Index

Show enters and exits. Hide enters and exits.

00:17:40wayneeseguinf
00:17:55kronos_vanoff
00:25:40lukemeliacongrats, guys. way to persevere to 1.0!
00:47:44evanlukemelia: thanks!
00:47:54evankronos_vano: did you put the magic coment in?
00:49:47kronos_vanoyep. q = +[ x , x.in([1,2,3]), y.in([5,6,7])] woks, q = +[ x + y, x.in([1,2,3]), y.in([5,6,7])] no
00:49:52kronos_vano*works
00:50:39kronos_vanoevan, why you never told about this future? This is awesome!
00:50:48evanit was just something i was playing with
00:50:54evanit's not on by default because it's just a toy
00:51:00evanfeel free to play with it though
01:17:43rb2kanybody got 1.0 to install with RVM?
01:18:03rb2ksince a few versions, I've got problems with the "env ruby" part of the script
01:44:37hammerdrGrats on getting 1.0 out. I have a rather odd question, I suppose, but is the rubinius VM targetable? E.g. could I write a byte code interpreter for another language that runs on rubinius?
01:45:58ezmobiusyeah totally
01:46:27ezmobiusi have a half started scheme compiler that compiles straight to rubinius bytecode and lets me call out to ruby
01:46:36ezmobiusbut i havent had time to work on it in almost a year
01:47:17ezmobiusthe vm is definitely pretty nice. it is specialized for ruby though. but ruby is a dynamic lang so I guess it depends on what kind of lang you want to write
01:48:18hammerdrgood stuff :) I was researching this a couple weeks ago when I was trying to find a ruby vm to target.. I couldn't find any docs that related to this though. Is there any laying around? An instruction set, at least? :)
02:11:20postmoderni just tried running rake on HEAD
02:11:27postmodernand got invalid byte sequence in US-ASCII
02:11:31postmodernunder ruby 1.9.1
02:12:05postmodernhttp://pastie.org/961221
06:12:54postmoderntrying to run specs of my libraries on rubinius 1.0
06:13:02postmodernbut im having trouble loading rbc files for YARD
06:13:45postmodernhttp://pastie.org/961359
06:19:40postmodernalso having trouble bundling nokogiri from rubygems.org
06:19:44postmodernhttp://pastie.org/961364
06:34:09brixenthe noko one says libiconv is missing, is it?
06:34:23brixenthe yard one, what command did you run?
06:47:54brixenpostmodern: hm, why is that trying to load a .rbc file directly?
06:49:04postmodernbrixen, libiconv is installed, but says version 0 of it
06:49:54brixenI don't know what that means
06:49:57postmodernah ha
06:50:03postmodernit's only the virtual package
06:50:15brixenok
06:51:47postmodernbrixen, whats odd is I was able to install nokogiri using rbx /usr/bin/gem install just fine
06:51:53postmodernbrixen, but it fails under bundle install
06:51:59brixenyeah, that is odd
06:52:09brixendoes it run from your gem install?
06:52:20brixenI honestly don't know much about bundler (yet)
06:53:11brixenbtw, rbx gem or rbx -S gem is all you need to invoke
06:53:49postmodernah -S, thats what i need
06:54:24postmodernah it looks like it was using the wrong cached gem
06:54:25brixenrbx gem is special, but rbx -S <name> is used for all binaries installed by gems
06:54:32brixenie, rbx -S rake, etc
06:55:21postmodernbrixen, Installing nokogiri (1.4.1.20100514000014) from .gem files at /home/hal/.bundle/rbx/1.8/cache with native extensions An exception occurred running /home/hal/rubinius/gems/bin/bundle
06:56:07brixenok
06:56:15brixenI don't know what to do with that
06:56:25brixendo you have steps I can use to set up the same env?
06:57:50postmodernbrixen, sure let me write up a bug
06:57:58brixenok cool
06:58:03brixenI'm going to sleep
06:58:03postmodernbrixen, nice thing about bundler, is it only needs a directory with a Gemfile
06:58:08brixenhave fun!
06:58:21postmodernoh wait i know why
06:58:37postmodernit's somehow still using the extra gem source from http://tenderlove.com
07:01:25postmodernbrixen, ok one bug down
07:01:46postmodernbrixen, just reported some Ruby behavior differences I found in my specs
07:01:56postmodernshould be trivial to fix
07:16:11postmodernsay how good is FFI on Rubinius?
07:16:23postmodernim having trouble just resolving the 'uint16' type under rubinius
07:16:28postmodernhttp://pastie.org/961382
07:16:34postmoderntrying to run specs on ffi-udis86
07:22:28postmodernalso has anyone thought about storing rbc files as .#{name}.rbc
07:37:19postmodernalso why is FFI::Struct.new returning a FFI::Pointer
07:37:26postmoderninstead of the class of the struct
07:37:38postmodern*of my defined struct
07:41:37postmodernah a benefit of a fresh release
07:41:50postmodernwatching the issues start to roll-in
10:06:12candellerare .rbc supposed to be treated like .rb files? that is, once i have one, do i invoke it 'rbx file.rbc' ?
10:12:07matthewdcandeller: I don't think so... but could be wrong
10:13:54candelleri am getting a buch of errors trying to do that - http://pastie.org/private/6jculhhqnq7vxymablnkhq - the path to the file in the .rbc file is treated like a regexp
10:18:20candeller(as long as what I am doing is legal that is)
10:30:27matthewdcandeller: Sounds like it's just trying to run it as a ruby script... which certainly suggests it isn't intended to work
10:35:06postmoderni really feel like the rbc files should be hidden
10:35:14postmodernprefixed with a '.'
10:35:18postmodernlike vim swp files
11:16:32dbussinkcandeller: you should use the rbc files directly normally
11:16:44dbussinkrbx will check if they exist and use them if needed
11:17:29dbussinkpostmodern: btw, on #293, are you sure you checked it with 1.8?
11:17:37dbussinkbecause 1.8 gives me a NameError just like rbx does
11:18:25candellerdbussink: you mean .rb files directly ? so, `rbx file.rb`, looks for an .rbc file, if there's one, it uses it, if not, makes one and uses it?
11:18:48dbussinkcandeller: yeah
11:18:53candellerok
11:19:07dbussinkyou shouldn't be concerned with invoking rbc files yourself
11:19:31badboyhm, I can't compile rubinius =/
11:19:57postmoderndbussink, just tested it on 1.8.7, same results
11:20:07dbussinkpostmodern: do you have a script for me?
11:20:12postmoderndbussink, i found this bug when running specs for the parameters and rprogram library
11:20:27postmoderndbussink, wait im confusing bugs
11:20:35dbussinkpostmodern: https://gist.github.com/23769bce3c5a11e3772c
11:20:35postmoderndbussink, but yes i do have a script for that bug
11:20:40badboyhttp://gist.github.com/402146
11:20:51dbussinkpostmodern: please attach stuff like that directly, because the basic cases seem to be fine
11:21:00dbussinkso we can't debug it properly with a reproduction
11:21:05postmoderndbussink, did you trie blah()
11:21:11postmoderndbussink, or blah { }
11:21:39postmoderndbussink, i think those invoke actual method calls
11:21:44dbussinkpostmodern: ah ok, please provide information like that in the issue :)
11:21:49dbussinkbecause now i have to ask about it :)
11:21:50postmodernok
11:22:17dbussinkbadboy_: looks a known issue: http://github.com/evanphx/rubinius/issues#issue/298
11:22:31dbussinklooks like gcc 4.5 has started to warn about stuff like that
11:22:54badboyok,
11:23:18badboychecked the issues half an hour ago, =D
11:23:18dbussinkpostmodern: are you adding the information to the issue?
11:23:37dbussinkbadboy_: also gcc 4.5?
11:23:49badboyyep
11:23:56badboyarch linux, too
11:24:18postmoderndbussink, doing so now
11:28:38dbussinkbadboy_: you can try after applying this diff: https://gist.github.com/a5d201ce95ce3c5eddc5
11:29:53badboyok
11:30:45dbussinkbadboy_: hmm, the diff is probably not correct though, but usually you don't use instrumenting
11:31:21badboyhm, the diff fixes the error with vm/instruments/stats.hpp
11:31:52dbussinkbadboy_: but somewhere else something pops up?
11:31:53badboyso the errors like "./vm/gen/instruction_implementations.hpp:92:16: error: value computed is not used" are still there
11:32:10dbussinkhmm, could you add that information to the ticket?
11:32:16dbussinkbadboy_: because i don't see it there
11:32:24badboyok
11:35:00badboydone ;)
11:38:45postmoderndbussink, updated: http://github.com/evanphx/rubinius/issues/#issue/293
11:38:47postmoderndbussink, test away
11:39:21dbussinkpostmodern: ok, cool, that at least makes it easier to fix :)
11:48:51dbussinkbadboy_: could you try with this? https://gist.github.com/2cfb09a7b143fce4876d
11:51:22badboyrake runs...
11:54:53dbussinkbadboy_: cool :)
11:58:13badboyhm, why is llvm 2.7 not supported yet?
12:01:45dbussinkbadboy_: because they changed a whole lot of things since 2.6
12:01:50badboyok
12:01:56dbussinkso it's some work to do the upgrade
12:02:48badboyok
12:03:54badboyrake aborted!
12:03:54badboyno such file to load -- ext/digest/sha1/sha1
12:04:06dbussinkbadboy_: you're using 1.9 to build it?
12:04:06badboy% bin/rbx -v
12:04:06badboyRuntime exception: Unable to load /home/badboy/code/rubinius/runtime/platform.conf, it is missing
12:04:16dbussinkcurrently only 1.8 is supported for building it
12:04:16badboyhm, yes. not possible?
12:04:26badboyoh, damn *using rake18*
12:04:29dbussinkno, they changed some tricky things that make it hard :(
12:04:48badboyyeah. and here on arch linux I'm living on the edge, so llvm 2.7 and ruby 1.9 =D
12:11:35badboyok, tests are running
12:13:04badboy3457 files, 13601 examples, 41231 expectations, 1 failure, 0 errors
12:17:20badboyProcess.setpriority sets the scheduling priority for a specified user FAILED
12:41:25dbussinkbadboy_: does that spec consistently fail?
12:43:07badboyweird. mom
12:43:10badboy-weird.
12:43:43badboyseems so
12:44:00dbussinkbadboy_: ah, well, it's probably not very critical, you can also check if it also fails with mri
12:44:09dbussinkbadboy_: but seems like it's running then :)
12:44:21dbussinkbadboy_: ./bin/rbx will give you an irb session
12:44:35dbussinkand you can use stuff like ./bin/rbx -S gem install some_gems to try stuff out
12:47:29badboyi know
12:47:39badboynot the first time I compiled and used it, but thanks ;)
12:48:03jaribwas the fix committed?
12:48:52jaribor is the first post-1.0 commit sacred maybe :)
12:51:18dbussinkbadboy_: ok, np, lot's of new people now with 1.0 :)
12:51:28dbussinkjarib: hehe, not committed (yet)
12:51:33badboyyep. that's what brought me back^^
12:51:43dbussinkjarib: first want to verify that it actually solves it, because i don't have a gcc 4.5 here handy
12:56:04dbussinkjarib: ah, i see you reported it originally :)
13:01:07jaribdbussink: let me know if i should test anything
13:01:16boyscoutDon't use a increment operation and the original value in the same expression - d711862 - Dirkjan Bussink
13:01:16boyscoutExplicitly cast the return value of stack_pop() to void* - 10f4032 - Dirkjan Bussink
13:01:31dbussinkthere we go, no virgin 1.0 in the repo anymore
13:01:39dbussinkjarib: that should fix the build issues
13:01:45jaribthanks, will try
13:08:55jaribit's building, my poor netbook is not enjoying it :)
13:10:10dbussinkjarib: hehe, shouldn't take too long (unless it wants to compile llvm)
13:11:05jaribno, it's fine
13:11:10jaribjust a slow machine
13:11:25boyscoutCI: rubinius: 10f4032 successful: 3457 files, 13596 examples, 41216 expectations, 0 failures, 0 errors
13:15:18duncanmvevan: hi
13:15:25duncanmvevan: what LLVM does 1.0.0 require?
13:24:52dbussinkduncanmv: 2.6
13:25:10kronos_vanoBut you can use prebuild one
13:27:15duncanmvkronos_vano: actually I am trying to build rubinius with a system installed 2.6 one
13:27:24duncanmv(it does not work right now)
13:27:33duncanmvand that would prevent adoption in linux distributions
13:28:10duncanmvI have a patch that makes the build system use the system one, but rbx-build (which just calls the vm, segfaults then)
13:38:07dbussinkduncanmv: could be that the system one is compiled with different options?
13:38:46dbussinkduncanmv: and it does work with the prebuilt llvm?
16:10:22duncanmvdbussink: uh no, prebuilt also makes the vm segfault on bootstrap
16:11:23duncanmvProgram received signal SIGSEGV, Segmentation fault.
16:11:24duncanmv0x000000000069ec9d in mp_copy ()
16:14:38brixenduncanmv: how do you know you are using the prebuilt?
16:14:59brixencan you gist your ./configure --show
16:15:19brixendbussink: hmm, I wonder if we should have a stack_discard() op instead
16:15:47dbussinkbrixen: that's maybe a better solution then yeah
16:15:51brixendbussink: because the (void*) did not work before for some libc method that returns a value
16:16:04brixendbussink: just a thought, we can ask evan
16:16:06dbussinkbrixen: maybe that was secretly a macro?
16:16:12brixencould be, yeah
16:16:21brixenstill, might be semantically nicer
16:16:41brixendbussink: thanks for hoping on the fixes though! :)
16:16:55brixenI should be around later to work on a few
16:17:56brixenwhat's the best way to install rmagick on os x?
16:20:58brixenoh der, it's ImageMagick
16:21:07brixenit's been about 5 years since I used this
16:26:34Defilerbrixen: homebrew
16:26:43Defilerand then brew install imagemagick
16:26:49Defileror brew install graphicsmagick if you need that
16:27:07Defilerafter that, the gem should install and build without further intervention
16:30:10brixenyeah, I need to get homebrew going
16:30:15brixenI just port installed it
16:30:28brixenneed to fix c-api issues so rmagick builds with rbx
16:31:38Defilerfun
16:31:43Defilerrmagick is a model citizen
16:33:08brixenindeed
16:33:17brixenwhen the masses want pain, we give them pain
16:33:20brixenI guess
16:33:36brixensomeday we'll be able to say, "no, sorry, you really don't want that and can't have it anyway"
16:33:42brixensomeday :)
16:34:49DefilerI am so pleased with how infrequently I have typed 'sudo' on my new laptop
16:44:30dbussinkbrixen: did feel a bit dirty when i was defiling the pristine 1.0 state of the master branch ;)
16:48:45cyndisis it intended that rubinius's behavior while mutating an array while looping through it doesn't need to match mri's? i vaguely remember reading so
16:49:30cyndismkrf is currently depending on the mri behavior and i'm making a patch for it
17:03:27Defilercyndis: Yeah, I think some of the MRI comments/docs even suggest that it is meant to be undefined
17:04:10Defilerrake abuses that behavior a little as well
17:04:10cyndisok, good :)
17:04:34cyndisyeah, this is actually in mkrf's rake helper
17:04:53cyndis"task 'lib'; directory 'lib; end" which i suppose must be just a silly mistake
17:04:56Defilerhaha yes that's where it is
17:05:09DefilerI remember I saw that while working on Enumerable#sort back in 2007
17:05:13cyndis:)
17:28:27duncanmvhow do I pass additional CFLAGS to the build?
17:28:48duncanmvCFLAGS and CXXFLAGS are ignored
18:33:45dbussinkduncanmv: why do you need that?
19:00:50duncanmvdbussink: distributions have a set of standard C flags
19:01:24duncanmvwhen you build rpm packages, the "lint" program complains if RPM_OPT_FLAGS flags where not used
19:08:57dbussinkduncanmv: there hasn't been a lot of work on package building for various distro's, so it probably needs work
19:10:50rueCFLAGS should not be ignored
19:32:04rolfbis 'gem install nokogiri' not supposed to work on the first run?
19:32:11rolfbit works on the second run
19:32:13rolfb:)
19:32:41rolfb(rbx 1.0)
19:55:18duncanmvI may have the first rubinius package for openSUSE
19:55:39duncanmvlocally osc builds it, lets see if the build service lint is as relaxed as the osc one :-)
19:56:41dbussinkduncanmv: cool :)
19:56:50dbussinkduncanmv: any fixes for the build process are welcome :)
19:58:54brixencyndis: modifying a collection while iterating it is undefined
19:59:04brixencyndis: we will not ever accept a patch for it
19:59:23brixencyndis: I'll get you the ruby-core ML post where Matz agrees it is undefined
20:00:39brixenduncanmv: the build process will be rewritten soon, so please document what issues you had, what you need, and why
20:01:25duncanmvbrixen: I think FAKEROOT and other stuff was the result of my feedback to evan ;-)
20:01:39brixencyndis: http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-core/23633
20:01:48brixenduncanmv: ok, cool
20:01:54duncanmvI was pleased to see that now rubinius uses the std /usr/bin/llvm-config :-)
20:02:10brixenduncanmv: the issue for us is, we don't always know why stuff is needed, so it's harder to architect a better build system
20:02:23brixenany help you can give us understanding is great
20:02:23duncanmvbrixen: I think the remaining issues are OS independent, like
20:02:32duncanmvwe have a repo with gems built against 1.8
20:02:44duncanmvright now we would need to rebuild all gems against rubinius
20:02:50brixenyep
20:02:53duncanmvso they go to the right path etc
20:03:03brixenyou need to do that anyway for any that have C ext
20:03:11brixenwe are very much against sharing gems between impl
20:03:16duncanmvwe need a way to share gems across interpreters, at least for ruby ones
20:03:27brixenI don't think that is a good idea
20:03:29brixenat all
20:03:29duncanmvif you want to isolate them , you have rvm
20:03:41brixenwhy make rvm a dependency on that
20:03:43duncanmvit may not make sense for gem installed gems
20:03:43brixenit should not be
20:03:51duncanmvbut if they are packaged as .deb or .rpm
20:03:58duncanmvthen you will have one gem per interpreter
20:03:58brixenthat's a problem yes
20:04:03duncanmvand that is not user friendly
20:04:13brixenindeed, it's sticky
20:04:29duncanmvand as I said, for pure ruby gems, it is not necessary
20:04:29brixenbut so is people confused why their gems don't work
20:04:38brixenand why this gem has a -rbx and that one doesnt'
20:04:39duncanmvjava works if you have a jar in /usr/share/something
20:05:04brixenmaking it easy on the packaging system should not be a concern
20:05:11brixenmaking it easy on the user should
20:05:26brixenI don't see that some_gem-rbx is that confusing
20:05:46brixenit seems *less* confusing
20:06:21brixenalso, I think .deb or whatever for gems is a terrible idea when gem already exists
20:06:30brixenI hate it on ubuntu and never use it
20:07:10duncanmvI know
20:07:17duncanmvbut it makes sense in some situations
20:07:21duncanmvif you build an appliance
20:07:27duncanmvusing susestudio.com for example
20:07:34brixenwith chef, I don't see that being an issue
20:07:38duncanmvgems packaged as rpms allows to use the system tools
20:07:39brixenbut I'm not a sysadmin
20:07:45brixenso I'm happy to be educated :)
20:07:54duncanmvand the main issue is, it may be a good idea or a bad one, but it is already done
20:07:59brixenchef allows you to do all of that and more, no?
20:08:14brixenbad ideas should not be perpetuated
20:08:14duncanmvSUSE for example ships susestudio.com onsite, and there we have to package everything as rpm in order to do updates with the system tools
20:08:21brixenI hate that argument
20:08:32brixenthe only reason not to fix something broken is laziness
20:08:56duncanmvwhat is the fix? gem rpms work really well, they are visible as rpms and are seen by gem too
20:09:12duncanmvit is the ruby interpreters that are broken
20:09:24duncanmvand authors havent talked each other where to put common compatible gems
20:09:24brixenyou can make this work fine with gem rpms: foobiedoo-rbx.rpm
20:09:26brixendone
20:09:45brixenwhen do you know that there will be an incompatibility?
20:09:53duncanmvwell, there should not be
20:10:05duncanmvat least the environment should not be designed under that assumption
20:10:08brixenbut that's an impossible future-proofing scenario
20:10:14duncanmvI mean , if it only works in foo
20:10:23duncanmvthen install it in /usr/lib/foo-vm/1.0/gems
20:10:26brixen"it works now, we hope it keeps working, if it doesn't, we'll make some exceptions"
20:10:26duncanmvbut if it is not known
20:10:52brixenwell, sounds like a lot of hope and prayers to me
20:11:05brixenand trying to wedge a solution into existing stuff
20:11:16brixenthat was designed when there was *one* Ruby
20:11:53duncanmvI guess then it would be impossible to have 2 rubys without rvm
20:12:14duncanmvand that would make rubinius always a 2nd class citizen in the linux world
20:12:31duncanmvok, rpm for 11.2 builds, for factory doesn't http://pastebin.com/NRUXzMah I guess g++ version
20:12:37brixenrbx can certainly accommodate many schemes, but we don't want to encourage usage that will lead to a lot of headache and confusion
20:13:51brixenugh
20:14:04brixenwas this after dbussink change?
20:14:09brixenduncanmv: are you using master?
20:14:27brixenman what horrid bt's
20:16:50brixenfuck that, I need food
20:16:50duncanmvno, that is an rpm of 1.0.0
20:17:01brixenI can't even read that stupid backtrace
20:17:06brixenbbiab...
20:17:11duncanmvit builds with openSUSE 11.2, but on openSUSE Factory (kind of HEAD distribution) it does not, same .spec file
20:17:25dbussinkbrixen: yeah, that's before my fixes
20:17:27evansorry, i'm headed out for the day
20:17:33evanbut that appears to be because of g++ 4.5
20:17:38evanwe'll have to deal with that next week.
20:17:39dbussinkduncanmv: i fixed those errors after the 1.9 release
20:17:46dbussinkevan: it's already fixed in master :)
20:17:48evanbrixen: GO ENJOY YOUR WEEKEND
20:17:57cyndisbrixen, yes, the patch i made is for mkrf, not rubinius
20:18:03brixenevan: :)
20:18:12brixencyndis: excellent, just need to get the word out there
20:18:16evansorry guys, but as brixen's boss, I must ask that he enjoy the fruits of his labor this weekend
20:18:21evanbugs can wait until monday.
20:18:34dbussinkevan: true, but wanted you to know that gcc 4.5 compile is already fixed
20:18:41evandbussink: yep, no prob.
20:18:53evanand with that, i'm off to buy a shirt and go to a wedding.
20:18:56evansee you all on monday.
20:18:56dbussinkduncanmv: so you could try again with current master
20:18:57brixendbussink: yeah, thanks for doing that
20:18:58dbussinkevan: have fun
20:21:27evandbussink: ok, real fast
20:21:29evanyour last commit
20:21:32evanis wrong-ish
20:21:38evanyou need to be doing
20:21:40evan(void)stack_pop()
20:21:42evannot
20:21:46evan(void*)stack_pop()
20:22:04evan"(void)anything" means "disregard the return value"
20:24:04evanok, i'm off. bye!
20:30:17kronos_vanobrixen, let me fix rmagick stuff
20:38:35slavahi evan
20:38:40slavaevan: I've always wondered
20:38:45slavawhat's the difference between foo() and (void)foo()
20:38:50slavadoes it just silence a compiler warning
20:40:20boyscoutProperly use (void) instead of (void*) - 9cc63a7 - Dirkjan Bussink
20:40:36dbussinkkronos_vano: if brixen isn't allowed to do anything, you'll have all weekend to fix it ;)
20:41:11kronos_vanodbussink, I just do not want to do double work)
20:41:14Maledictusslava: I don't think there is even a compiler warning
20:42:33dbussinkMaledictus: apparently there is one with gcc 4.5
20:42:55Maledictusoh, there will be gazillion lines of code which will trigger that
20:45:04kronos_vanormagick uses rb_frame_last_func
20:45:13kronos_vanohm...
20:45:15dbussinkMaledictus: dunno what cases exactly will trigger the warning, but apparently we had some
20:50:06boyscoutCI: rubinius: 9cc63a7 successful: 3457 files, 13596 examples, 41216 expectations, 0 failures, 0 errors
20:56:56dbussinkkronos_vano: shouldn't be that hard to add from what i see
20:57:57kronos_vanodbussink, http://gist.github.com/402392
20:58:07kronos_vanoreturn call_frame->name(); doesn't work
20:59:36dbussinkkronos_vano: does that complain it returns a symbol not an ID?
20:59:43dbussinkkronos_vano: you should be able to cast that safely
21:00:10kronos_vano->name() returns Symbol*
21:00:16kronos_vanoso it is ID
21:00:18dbussinkkronos_vano: if you look at rb_id2name in vm/capi/util.cpp:8 you can see the reverse being done
21:00:32kronos_vanoah
21:00:36kronos_vanogot it
21:00:39Helena79Discounts!! Our Special Limited Time Offers Up To May,22!!!New BranD!! Notebooks,Plasma and LCD TV's.Buy your electronic needs at our unique prices. Laptop Sony VAIO® VGN-FW590FFD-575,57$!!!Apple MacBook® Air MC234LL/A-695,27$!!! http://www.elplace.com/
21:01:11dbussinkkronos_vano: but other than that, this approach looks good to me
21:04:07dbussinkkronos_vano: be sure to add specs for it too and run them with mri
21:04:11dbussinkto verify it it's correct
21:23:00kronos_vanoSymbol not found: _rb_cFalseClass :/
21:23:34kronos_vanoruby.h#399 it is defined
21:24:08dbussinkkronos_vano: yeah, that's weird
21:24:17dbussinkalthough the symbol shouldn't be there, since it's a macro
21:24:26dbussinkso it should have been expanded
21:27:43ddfreyneAs I understand it, Rubinius uses green threads at the moment, but there are plans for a m:n model allowing usage of multiple kernel threads without a GIL… is this correct? I found some documentation on concurrency, but this file (doc/vm/concurrency.txt) is over two years old
21:28:05kronos_vanoddfreyne, Rubinius uses native threads
21:28:19ddfreynewith or without GIL?
21:28:24kronos_vanowith
21:28:30kronos_vanoBut patches are welcome ;)
21:28:41kronos_vanoevan (C)
21:29:08ddfreyneaww
21:29:54ddfreyneWhy does the GIL actually exist? backwards compatibility with existing code that isn’t threadsafe?
21:29:56brixenddfreyne: we will be removing the GIL over time
21:30:03brixenyes, code is not thread safe
21:30:12brixenso we need to add finer grain locks in those places
21:30:31brixenevan also implemented experimental support for 1.9 Fibers
21:30:47brixenit is possible that we could add M:N based on that
21:30:54brixenbut that's up in the air
21:31:00brixenthe GIL we will remove for certain
21:31:10kronos_vanobrixen, any ideas with rb_cFalseClass?
21:31:21brixenkronos_vano: ideas about what?
21:31:23ddfreyneThe reason why I’m asking is that I have been working on a Ruby program that could absolutely use multithreading to speed up the app quite a bit… but none of the Ruby interpreters allow it
21:31:46brixenddfreyne: you should be using actors anyway ;)
21:31:47ddfreyne(with the exception of jruby, but the JVM startup time is so long that it is actually faster to use MRI instead)
21:32:03kronos_vanobrixen, http://gist.github.com/402409
21:32:04ddfreynebrixen: oh, yeah
21:32:17kronos_vanoarr
21:32:27brixenkronos_vano: probably just need to add it then
21:32:40ddfreynehttp://github.com/evanphx/rubinius/blob/master/doc/vm/concurrency.txt needs to be updated ;)
21:32:47kronos_vanoit is already exists, but I understand the problem)
21:33:17brixenddfreyne: yes, I kno
21:33:26brixenddfreyne: I updated most of doc/ but doc/vm
21:33:34brixenddfreyne: feel free to submit patches ;)
21:33:35ddfreyneahh, cool
21:33:47ddfreynebrixen: I would, but I have little knowledge about rubinius’ internals
21:33:50kronos_vanoHere we go: Symbol not found: _rb_int2big :)
21:34:01kronos_vanoit is easy
21:34:03brixenddfreyne: heh, yes, it's a bit of a catch22 huh :)
21:34:24brixenkronos_vano: thanks for working on rmagick, but don't forget you owe us the other patch too
21:34:32brixenI forget what it was now...
21:34:35brixen:)
21:35:29kronos_vanobrixen, Which one?
21:35:59kronos_vanoah
21:36:05kronos_vanoit is about shifting bits
21:36:12brixenah yeah, that's the one
21:36:48kronos_vanoIt is ready, just needed some tests.
21:36:53brixenok
21:37:09brixenI guess you don't really know if it's ready then, huh? ;)
21:37:17kronos_vano:D
21:37:27brixenok, I'm going to run some errands
21:37:34brixendon't burn anything down :)
21:37:34ddfreynedarn, deleting global_lock.cpp and .hpp didn’t remove the GIL
21:37:40ddfreyne(no, I wasn’t being serious :P)
21:37:58brixenddfreyne: that approach has been offerred many times by dwaite
21:38:05brixenbut we keep rejecting it :)
21:38:23ddfreynehigh fives dwaite
21:39:57ddfreynethe rubinius code certainly is a lot more pleasing to read than the MRI source code
21:40:25kronos_vanoheh 300th ticket
21:56:21duncanmvwonders why g++ warns there
21:56:47dbussinkduncanmv: where?
22:01:26kronos_vanoeverywhere ^^
22:02:25duncanmvahhh grrrrr
22:02:27duncanmvit was fixed in head
22:02:33duncanmvbut it is present in 1.0.0
22:02:46duncanmvthe behavior is actually in the gcc 4 porting guide
22:03:01duncanmv"You cannot modify variables as part of passing them to a function"
22:03:09dbussinkduncanmv: well, last version i ever tested with was with 4.4
22:03:16dbussinkduncanmv: i read it up yeah, it's unspecified
22:03:26dbussinkbut they didn't start warning about it in 4.5 apparently
22:03:31dbussinkuntil 4.5 i mean
22:04:27dbussinkduncanmv: it's actually modifying them in a statement where it also occurs in another place in the same statement
22:05:14duncanmvthinks about applying all head patches to the 1.0.0 rpm package
22:05:28duncanmvwould it work with llvm 2.7?
22:05:32dbussinkduncanmv: probably the best solution
22:05:34dbussinkduncanmv: nope
22:05:42dbussinkllvm 2.7 support isn't trivial
22:05:45dbussinkso that will take some work
22:05:52dbussinkdunno when evan wants to take that on
22:05:54slavawhat hapened in llvm 2.7?
22:06:08dbussinkthey shuffled around stuff in the api from what i know
22:06:22dbussinkit's at least not compatible with 2.6
22:07:47dbussinkduncanmv: the only changes in master are the gcc 4.5 build fixes
22:08:38dbussinkbut i'm going to get some sleep
22:08:42dbussinknite ppl!
22:10:29duncanmvbye
23:28:09boyscoutAdded HAVE_STDARG_PROTOTYPES for C-API. Closes #300. - 3c2edb3 - Brian Ford
23:28:09boyscoutAdded OBJ_FREEZE for bluecloth gem. - fbbdc09 - Brian Ford
23:29:57kronos_vanobrixen, around?
23:30:53brixenkronos_vano: sup?
23:31:12kronos_vanoAny ideas where I can find rb_io_stdio_file function description?
23:31:38kronos_vanoI grep whole 1.8.7 source - it is ghost!
23:31:52kronos_vanono such function.
23:32:26brixenhmm, good question
23:32:30brixenrmagick uses this?
23:32:41brixeneven google code search doesn't find it
23:32:46kronos_vanoarr. I get it
23:32:52kronos_vanoit is 1.9 function
23:33:13kronos_vanohttp://gist.github.com/402509
23:33:14brixensomeone should tell google that :)
23:33:33kronos_vanojust need to implement GetReadFile and GetWriteFile
23:37:13boyscoutCI: rubinius: fbbdc09 successful: 3457 files, 13596 examples, 41216 expectations, 0 failures, 0 errors
23:52:30kronos_vs_rmagicruby-1.8.7-p249 > require 'rmagick'
23:52:30kronos_vs_rmagic => true
23:59:22brixenkronos_vano: sweet!
23:59:50kronos_vanobrixen, rmagick doesn't have tests :(