp engine for rubinius?
00:02:12jinjing enters the room.
00:02:15zenspideris that pulled from C or a ruby port of oni?
00:07:28zenspiderwho was just working on the syslog ffi stuff?
00:07:40drbrainseydar
00:07:50drbrain... Ary Brown
00:09:30zenspiderany idea how far he god?
00:09:33zenspidergot
00:09:43drbrainseems to work now
00:10:04zenspideroh. n/m. the bug is from him and it has a patch attached... he's got commit now right?
00:10:09zenspiderI'll assign to him
00:10:14drbrainyep
00:10:25brixenzenspider: one problem is that not everyone can close tickets
00:10:34brixenand I don't know exactly what perms it takes
00:10:43brixenI guess if you're invited to the project
00:11:25zenspideryeah. I can't assign to him... I'll see about promoting him
00:12:55drbraindoes anybody here have some familiarity with the compiler?
00:13:33zenspidera bit... what's up?
00:14:01zenspiderI'm kinda a post-workout tard still
00:14:04drbrainI'm trying to get order-of-operation to be correct
00:14:15drbrainso I went to Call#emit_args
00:14:27drbrainand changed from reverse_each to each
00:14:40drbrainbut then the args are backwards for calling
00:15:19drbrainI guess I've got to change the send_stack, etc opcodes too...
00:15:25zenspideris this the masgn bug? or something different?
00:15:35radarek leaves the room.
00:15:40drbrainsimilar to the masgn bug
00:16:05drbrainstuff = %w[a b]; p stuff.shift, stuff.first
00:16:23drbrainMRI reports a, b, rubinius a, a
00:17:14GMFlash leaves the room.
00:17:29zenspiderYAY! no unassigned open tickets!
00:18:06zenspideroh. that. that bug is gonna suck ass.
00:18:24Defileryes. yes it is.
00:18:58zenspiderI think it is gonna bite us a lot more than we think... it might be worth it to do the arm wrestling now and match ruby.
00:19:17drbrainI refactored part of RubyGems and hit it
00:20:02zenspiderI think evan did it reverse for efficiency reasons... I'd rather throw that out and be more compatible
00:20:19zenspiderbut.. I'm not gonna be able to help you today on that one... no freakin' way. :)
00:21:14drbrainI can't figure out how cpu_send_message works now
00:23:52enebo_ leaves the room.
00:25:19GMFlash enters the room.
00:30:37wmoxam leaves the room.
00:31:08nicksieger leaves the room.
00:33:39cremes enters the room.
00:33:49rubuildius_ppc enters the room.
00:35:12ragnard leaves the room.
00:40:03DefilerCan someone explain to me why we switch to RbYAML?
00:40:08Defilerswitched, rather
00:40:23DefilerBecause we spent a fair amount of brainpower going to syck, for good reasons
00:41:00drbrainDefiler: syck was 90% broken when I tried to use it
00:41:18drbrainas in, none of the fancy stuff that syck was supposed to provide worked
00:41:58DefilerOK
00:42:34drbrain#to_yaml_properties, in particular, IIRC
00:42:50drbrain#to_yaml/YAML.load worked, but not the custom serialization
00:43:28drbrainthere are other problems with RbYAML, so if we could go back to syck, and have the fancy stuff work, I'd be happy
00:44:45DefilerOK. I ask, because RbYAML is pretty damn buggy
00:44:53Defiler..and I was making sure it made sense to take the time to fix all that
00:46:37DefilerAlso, as I recall, RbYAML is unbelievably slow
00:47:36rubuildius_ppcBrian Ford: ba1affdd5; 1723 files, 5813 examples, 19808 expectations, 0 failures, 2 errors; http://pastie.caboo.se/paste/159582
00:47:47drbrainDefiler: well, it is pure-ruby
00:47:52DefilerI mean, beyond that
00:47:52drbrainrootprry
00:47:56drbraingah
00:48:02cremesyay! looks like libffi did cure the ppc ills
00:50:10drbrainyay! I have a FreeBSD-7 paralells image
00:50:10cavalle leaves the room.
00:51:25brixencremes: would you be able to investigate those 'characterSpecial' spec failures? perhaps we need a platform guard
00:51:38brixencremes: unless we've messed up the st_mode flags
00:51:50brixenDefiler: yeah, we should get syck running properly
00:51:50cremesbrixen: yessir; i will check into them later tonight
00:52:01brixencremes: awesome, thanks!
00:52:55brixenI think I'll take a shot at integrating 1.9's strtod
00:53:05Defilerbrixen: I agree. Want to help?
00:53:06brixenit's just such a huge mountain of ugly C code
00:53:10brixenDefiler: sure
00:53:11Defilerbrixen: Or are you saying you are on it?
00:53:20brixenheh, I'll help ya
00:53:28brixenjust saying, we should use syck imo
00:54:51brixenDefiler: the 'huge mountain..' comment was for 1.9's strtod, not sych
00:54:55brixenack
00:55:06brixenanyway, in case that was confusing
00:55:13brixenbreak-time, I can't type
00:56:09headius leaves the room.
00:58:36jinjing leaves the room.
01:04:49wifelette leaves the room.
01:06:37olabini leaves the room.
01:12:28agile enters the room.
01:13:56headius enters the room.
01:17:42ttmrichter_ enters the room.
01:24:43ttmrichter leaves the room.
01:28:19AndrewO enters the room.
01:29:37olabini enters the room.
01:41:00headiusevening
01:51:58ezmobiushola
01:57:47imajes leaves the room.
01:59:01headius_ enters the room.
01:59:01headius leaves the room.
02:05:48evanevening.
02:07:03headiushiya evan
02:07:57evanhey
02:08:44headiusso why did you have to switch to libffi?
02:09:22evanlightning doesn't support long long at all
02:09:37evanso, i'd have to hobble getting that type working on whatever platforms we wanted
02:09:55evani've used libffi in the past
02:09:57evana few years ago
02:10:25headiusahh, I see
02:10:29evani wanted to go the stub route from the beginning because, initially, the code was cleaner, and my thinking was it was more performant
02:11:04evanbut with all the problems, I decided to just go the libffi route for now
02:11:08evanfor all platforms
02:11:19evanwe can reinvestigate the stub route later
02:11:28evanthe stubs are, atm, a failed experiment.
02:12:24headiusthat's too bad
02:12:35RyanTM leaves the room.
02:12:42evanwe learned quite a bit in the process of working on them
02:12:43headiuswhat did you need long long for?
02:12:47evanso the knowledge is for not
02:12:55evaner. not for not.
02:12:57evananyways.
02:13:06RyanTM enters the room.
02:13:08evanwe needed long long because off_t on darwin (and linux) is uint64_t
02:13:26evanand most of the file APIs use off_t
02:13:34brixenevan: did you have any useful dtrace scripts you cooked up?
02:13:40evanno
02:13:44brixenok
02:13:47evanmy plan is to work on that at the sprint
02:13:54brixenbut the probes we have work ok, yes?
02:14:02evanyep
02:14:06brixenok cool
02:14:42therealadamevan: ping?
02:15:18headiuswayne meissner from jna is toying around with jna improvements...he mentioned an interest in lightning, so I'll pass along the long long info to him
02:15:24evantherealadam: sup yo
02:15:31therealadamevan: pm you?
02:15:31evantherealadam: got your emizle
02:15:35evango for pm
02:15:37therealadamahh excellent
02:17:12headiushey, what do you guys use for live profiling
02:17:28agile leaves the room.
02:18:04evanheadius: yeah, lightning has problems
02:18:09evanheadius: also, it's GPL v3 now
02:18:31tarcieriwtf?
02:18:39tarcierilightning was lgpl before, right?
02:19:06agile enters the room.
02:19:40evanyep
02:20:20tarcieriwhy would they do that?
02:20:52evanno clue
02:21:17imajes enters the room.
02:23:15headiusanyone? live profiling?
02:23:22headiusI don't know my C tools all that well
02:23:30evani use shark to profile C
02:23:33evanit's the bomb.
02:24:02VVSiz_ enters the room.
02:24:40brixenevan: can I bump this up on your list? http://rubinius.lighthouseapp.com/projects/5089/tickets/354-bug-eval-instance_eval-fails-to-correc tly-modify-toplevel_binding
02:25:07evanyou'd like me to reprioritize and make this my top action item? </fight club>
02:25:13brixenheh
02:25:37evansure, i can look now.
02:25:41evanwhats the issue?
02:25:44brixencool, thanks
02:26:15headiusdoes shark do live profiling?
02:26:16brixenevaling something like 'include Z' in the toplevel binding
02:26:22headiusI thought it only did snapshots
02:26:24evanheadius: oh, no.
02:26:29evanalmost nothing does live profiling.
02:26:36headiusoh, hmm
02:26:50tarcieriby "live profiling" do you mean something ala dtrace?
02:26:52headiusok...I've just gotten used to live profilers on the JVM
02:27:17headiusI mean I can sit and watch invocation counts tick and see percentages change over the run of an app
02:27:21brixenheadius: what do you use? a particular one or something in the jvm?
02:27:22evanC profilers are built with zero runtime in mind
02:27:25headiusdtrace probably has a way
02:27:26evanyou could do it with dtrace probably
02:28:06headiusI've been using the live profiling tools for JRuby because I want to cut out the first 10 seconds of warmup and so on
02:28:17brixenheadius: this was interesting: http://searchsecurity.techtarget.com/news/article/0,289142,sid14_gci1301804,00.html
02:28:48headiusahh, yes
02:29:05brixenheadius: live profiling tools 'for' jruby on 'in' jruby, do you have tooling impl in jruby?
02:29:18evanheadius: the jvm startup or the jit warmup?
02:29:32headiuswe don't have to have tooling support in jruby
02:29:44headiusevan: bit of both
02:30:15headiusas JRuby runs I can see ruby methods enter the profiling results
02:30:27brixenheadius: do we have to play 20 questions or can you point me to docs on what you use? :P
02:30:36headiusbecause they compile and become just another Java method
02:30:50headiusI've been using NetBeans, YourKit, JProfiler at various times
02:30:56brixenheadius: ah ok
02:30:58headiusthere's probably a dozen "good" tools
02:31:08brixenheadius: my next question was animal or mineral :P
02:31:23evanbigger than a bread box?
02:31:28headiussampling, full timing, inclusive/exclusive filtering of code
02:31:42imajes leaves the room.
02:31:45evanheadius: so, the live profilers use sampling?
02:31:55headiusthey can go either way
02:32:07headiusI generally don't use sampling because it's usually wrong
02:32:18evanheadius: oh, another question: the part that displays the data is running an a seperate jvm, using the jvm's C profiling hooks, yes?
02:32:32headiusno, over a socket, using JVM tooling APIs
02:32:44evan2 jvm's then
02:32:47headiusyew
02:32:48headiusyes
02:32:59evanwhats the different between the C profiling hooks and the tooling API?
02:33:04evanurls are fine
02:33:21headiustooling API can connect across the network to a remote machine the same as to a local JVM
02:33:43headiusthe profiling hooks run in the target JVM...tooling API just feeds information across
02:33:47evanis the tooling API on top of the C hooks then?
02:34:14headiusI'm not sure at what level it gets into C
02:34:47evanhttp://java.sun.com/javase/6/docs/technotes/guides/jvmti/
02:34:49evanthat?
02:34:52headiusyes
02:34:54evanthat looks to be at C level
02:35:02evanoh oh
02:35:03evannm.
02:35:06evanby native, they mean java.
02:35:31headiuswell, there's a C side and a Java side
02:35:35headiusyou can access it from either
02:35:47evangotcha
02:35:51headiusit's an implementation detail whether the guts are java or C
02:35:56evanlooks like a fun airplane project
02:36:04evanputs it on his list for the flight sunday night
02:36:41headiuschad and rich implemented some other protocol in Ruby at one point
02:36:48headiusthe protocols are usually pretty straightforward
02:37:09evanyeah
02:41:17VVSiz leaves the room.
02:41:45nicksieger enters the room.
02:55:49evanah bugger.
03:00:28_mutle enters the room.
03:02:01jptix leaves the room.
03:08:09mutle leaves the room.
03:11:33benburkert enters the room.
03:24:26aotearoa enters the room.
03:24:35benburkert leaves the room.
03:24:41kevwil enters the room.
03:25:48benburkert enters the room.
03:26:40imajes enters the room.
03:33:15therealadam leaves the room.
03:36:30yipstar leaves the room.
03:41:04rue
03:41:12rueOps
03:43:30VVSiz enters the room.
03:45:59rueevan: When you have a moment, I want to see what if anything you have been plotting along the lines of `shotgun/rubinius --cluster 2x5 --require 'httpserver'`
03:46:34rueevan: Also, should I reserve days in mid/end March?
03:46:41rueOh
03:47:23rueAnd if anyone has Mac Apps You Cannot Live Without, suggestions appreciated. I typically prefer keyboard/cli over GUI, also moonlit walks on the beach
03:47:37rueQuickSilver has been suggestimated
03:48:59imajes leaves the room.
03:49:51DefilerQuicksilver is great.. SteerMouse is required if you are picky about your mouse pointer movement
03:49:55Defiler(though it sounds like you are not)
03:50:10DefilerYou will want to install Perian for video playback
03:50:20kevwil leaves the room.
03:50:54DefilerOnyX is a good adjust-weird-MacOS-setting tool
03:51:16DefilerSpirited Away is a must-have as well, if you have a laptop
03:51:27Defiler(hides unused windows automatically, based on your preferences)
03:51:59imajes enters the room.
04:01:41VVSiz_ leaves the room.
04:14:27jinjing enters the room.
04:17:20MenTaLguY enters the room.
04:21:45headius_ enters the room.
04:21:45headius leaves the room.
04:32:35rueSpaces works nicely enough
04:57:16headius leaves the room.
04:59:41be9 enters the room.
05:06:55ezmobius leaves the room.
05:09:12imajes leaves the room.
05:13:35jinjing leaves the room.
05:21:49benburkert leaves the room.
05:27:32evanrue: whats --cluster do?
05:27:48evanthe answer is likely no, nothing plotted
05:28:32rueevan: I was thinking offering a builtin server/distributed/cluster processing mode
05:29:58AndrewO leaves the room.
05:30:02rueevan: Sorta musing, started with figuring that it seems wasteful to reimplement the daemon backend separately with libev since Rubinius itself is based on that model
05:31:19GMFlash leaves the room.
05:31:19mernen leaves the room.
05:31:19srbaker leaves the room.
05:31:19gdagley leaves the room.
05:31:19d2dchat leaves the room.
05:31:19context leaves the room.
05:31:19zuwiki leaves the room.
05:31:19mass leaves the room.
05:31:19mojombo leaves the room.
05:35:36evani don't see a reason for it to be 'builtin'
05:38:25GMFlash enters the room.
05:38:25mernen enters the room.
05:38:25srbaker enters the room.
05:38:25gdagley enters the room.
05:38:25d2dchat enters the room.
05:38:25context enters the room.
05:38:25zuwiki enters the room.
05:38:25mass enters the room.
05:38:25mojombo enters the room.
05:39:21mass leaves the room.
05:39:21zuwiki leaves the room.
05:39:21context leaves the room.
05:39:21gdagley leaves the room.
05:39:21srbaker leaves the room.
05:39:21GMFlash leaves the room.
05:39:21mernen leaves the room.
05:39:21mojombo leaves the room.
05:39:21d2dchat leaves the room.
05:39:24mojombo enters the room.
05:39:26GMFlash enters the room.
05:39:31context enters the room.
05:39:32mass enters the room.
05:39:32zuwiki enters the room.
05:39:35mernen enters the room.
05:41:49dewd enters the room.
05:46:22mae leaves the room.
05:56:09mernen leaves the room.
05:57:25srbaker enters the room.
06:05:05rueevan: Mm, "builtin" meaning "ships with." I just hate to reproduce the arch if I can reuse/improve it in Rubinius instead. Maybe an extension? Do you know if anyone has tried to embed Rubinius or "drive" Rubinius through the C side?
06:05:17rueMy machines are a mess except for the Apache work
06:12:25ezmobius enters the room.
06:21:05jluxenberg enters the room.
06:46:21evanrue: sure, reusing the infrastructure we already have is a big plus
06:46:57ezmobiushey evan how was the cruise?
06:47:11evanrue: no, no one has embedded rubinius yet (to my knowledge)
06:47:12evanezmobius: fun!
06:47:53evanezmobius: a little tedious at times, but great
06:48:00ezmobiuscool
06:48:53evanhows things for you?
06:49:05dkubb enters the room.
06:49:05ezmobiusgood, crazy busy
06:49:17ezmobiusjust got a test cluster built here at the office
06:50:14evanoh nice.
06:50:52ezmobiusnow we can do TDD(test driven deployment)
06:50:57ezmobius:P
06:52:27evanheh
06:54:47drbrainevan: I'm trying to fix order-of-operations
06:55:12drbrainI've got the appropriate changes to the compiler, but can't figure out where args get turned into locals
06:55:20drbrain(specifically for calls)
06:56:28evanwhich order-of-operations?
06:56:54drbrainarr = %w[a b]; Array[arr.shift, arr.first]
06:57:00evanyikes
06:57:03evanit's not possible.
06:57:07drbrainshould be %w[a b], is %w[a a]
06:57:07evanwithout adding new instructions.
06:57:37drbrainif we know the arg count, why can't we fill in the locals backwards?
06:57:43drbrainor does that not happen?
06:57:45evanfill in the locals where?
06:57:50evanin the VM?
06:57:52drbrainI don't know
06:57:57drbrainyeah
06:58:02evanit's not in the VM
06:58:07evanit happens in bytecode.
06:58:14drbrainat least, that's what I was expecting to find, but it got mysterious
06:58:25evanbecause it's not there.
06:58:41evanevery method emits bytecode to pull args into locals itself
06:58:52drbrainah, ok
06:59:06evanand it expects the args to be on the stack, top of the stack == 1st arg
06:59:17drbrainyeah
06:59:26evanso changing this is radical
06:59:47drbrainI think it's time
07:00:26drbrainI will go look at def next
07:00:40evanwell, since it's radical, I think we need to fix the max arg count problem at the same time
07:01:08wifelette enters the room.
07:03:03evanit's not as easy as just running the code to assign locals from the stack in reverse order
07:03:08evansadly
07:03:21evanbecause defaults have to be run left to right as well
07:03:27drbrainI didn't expect it to be easy
07:03:37evanand defaults have to be able to access lvars to their left
07:03:53evanwell, whats your idea?
07:04:21drbrainI don't know enough about the compiler to have anything more than the most basic ideas
07:05:41evanso, the simplest that works, is to alter the compiler to emit code left to right for the args
07:05:59evanthen rotate the values on the stack inside the check_args instruction
07:06:27evanso that the called method still sees the args in the top => 1 order
07:06:59evandrbrain: use ./shotgun/rubinius describe
07:07:07evanto get a feel for the bytecode output for handling args
07:07:29drbrainthe call-side part was clear
07:07:39drbrainbut the recieve-side part wasn't
07:07:54evanthe receive side is actually easy
07:08:06evanit's just that the method expects the args are on the stack already
07:08:09evanthats all.
07:08:23drbrainfor an Array, I just changed reverse_each to each
07:08:39drbrainI haven't tracked down the other things, though
07:09:00evanhow do ya mean, for an Array?
07:09:00drbrainConcatArgs and Splat
07:09:33evanthose are a whole nother thing
07:09:35drbrainif @arguments.kind_of? Array then @arguments.each do |x| x.bytecode(g) end
07:09:43drbrain~ line 1820 of bytecode.rb
07:10:25evanyeah, thats what I mean
07:10:27drbrainalso, can we get some RDoc in the compiler?
07:10:50drbrainI'm groping in the dark alot
07:11:00evanConcatArgs and Splat are going to be uggers.
07:11:06evansure, add rdoc as much as ya want
07:11:17evanby adding, you mean me adding rdoc, yes?
07:11:20drbrainI'm not fully qualified yet
07:11:53drbrainwell, you or whoever else that's qualified
07:12:37evansure
07:12:43drbraineven something like "Call emits the arguments then the reciever then the send" would help some
07:13:13evanso mainly, you're looking for how they effect the stack
07:13:20drbrainactually, most helpful would be Generator
07:13:23evanyou should be looking at the instruction docs
07:13:42evanbecause it's easy to work backwards from the instruction docs to the Generator
07:13:57drbrainthey're not always 1:1
07:13:58evanbut no problem, i'll add rdoc to the Generator
07:14:07evanno, they're not, but largely they are
07:14:30drbrainfor the 1:1 ones, See: Instructions#blah would be fine
07:14:43drbrainor whatever the class is, I forget
07:14:56jtoy enters the room.
07:15:09evanright
07:16:24evanwell, take a crack at it
07:16:27evanif you want
07:16:33evani've been mulling it over for a while now
07:16:37drbrainI do
07:16:47evani'd planned on just doing it at the sprint
07:16:52drbrainthe implementing
07:16:55evansince you'll be there, perhaps that would be a good place
07:17:02drbrainok
07:17:11drbrainI'll wait, and unfactor RubyGems then
07:17:20evanunfactor?
07:17:24drbrainbut, I suspect we'll not get enough time to work on it
07:17:34GMFlash leaves the room.
07:17:49drbrainyes, I refactored part of RubyGems tar stuff, since it's terrible and untested and causing problems with Rubinius
07:17:50evanwell, just you and I will have some time outside the sprint
07:17:53evanto pair on it
07:18:00drbrainbut, the nice, simple refactored code runs into the order-of-operations problem
07:18:03evanso bring code and ideas
07:18:32drbrainhell, the one-liner is a good enough bit of code to fix it
07:18:46drbrainI chose shift/first instead of shift/shift, so the problem is more-clear
07:18:49evanthe reverse_each => each?
07:18:59drbrainno,
07:19:12drbrainarr = %w[a b]; Array[arr.shift, arr.first]
07:19:18evanoh, you mean thats a good enough case
07:19:21evangotcha
07:19:25drbrainyeah
07:19:51drbrainthat's the smallest test case I can think of
07:20:17evanarr = %w[a b]; p(arr.shift, arr.shift)
07:20:22evanis the testcase i've used before for this
07:20:35drbrainbut, not good inside specs
07:20:45evanvery true
07:27:32drbrainMonday I'll remove the zip code
07:29:50evanok, sweet.
07:29:56evani've got that on the todo list.
07:30:02evani'm ripping out lightning atm.
07:30:31drbrainI decided to leave in a transition period, just in case
07:30:47tarcieriit sucks you can't use llvm
07:38:50aotearoa leaves the room.
07:39:34boyscout2 commits by Eric Hodel
07:39:35boyscout * Fix some uninitialized ivar warnings; 934dfd4
07:39:36boyscout * Trim trailing whitespace; 1c8080c
07:45:32peglegrot enters the room.
07:46:06evandrbrain: so, i've been reviewing how others handle call convention wrt to args
07:46:34drbrainyeah?
07:47:15evanseems like all VM based languages have the VM apply to the args into locals itself
07:48:18drbrainwell, considering we're trying to get rid of a C-based VM, I prefer changing the bytecode
07:48:20evannone do it the way we do currently, ie, just have the args on the stack when the method is called
07:48:42drbrainyeah
07:49:03evanwell, it's a trade off
07:49:48evanbecause if we do it in bytecode, then we might have to add more VM code to do it correctly
07:49:57rubuildius_ppcEric Hodel: 934dfd43a; 1723 files, 5813 examples, 19808 expectations, 0 failures, 2 errors; http://pastie.caboo.se/paste/159697
07:50:24evanwhere as putting the args directly into locals is probably a smaller amount of VM code
07:50:34drbrainok
07:51:35drbraincrap
07:51:40drbrainruby 1.9.0-1
07:51:56drbrainhad I known, I'd have imported my latest RubyGems changes
07:51:59evanwe'll still have bytecode to process the defaults
07:52:00drbrainon Wed.
07:52:07evanwhat happened?
07:52:21drbrainbugfix release for 1.9.0
07:52:39evanah, ack.
07:52:42drbrainand, on Wed. I finished up a bunch of fixes for RubyGems
07:56:50evanhm, does 1.8 have a max number of args?
07:57:39tarcieriif it does I've never hit it...
07:57:40tarciericl
07:57:42tarcierigeh
07:58:57tarcieri>> foo(*(['X'] * 16777216))
07:58:57tarcieriIllegal instruction
07:59:00tarcieriheh
07:59:07TheVoice leaves the room.
08:03:56drbrainI don't believe it does
08:10:07evanhuh!
08:10:14evani finally found out what the NODE_ALLOCA is for!
08:11:00drbrainoh?
08:11:26evanwhen you call a method, if the arguments were simple, ie:
08:11:29evanblah(1,2)
08:11:48evanit calls alloca() to create a buffer to hold 1 and 2
08:11:59evanand, since 1 and 2 might be real objects, it has to be sure the GC sees them
08:12:16evanso it makes a new node, ie, NODE_ALLOCA
08:12:29evanand shoves the pointer it got from alloca() into it
08:12:37drbrainnice
08:12:52drbraintomorrow I should have a FreeBSD 7.0 parallels image
08:13:02evanand, the a pointer to the new node is on the stack, so the ruby GC sees it
08:13:05evanand keeps the arguments alive.
08:13:24evanit's funny how much they've hacked around the GC architecture
08:13:53drbrainwhich is also why it hasn't been changed
08:14:26evanyep
08:15:08drbrainexercisez
08:42:33GMFlash enters the room.
08:52:33peglegrot leaves the room.
09:04:03RyanTM leaves the room.
09:04:05GMFlash leaves the room.
09:04:23GMFlash enters the room.
09:07:11crossblaim enters the room.
09:08:47rueWoo woo
09:11:45cavalle enters the room.
09:25:56GMFlash leaves the room.
09:27:45ezmobius leaves the room.
09:29:04peglegrot enters the room.
09:35:41wifelette leaves the room.
09:35:54wycats leaves the room.
09:44:55sfaxon leaves the room.
09:46:36rueSleepy, be back
09:49:07GMFlash enters the room.
09:58:58Fullmoon enters the room.
10:21:26GMFlash leaves the room.
10:35:03rubuildius_ppc leaves the room.
10:45:54w1rele55 enters the room.
10:50:07msimbartl enters the room.
11:01:23crossblaim leaves the room.
11:04:24thehcdreamer enters the room.
11:05:16hassox enters the room.
11:07:04Fullmoon leaves the room.
11:21:31langenberg enters the room.
11:23:42aotearoa enters the room.
11:24:25ragnard enters the room.
11:25:05langenberg leaves the room.
11:25:37langenberg enters the room.
11:31:40headius enters the room.
11:32:38aotearoa leaves the room.
11:50:53Fullmoon enters the room.
11:55:14langenberg leaves the room.
11:56:22langenberg enters the room.
11:58:13wycats enters the room.
11:58:47Fullmoon leaves the room.
12:24:01crossblaim enters the room.
12:39:38crossblaim leaves the room.
12:40:36thehcdreamer leaves the room.
12:48:50radarek enters the room.
12:51:04mediogre enters the room.
13:26:52jptix enters the room.
13:29:38joachimmjp_tix: hello.
13:29:53jptixhi
13:36:52w1rele55 leaves the room.
13:42:00imajes enters the room.
13:53:14ttmrichter_ leaves the room.
13:54:00ttmrichter_ enters the room.
14:09:06Phoop leaves the room.
14:09:58chris2 enters the room.
14:13:13Phoop enters the room.
14:15:45ctennis enters the room.
14:28:59agile leaves the room.
14:35:09msimbartl leaves the room.
14:36:01crossblaim enters the room.
14:36:25crossblaimhi
14:51:05nicksieger leaves the room.
14:51:08AndrewO enters the room.
14:52:40AndrewO leaves the room.
14:54:32enebo enters the room.
14:58:56dodecaphonic leaves the room.
14:59:41jinjing enters the room.
15:09:23crossblaim leaves the room.
15:10:07headius leaves the room.
15:19:14jinjing_ enters the room.
15:20:54jinjing leaves the room.
15:22:26headius enters the room.
15:22:51headius leaves the room.
15:22:59rubuildius_ppc enters the room.
15:23:23headius enters the room.
15:28:36wmoxam enters the room.
15:29:02jinjing_ leaves the room.
15:35:43rubuildius_ppcEric Hodel: 934dfd43a; 1723 files, 5813 examples, 19808 expectations, 0 failures, 2 errors; http://pastie.caboo.se/paste/159758
15:38:09msimbartl enters the room.
15:41:56langenberg leaves the room.
15:46:06gnufied leaves the room.
15:52:02jtoy leaves the room.
15:59:20wmoxam leaves the room.
16:25:21boyscout1 commit by Chuck Remes
16:25:22boyscout * Fixes a race condition on OSX when "find"-ing character devices; 25cfa6a
16:26:43crossblaim enters the room.
16:29:19thehcdreamer enters the room.
16:29:36mediogre leaves the room.
16:37:57langenberg enters the room.
16:40:29rubuildius_ppcChuck Remes: 25cfa6a96; 1723 files, 5813 examples, 19808 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/159781
16:40:49cremesyes! no more errors on OSX PPC!
16:50:27GMFlash enters the room.
16:59:34imajes leaves the room.
17:01:49headius leaves the room.
17:02:19headius enters the room.
17:19:52GMFlash leaves the room.
17:23:18mernen enters the room.
17:23:44GMFlash enters the room.
17:24:04thehcdreamer leaves the room.
17:24:08obiejuan enters the room.
17:25:54agile enters the room.
17:26:04headius_ enters the room.
17:26:11headius leaves the room.
17:26:20headius_ leaves the room.
17:26:49headius enters the room.
17:29:03benburkert enters the room.
17:30:27radarek leaves the room.
17:33:17enebo leaves the room.
17:35:14obiejuan leaves the room.
17:39:38crossblaim leaves the room.
17:41:58dodecaphonic enters the room.
17:50:02jptix leaves the room.
17:51:25Ingmar leaves the room.
17:51:58Ingmar enters the room.
17:52:17crossblaim enters the room.
17:53:45d2dchat enters the room.
18:01:34wmoxam enters the room.
18:03:44nicksieger enters the room.
18:03:49RyanTM enters the room.
18:04:12jptix enters the room.
18:09:38crossblaim leaves the room.
18:14:25msimbartl leaves the room.
18:15:27jptix_ enters the room.
18:16:53peglegro1 enters the room.
18:16:53peglegrot leaves the room.
18:17:35benburkert leaves the room.
18:19:12benburkert enters the room.
18:28:37benburkert leaves the room.
18:34:01VVSiz_ enters the room.
18:36:51benburkert enters the room.
18:40:38VVSiz leaves the room.
18:40:45jptix leaves the room.
18:41:11VVSiz enters the room.
18:43:27cavalle_ enters the room.
18:46:37sfaxon enters the room.
18:47:39crossblaim enters the room.
18:50:26cremes leaves the room.
18:54:04GMFlash leaves the room.
18:55:04VVSiz_ leaves the room.
18:59:29cavalle leaves the room.
18:59:48nemerle enters the room.
19:00:12jacen_ leaves the room.
19:04:12srbaker leaves the room.
19:05:05FoobarWidget enters the room.
19:08:39srbaker enters the room.
19:12:46GMFlash enters the room.
19:15:58jluxenbergthere's a bug in the Marshal module
19:16:35jluxenbergI'm trying to figure out how to fix it but its not immediately obvious.
19:17:35crossblaim leaves the room.
19:18:06brixenjluxenberg: do you have a spec that demonstrates the bug?
19:18:31dkubb leaves the room.
19:18:44jluxenbergthis code fails: http://pastie.caboo.se/159837
19:18:54jluxenbergI can make a spec in a minute, never used rspec before
19:19:34brixenahh subclassing Hash
19:19:46jluxenbergthe offending piece of code is the first line of Marshal::State#construct_hash
19:19:47brixenyeah, wouldn't surprise me if there were issues
19:21:01jluxenbergif I make a spec will they put it in the test suite?
19:21:56brixensure
19:22:49brixentake a look at spec/ruby/1.8/core/marshal
19:23:39jluxenbergok
19:24:20brixenin that dir, fixtures/marshal_data.rb there's already a class UserHash < Hash
19:24:33jluxenbergah
19:24:35brixenyou could probably add the def initialize there and update the expected value below
19:24:46jluxenbergyeah, good idea
19:25:05jluxenbergwhere do I send patches? I guess I need to get the GIT version of the code as well
19:25:24brixenyeah, best is get git, git clone, and use git-format-patch
19:25:28brixenone sec..
19:25:49brixenhttp://rubinius.lighthouseapp.com/projects/5089/using-git
19:25:56jluxenbergah, good
19:25:56brixensome other docs there on specs
19:26:09brixengotta run some errands, but somebody will probably be here to help ya
19:26:55peglegrot enters the room.
19:27:17jluxenbergcool, thanks!
19:27:24brixennp, thank you!
19:33:55rudebwoy leaves the room.
19:35:19srbaker leaves the room.
19:38:03enebo enters the room.
19:39:14chris2 leaves the room.
19:42:41Defilerjluxenberg: The way this works is.. you 'clone' the main repository.. make changes.. and then use 'git commit' to commit them to YOUR copy
19:43:03Defilerjluxenberg: Then you use git format-patch to make an emailable patch out of the difference between your repository and the main one
19:43:10jluxenbergah, I see
19:43:16peglegro1 leaves the room.
19:43:24jluxenbergDefiler: can I skip the clone step if I'm only making a small change?
19:43:26DefilerThat way, when I (or someone else) accepts it to the main repository, you get credit
19:43:38DefilerWell, you have to clone to have a copy of rubinius
19:44:04DefilerSo, yeah.. even for a small change, if you downloaded the tarball, you will need to fetch the latest version from git
19:44:45jluxenbergDefliner: oh, what I meant was do I have to "checkout"
19:45:04Defilergit clone == svn co
19:45:10Defiler..and yeah, you need a working copy
19:45:38jluxenbergDefiler: in the workflow description it says to git checkout (kind of like a branch?), make change, commit to my local copy, then "rebase" my branch?
19:45:45rudebwoy enters the room.
19:46:33rudebwoy leaves the room.
19:46:34DefilerOh, I see what you are asking
19:46:42DefilerNo, you don't need to bother with that, though it is pretty easy
19:46:47jluxenbergokay
19:47:04DefilerWhen you have direct committer access, you will want to do that
19:47:11Defiler..but just for producing a patch, it is not necessary
19:47:25rudebwoy enters the room.
19:48:23jluxenbergworking on it now
19:54:36wmoxam leaves the room.
19:57:54boyscout1 commit by Wilson Bilkovich
19:57:55boyscout * Implement UNIXSocket#peeraddr and UNIXSocket#addr; 7c01bea
20:02:59srbaker enters the room.
20:03:46jluxenbergcan anyone explain this syntax to me? ah, I see,
20:03:53jluxenberghaha oops, wrong paste.
20:04:00jluxenbergthis: n = class << o2;ancestors;end
20:04:05jluxenbergwhere o2 is some object
20:07:55jluxenbergis it the same as o2.class.ancestors ?
20:08:07peglegro1 enters the room.
20:08:11DefilerThat is returning the ancestors of o2's metaclass
20:08:21DefilerWhich generally will be the same list as o2.class.ancestors, but not always
20:08:48jptix enters the room.
20:09:14jluxenbergah, I see
20:09:27jluxenbergin this case, the class has been extended so the metaclass is different from the class
20:09:42Defilerjluxenberg: e.g. http://pastebin.com/m267753c1
20:10:04jluxenbergright, cool
20:10:18rue Morn
20:10:30Defilero7
20:13:28Fishy5 enters the room.
20:14:38jluxenbergdoes anyone know if rake has a -j option like make?
20:15:26DefilerIt does not
20:15:51DefilerIt has a 'parallel task' type that you can use, but it just satisfies dependencies in parallel, it doesn't really try to thread it all up
20:16:17jluxenberghmm, that sounds similar though. make just spawns multiple processes in parallel
20:16:39rubuildius_ppcWilson Bilkovich: 7c01bea10; 1723 files, 5813 examples, 19808 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/159853
20:16:51DefilerIt isn't the same, sadly
20:16:57DefilerIt wouldn't be hard to add, though
20:16:59Toba enters the room.
20:17:04wmoxam enters the room.
20:17:23DefilerWhat we should probably do, though.. is expose a JOBS environment option
20:17:31DefilerSince 'rake' is just calling out to make here
20:17:36jluxenbergyeah, fair enough
20:17:45jluxenbergwhen you say it satisfies them in parallel, what do you mean? does it spawn a bunch of rake processes?
20:18:17skaar enters the room.
20:18:53benburkert leaves the room.
20:18:54DefilerIt doesn't, no. As I recall, it just makes some threads to execute the task objects
20:19:08DefilerIt has been a while since I looked at it.. I know we explored it when desigining Vlad
20:19:20benburkert enters the room.
20:19:38benburkert leaves the room.
20:19:44jluxenbergI'm running rake spec:core, and I get an error that says "Unable to encode stream: <long obj.inspect dump>"
20:20:37DefilerYeah, that is known/expected
20:20:43DefilerWe don't pass everything in spec/core right now
20:20:53Defiler./bin/mspec ci should pass without errors, though
20:21:06Defiler(which I believe is still aliased as 'rake spec', but I could be wrong)
20:21:29jluxenbergah, so that error is from a spec test?
20:21:51jluxenbergthe weird thing is the rake process never finishes, I have to control-c to get it to end with a seg fault
20:21:59DefilerYeah, that is also known
20:22:04DefilerOr, rather.. same problem
20:22:14DefilerThe spec in question fails so hard that it hangs the spec run
20:22:28DefilerFeel free to fix it :)
20:23:01jluxenbergah, I see
20:23:26DefilerMake sure ./bin/mspec ci passes before you hit 'commit', though
20:23:30DefilerThat is the important part, for now
20:24:00jluxenbergyeah, I just made a new spec so I want to make sure it fails in the right way
20:24:17peglegrot leaves the room.
20:24:50DefilerOh, and you can do: ./bin/mspec ci -t r
20:24:54Defilerto run the suite under ruby 1.8
20:24:59jluxenbergok
20:25:00Defiler(Which should also pass 100%)
20:25:09DefilerThat is worth running to make sure you wrote the spec correctly
20:25:40jptix_ leaves the room.
20:31:51jluxenbergso I made a change to core/marshal/load_spec.rb
20:32:06jluxenbergwhich will cause mspec ci to fail
20:33:05be9 leaves the room.
20:33:06DefilerYou can do: bin/mspec tag --add fails spec/core/marshal/load_spec.rb
20:33:13jluxenbergah, ok cool
20:33:16Defiler..which will mark any specs that fail on that run as 'failing'
20:33:30DefilerThen re-run bin/mspec ci to make sure it is correctly excluded
20:33:38jluxenbergok
20:34:31Defilerthis page may be of interest generally: http://rubinius.lighthouseapp.com/projects/5089/specs-runners
20:34:45DefilerThey have gotten a little option-heavy over time
20:35:37riles enters the room.
20:35:53therealadam enters the room.
20:35:59jluxenberglooks like there is one ERROR spec that I didn't modify: spec/ruby/1.8/core/process/setrlimit_spec.rb
20:36:16riles enters the room.
20:39:16DefilerWhat platform are you on?
20:39:49jluxenbergUbuntu linux
20:39:53Defiler..and can you paste the failure output at pastie.org or some other paste site?
20:39:57jluxenbergyea
20:40:12DefilerThat spec has been irritating on various Linux systems, and I would like to take this opportunity to nail it
20:40:16wifelette enters the room.
20:40:19jluxenberghttp://pastie.caboo.se/159858
20:42:02jluxenbergso if one of these specs says it "FAILED", does that mean its in the fail list?
20:42:12jluxenbergbecause one of the new specs I just wrote says it "FAILED"
20:42:19jluxenbergbut I definitely haven't added it to the fails list
20:45:12ataiar enters the room.
20:45:51DefilerNo, that means it is on the 'not failing' list, but failed anyway
20:46:00DefilerSo, unexpected failure
20:46:22jluxenbergso ERROR and FAILURE are just two different kinds of unexpected failure?
20:47:02jluxenbergin that case theres another test case that failed unexpectedly: http://pastie.caboo.se/159860
20:47:59Defilerfailure means an expectation was not met
20:48:06Defilererror means an exception was raised
20:48:07jluxenbergah, I see
20:48:24DefilerOK, yeah.. that one you just pasted is known, and has a ticket open
20:49:38cremes enters the room.
20:49:39radarek enters the room.
20:50:03Defilercan you type 'man 2 setrlimit' and paste the man page that it shows you?
20:51:40DefilerAlso, does that same spec fail for you when you run it with: bin/mspec ci -t r spec/ruby/1.8/core/process
20:56:51jluxenbergone sec
20:57:07ataiar leaves the room.
20:57:38jluxenbergDefiler: man page: http://pastie.caboo.se/159861
20:58:08DefilerThanks. I will take a peek in just a moment. This spec is using a weird feature, so I am googling it up
20:59:07jluxenbergDefiler: it still fails when run with mspec ci -t r ...
20:59:14DefilerOK, that is good
20:59:27DefilerMeans it is an incorrect spec (on your system, at least), not a rubinius bug
21:01:42jluxenbergwhat does ci -t r do?
21:04:47Defilerruns it under ruby 1.8 instead of rubinius
21:04:59DefilerThere are -t options for the various ruby implementations.. -t j for jruby, etc
21:05:35jluxenbergah, I see
21:08:28skaar leaves the room.
21:10:05Defilerjluxenberg: can you paste the output of 'uname -a' ?
21:10:13DefilerShould be one line, so you can just put it in the channel
21:10:30jluxenbergLinux jared-desktop 2.6.22.8-custom92507 #1 SMP Tue Sep 25 23:43:57 EDT 2007 i686 GNU/Linux
21:13:41cavalle enters the room.
21:14:43Defilerjluxenberg: Could you open up that failing spec file (process/setrlimit_spec.rb)
21:15:03Defiler..and above line 73 (the call to Process.setrlimit), put this:
21:15:09Defiler puts "RLIMIT_MEMLOCK: lim: #{lim} max: #{max}"
21:15:23Defiler..run it again, and paste what that prints?
21:15:35jluxenbergok, one second
21:15:51obvio enters the room.
21:16:58jluxenbergRLIMIT_MEMLOCK: lim: 140741783322623 max: 0
21:17:30DefilerOK, yeah.. that's the problem
21:17:35Defilerlim isn't allowed to be greater than max
21:18:14Defilermy box prints: RLIMIT_MEMLOCK: lim: 9223372036854775807 max: 9223372036854775807
21:20:27boyscout1 commit by Wilson Bilkovich
21:20:28boyscout * Tweak Process.setrlimit spec for odd Linux platforms; 6039a3b
21:20:29DefilerThat should fix it
21:21:31DefilerYou can do: git fetch && git rebase origin
21:21:36Defilerto integrate that change to your copy
21:21:51Defiler(all your local changes must have been committed first)
21:22:00DefilerSo you can remove that 'puts' line I had you add
21:23:32jluxenbergok
21:23:46jluxenbergwhat is "origin"?
21:24:16DefilerThat is the name git assigns to the URL you cloned from
21:24:39DefilerSo what that command is doing is downloading any updates, without merging them into the branch you are working on
21:25:01Defiler..and then you are saying "modify my working copy so that the commits I have done come AFTER whatever is in the origin copy"
21:25:17jluxenbergah, ok
21:25:18DefilerIt's pretty cool.. it's like doing "svn up" without needing to merge by hand
21:25:53DefilerYou can also do git rebase --interactive, which is crazy awesome, and lets you move your commits around
21:26:03Defilerbut you won't need that today
21:31:00DefilerBut please let me know if that change makes the spec pass for you
21:31:25jluxenbergokay, one sec
21:32:25jluxenbergwhen I do the git rebase, it says "... needs update"
21:32:40Defiler'git status' should show that nothing has changes
21:32:48Defiler('git diff' is another way to check)
21:33:03DefilerWhich files does git status list out as being modified?
21:33:14jluxenbergvarious .h files created by the build process
21:33:50Defilerreally?
21:33:57DefilerCan you paste the output of 'git status' for me?
21:34:01DefilerThat sounds like a build problem
21:35:04jluxenberghttp://pastie.caboo.se/159879
21:35:08rubuildius_ppcWilson Bilkovich: 6039a3bd4; 1723 files, 5813 examples, 19808 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/159880
21:35:30jluxenbergwoah something weird is going on in lightning/doc
21:35:38DefilerAah
21:35:44Defilerlightning went away yesterday
21:36:36jluxenbergbut the spec still fails, same error
21:36:48Defilertry this: git checkout shotgun/external_libs
21:36:54DefilerYou haven't gotten the updated copy of the spec yet
21:36:58DefilerBecause the 'rebase' didn't finish
21:37:03jluxenbergoh, ok
21:37:09jluxenbergwhat after the checkout?
21:37:24Defilerrun git status again
21:37:35Defilerand see if 'changed but not updated' is now empty/gone
21:38:16jluxenbergit's gone
21:38:27Defilercool
21:38:33Defilernow type 'git rebase origin'
21:38:50DefilerAssuming that doesn't complain, you can type 'rake' and then re-run that spec
21:39:03jluxenberghttp://pastie.caboo.se/159881
21:39:07jluxenbergerrored out
21:39:45Defilerwtf
21:40:18Defilertry opening that file in your editor, and see if it has a section it wants you to edit
21:40:38jluxenbergyeah, http://pastie.caboo.se/159882
21:40:44jluxenbergag shit
21:40:45jluxenbergsorry
21:40:46jluxenberglol
21:40:57DefilerTold you to remove that line :)
21:41:51DefilerOh, and after this part works..
21:41:52jluxenbergweird
21:42:00DefilerI will have you type: "spec/tags/ruby/1.8/core/marshal/
21:42:03Defilererr..
21:42:04Defilergit add spec/tags/ruby/1.8/core/marshal/
21:42:07jluxenbergrake doesn't do anything even though the files have changed
21:42:15DefilerThat's fine.. probably means only specs changed
21:42:25DefilerJust wanted to make sure nothing in the core had changed since you started this
21:42:51jluxenbergah
21:42:54jluxenbergit still fails
21:43:11Defilerdo 'git fetch' and then 'git rebase origin' again just to make sure it is all up to date
21:43:22DefilerSince you had trouble the first time. I think you are OK now, but we should make sure
21:43:54jluxenbergoh, it is saying "spec/ruby/1.8/core/process/setrlimit_spec.rb: needs update" when I "git rebase origin"
21:44:07jluxenbergis that stopping the rebase from happening?
21:44:09Defileryes
21:44:22Defilerdo: git checkout spec/ruby/1.8/core/process/setrlimit_spec.rb
21:45:50Defilerthen git rebase origin again
21:46:47jluxenbergalright should be fixed now
21:46:54jluxenbergbut the spec still fails
21:47:02jluxenberggit rebase origin says "Current branch master up to date"
21:47:35jluxenbergand "git diff spec/ruby/1.8/core/process/setrlimit_spec.rb" outputs ""
21:47:42jluxenbergso I'm pretty sure we have the same file there
21:49:46DefilerOK
21:49:52DefilerYou can make totally sure by looking on line 73
21:50:05Defiler..and if you see max = lim if lim > max then it is right
21:50:26DefilerOr, well, at least includes the change I added
21:50:27jluxenberg max = lim if lim > max # EINVAL is raised if this invariant is violated
21:50:34DefilerOK. Cool. You are up to date
21:50:36MacRue enters the room.
21:50:42obvio171 enters the room.
21:50:46DefilerSo, that didn't fix it.. i will install Ubuntu and track it down here
21:50:57DefilerSo we can ignore that for now. Sorry for the wild chase
21:51:01Toba leaves the room.
21:51:17DefilerNow, you can do that 'git add' on spec/tags/ruby/1.8/core/marshal
21:51:20MacRueEw, Colloquy plays sounds
21:51:39Defiler..and commit that as well, and use git format-patch again to make a patch for that as well
21:52:10jluxenbergI already made a patch by doing "git-format-patch HEAD^ spec/tags/ruby/1.8/core/marshal"
21:52:13jluxenbergis that wrong?
21:52:33ataiar enters the room.
21:53:11foysavas leaves the room.
21:56:16obvio leaves the room.
21:56:21jluxenberghere's the ticket: http://rubinius.lighthouseapp.com/projects/5089/tickets/377-patch-added-specs-for-marshal-ing-a-su bclassed-hash-object
21:56:28foysavas enters the room.
21:59:51Defilerjluxenberg: Your patch isn't wrong.. but I noticed when you pasted 'git status'..
22:00:05Defiler..that the change you had made using that bin/mspec tag command had not gotten committed
22:00:19DefilerBut I can fix that here when I apply it
22:00:22DefilerSo don't bother for now
22:00:38jluxenbergah, okay. where is that data stored, for future reference?
22:00:41ataiar leaves the room.
22:00:58Defilerspec/tags
22:01:04DefilerYou can see it in 'git status'
22:01:05jluxenbergah, makes sense
22:01:09jluxenbergah, okay
22:01:31GMFlash leaves the room.
22:01:45DefilerOh, I see what happened.. you committed that 'puts 'RLIMIT'' line in this patch
22:06:18jluxenbergoh, crap yeah I didn't catch that
22:06:19jluxenbergsorry
22:06:49DefilerTrying to improve my git-fu so that I can easily fix that
22:06:56therealadam leaves the room.
22:13:12Defilerjluxenberg: So.. this spec fails on ruby 1.8 for me
22:13:14DefilerIt passed for you?
22:13:28jluxenbergyeah
22:13:32jluxenberglet me double check.
22:13:40benny enters the room.
22:13:42Defilerhttp://rafb.net/p/RZHNhD90.html
22:14:31bennyhey there... rubuildius will be down until tuesday (the server needs to be booted with a password, I live in germany after all ;-))
22:14:39bennyso, sorry about that
22:15:04Defilerjluxenberg: OK, I changed it from checking that the thing was first in the ancestors list, to checking that it was included in the list
22:15:24Defilerbenny: No problem. That will give people an incentive to finish the new bot setup :)
22:15:26jluxenberglet me figure out why it's not first
22:15:51Defiler[Meths, UserHashInitParams, Hash, Enumerable, Object, PP::ObjectMixin, Kernel]
22:15:58DefilerIs the list I get on 1.8.6
22:16:08jluxenbergokay I see
22:16:11jluxenbergyeah that's right
22:16:15dewd leaves the room.
22:16:33jluxenbergmaybe change it so that it checks that its second in that list?
22:16:40DefilerOK
22:16:49cremesbenny: rubuildius_ppc will carry the load in the interim!
22:16:52bennyMwuwhahaha, my plan worked perfectly.
22:17:09cremesbenny: you crafty devil...
22:17:32bennywell, being disconnected by irc gives rise to the inner evil
22:17:32d2dchat leaves the room.
22:18:45DefilerWow, we have a truly unacceptable number of CI failures under ruby 1.8.6
22:18:59DefilerWe need to fix that before 1.0preview
22:19:18boyscout2 commits by Wilson Bilkovich
22:19:19boyscout * Tweak new Marshal spec to pass on MatzRuby; 70aa320
22:19:20boyscout * Tag new failing Marshal spec; 35476e1
22:19:36DefilerAah.. looks like a boyscout bug, that it didn't report yours
22:20:20jluxenbergs'ok
22:21:36DefilerI updated the ticket. Thanks for the spec
22:21:46jluxenbergwelcome
22:22:01ataiar enters the room.
22:22:12ataiar leaves the room.
22:25:05DefilerI made a ticket for that setrlimit thing
22:25:07Defilerhttp://rubinius.lighthouseapp.com/projects/5089-rubinius/tickets/378-incorrect-process-setrlimit-s pec-on-some-linux
22:25:12DefilerI'll take care of it soon
22:26:08headius1.0 preview?
22:29:54rubuildius_ppcWilson Bilkovich: 70aa320f7; 1723 files, 5814 examples, 19809 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/159903
22:35:46cavalle leaves the room.
22:36:20cavalle enters the room.
22:36:47enebo leaves the room.
22:36:50Defilerheadius: our next milestone
22:37:12headiusis that coming soon? I thought 1.0preview was the version you guys were going to have out in early jan
22:37:19DefilerMay 16th
22:37:27headiusfor railsconf eh?
22:37:30Defileryar
22:37:37headiusrunning rails a kajillion times faster than anything?
22:37:49DefilerAs long as we beat that crapy JRuby pile
22:37:51DefilerI mean.. uhh...
22:37:56Defilerbut yes, running Rails
22:38:02Defilerat least, I have tickets that require such
22:38:12headiusI'm just happy jruby's the one to beat
22:38:31headiushave you guys started running rails tests?
22:38:34Defileryes
22:38:48Def