Index

Show enters and exits. Hide enters and exits.

00:03:04somebodykhaase, I can not reproduce this on head
00:19:39evankhaase: still around?
00:19:50khaaseevan: yes
00:19:59khaasei cannot update to latest head
00:20:01evani figured this was easier that exchanging issue comments :)
00:20:05khaasemelbourne is not compiling
00:20:05evanoh? why not?
00:20:17evanplease gist the output
00:20:49khaasehttp://gist.github.com/317955
00:21:19evanuuumm
00:21:31evanthats bizarre.
00:21:41evanare you doing this directly or under rvm?
00:22:29khaasedirectly
00:22:38khaasesince rvm did fail, too
00:22:48evanweird.
00:22:49evanum
00:22:53evancould you try a fresh clone?
00:23:09khaaseyes
00:23:25evank
00:23:41evanalso if you're doing it directly
00:23:46evandisable rvm in that shell
00:23:53evanrvm confuses things because it sets a bunch of env vars
00:24:06khaaseok
00:25:23wayneeseguinsomething need adjusting?
00:26:19evanwayneeseguin: no, it's rvm setting GEM_HOME etc
00:26:26evanyou said there was an rvm command to unset all those
00:26:29evanwhat was that command?
00:27:46wayneeseguinrvm system
00:27:56wayneeseguinshould put you back in the 'pre-rvm' state
00:28:04wayneeseguinor 'rvm reset' to completely nuke it all
00:28:34evank
00:28:48khaasewayneeseguin: rvm system does that only for the current shell, right?
00:30:50wayneeseguinkhaase: correct
00:31:01wayneeseguin'rvm reset' would blow away for all future shells
00:33:36khaasecompiling rubinius takes quite some time these days
00:33:47wayneeseguintrue but still not as long as ruby-head
00:33:53wayneeseguinunless you have to compile llvm
00:33:55wayneeseguinthat's a beast
00:34:26evankhaase: shouldn't take more than 5 minutes
00:34:43evanhopefully you're not compiling LLVM
00:34:54evanif so, what platform are you on?
00:35:33khaaseevan: I'm not. Also, melbourne compiled just fine, as it seems.
00:35:46khaaseosx 10.6
00:35:55khaasestarted running tests
00:36:04evanok
00:36:09evanshouldn't have taken too long to compile
00:36:31evanunless 5 minutes it quite a long time.
00:36:59evans/it/is/
00:37:15khaasedunno, felt like a bit more than 5 minutes, but haven't watched the clock.
00:38:04evan*shrug*
00:38:24khaasehmmm... works on latest head.
00:38:49khaasequestion is... is that due to some update or since i did not use my rvm stack?
00:38:54evani actually remember this bug
00:38:59evanbut i fixed it a while back
00:39:01evanlet me see when.
00:40:03wayneeseguinkhaase: are you installing 'rbx' or 'rbx-head' ?
00:40:15khaasewayneeseguin: rbx-head
00:40:20evankhaase: git show bb511f8c
00:40:30evanthats me fixing that bug
00:42:11khaasenice. hmmm... i did rvm remove rbx-head in the meantime, so I cannot figure out when I last updated it.
00:42:59evanok
00:44:42khaasegonna play around with it a bit (i.e. try to install head with rvm, then run my real code against it)
00:44:54evank
00:44:56khaasebtw, in rvm it's mkmf.rb can't find header files for ruby at /Users/konstantin/.rvm/rubies/rbx-head/include/ruby.h
00:45:20evanwe don't have include/ruby.h
00:45:21evanthats why.
00:45:31evanwayneeseguin will have to look at that
00:45:40evankhaase: i'm going to close this issue, ok?
00:46:03khaaseevan: ok
00:46:18khaaseevan: thanks
00:46:22wayneeseguinexcept that *rvm* doesn't do anything with a 'ruby.h' and/or 'mkmf.rb'
00:46:36evannp
00:46:51evankhaase: are you using --prefix and install?
00:46:59wayneeseguinrvm will use --prefix
00:47:07khaaseevan: rvm does
00:47:55evanok
00:48:04wayneeseguintries a new 'rvm install rbx-head '
00:48:09evankhaase: so, rbx-head on rvm can't install any extensions?
00:49:12khaaseseems likely. it fails during rbx-build extconf.rb.
00:49:22evanhrm.
00:49:30evanoooh
00:49:32evancould you gist the output?
00:49:36evani just added rbx-build actually
00:49:40evanmaybe I screwed something up.
00:49:44wayneeseguinkhaase: if you do with '--trace' it'll show all commands
00:52:23khaaseevan: http://gist.github.com/317974
00:52:55evanmmmmmmmmmm
00:52:57evani did this.
00:52:59evanwayneeseguin: you're off the hook.
00:53:04evangive me a sec
00:53:05wayneeseguinw00t!
00:53:15wayneeseguinwipes the sweat off his brow
00:54:08wayneeseguinhaha
00:54:09wayneeseguinmkmf.rb can't find header files for ruby at /Users/wayne/.rvm/rubies/rbx-head/include/ruby.h
00:54:12wayneeseguinI got the same error
00:54:31evanyeah
00:54:53evanmkmf.rb is a pain in my ass.
00:55:12kronos_vano))
00:55:22khaasealso, i have that extremely verbose output of --trace now, if anybody is looking for some evening reading.
00:55:36evanok
00:55:39evankhaase: pretty sure i've got it.
00:55:51evangive me a sec, i'll push a change.
01:04:20khaaseevan: no hurry. think i gonna leave soon anyway, farbuary has been gone for some time now around here and it's like monday, which sucks. will have a look at it tomorrow.
01:04:34evanok
01:04:37evani've got it fixed
01:04:40evanabout to push it
01:04:47boyscoutAdd RBX_CAPI_DIR to assign in building extensions - 07d613e - Evan Phoenix
01:04:55evanupdate and give that a shot
01:07:54khaaseevan: yay, worked.
01:08:14evancool
01:08:53khaasebah... no method 'nth_ref' on an instance of Rubinius::VariableScope. (NoMethodError) when running some real code
01:09:18evanyou've got stale .rbc files
01:09:30evanvery stale
01:09:52evanbecause newer .rbc files are automatically detected as bad
01:10:01boyscoutCI: rubinius: 07d613e successful: 3037 files, 11971 examples, 36275 expectations, 0 failures, 0 errors
01:10:03evando a find dir -name "*.rbc" -delete
01:10:06evanand you'll be fine
01:11:03khaasei usually have those added to my rake clobber task. however, that sometimes spares the rakefile.compiled.rbc for some reason.
01:11:13evanah
01:11:14evanyeah
01:11:23evanwe've solved the problem now
01:11:33evanvia some signature/versioning magic
01:11:50evancompiler changes updated the signature, which invalidate old .rbc files
01:11:57khaasenice
01:12:16khaasehmm... still persists. i think it's caused by some old gem's rbc-files
01:12:44evanyeah, you might have to hunt them down a bit
01:12:46evanit's a pain, i know.
01:12:52evanbut thankfully this should be your last time :)
01:13:44khaaseyep. and my specs pass. meaning that bug is definitely gone. thanks again, evan.
01:14:20evanno prob!
01:14:28evanthanks for running your specs and letting us know
03:45:36heftigso I'm trying to compile rubinius, but I'm running into errors because of unused or uninitialized variables
03:46:19heftigi see that assertions are used everywhere, but those lines are apparently disabled because the code is compiled with -DNDEBUG
03:46:34heftighowever, i have no idea where this flag is set
04:25:08heftiganyone?
05:32:24brixenheftig: if you are building with a system llvm, that may be setting NDEBUG
05:32:50heftigah right, it does
05:32:57brixenheftig: try ./configue --skip-system if you are using a system llvm
07:30:18boyscoutAdded stdin and snippet support to compiler CLI. - a800d9a - Brian Ford
07:30:18boyscoutAdded SexpPrinter to render an AST. - c4c389e - Brian Ford
07:30:18boyscoutAdded String#to_sexp and File.to_sexp. - 9ff2fee - Brian Ford
07:30:18boyscoutStandardize some AST class names. - 2bbdc8d - Brian Ford
07:30:18boyscoutFixed misspelling of snippet. - 9ff8193 - Brian Ford
07:37:07boyscoutCI: rubinius: 9ff8193 successful: 3037 files, 11971 examples, 36275 expectations, 0 failures, 0 errors
08:46:28dbussinkevan: ping?
16:56:05evanmorning
17:03:27rueI reject your timezone and substitute my own
17:11:32evansolar position substitution approved
17:12:05rueExcellent, now we can blow something up
17:12:49evancan we use the giant magnifying glass to blow up some asteroids?
17:19:13brixenmorning
17:21:17evanbrixen: good job on #to_sexp !
17:21:41brixenheh, thanks
17:21:47brixenruby sure has a lot of syntax
17:22:24evanyeah
17:24:22BrianRice-work:)
17:25:42brixengoing from the rather nice AST back to the sometimes wacky MRI parsetree was much much easier than vice-versa
17:25:57evani can tell, it only took you a couple days
17:26:02brixenheh
17:26:21evani'm pushing through getting in_rescue removed
17:26:25brixenI forget how many hours I spent on the 120 lines of the MultipleAssignment node
17:26:25evanshould be done this morning
17:26:32brixenahh cool
17:27:33brixenI'm going to work on defined?
17:27:38brixen#122 and other holes
17:27:50evanrad.
17:27:55evanthats perfect.
17:30:38brixenevan: do you want lib/bin/describe.rb anymore? you can now do: echo 'foo' | bin/rbx compile -[whatever] - to read stdin
17:30:53brixenor bin/rbx compile - and it will block reading stdin
17:30:59evanyou can nuke it
17:31:03brixenk
17:31:23brixenstill would love to make an irb plugin to enable interactive compiler
17:31:35evanyeah
17:31:39evanwe shoud make one
17:31:39brixenwhat happened to wycats refactor of irb?
17:31:46evanstill on github I think
17:31:51brixenah ok
17:50:41brixenwhat the hell does "returns 'method' when defined?(Kernel.puts) is sent (attribute)" even mean?
17:50:50brixenstarts rewriting defined? specs
17:50:54brixenle ugh
17:51:38brixenthankfully I have Goldfrapp to keep me company
17:59:01dbussinkevan: got a question on the gem precompilation, is it by design that it swallows any exceptions that have occured?
17:59:17evannope
18:00:47dbussinkevan: because i changed it to throw an exception for #142, but it silently continued
18:01:12evanlets see....
18:01:31evanah right
18:01:32evanwell
18:01:34evanhm
18:01:42evansince this is ahead of time
18:01:47evanit should probably swallow it yes.
18:01:50evanand let it fail later
18:01:54evanwhen they try to run the code
18:01:58evanthe precompiling is a convience
18:02:20dbussinkhow is compilation for gems handled security wise? ownership of rbc's etc?
18:03:03evanby the user doing the install
18:03:06evanwhich is the point
18:03:14evansince the gem installation might be done by root
18:03:23dbussinkexactly, but if precompilation failed that can cause issues then
18:03:23evanand thusly go to places a user can't write
18:03:29evannah
18:03:30evanit's not an issue
18:03:37evanthat file will always just load cold
18:03:41evanand fail then
18:03:42evanno biggy.
18:03:50dbussinki'm just wondering whether fail early and fail fast might be the better approach initially
18:04:26brixenyeah, probably should
18:04:44evani don't think so.
18:04:45brixendoes the precompilation tie into gem error handling?
18:04:57evanfailing during gem installation is a lot more complicated
18:04:58evana lot.
18:05:08evanfailing at file require time is easy.
18:05:11evanand clear to the user
18:05:55brixentrue, it's not actually related to installation
18:06:19brixenand the gem wouldn't need to be re-installed when the rbx error is fixed
18:06:47brixenie, the error is in rbx, not in the gem or in the installation
18:09:36dbussinkdoes this make sense for people? https://gist.github.com/9a84aa7eaa6494199464
18:10:43evanright, since it's a rbx error, we should swallow it and allow it to pop up later if need be.
18:11:08evandbussink: yeah, thats fine
18:11:19evanwe'll still need to audit the call path for var_table_add though
18:11:24evanthere are other things we're leaking
18:14:48brixenmaybe we should add a little pool memory structure that can be called to free everything when we bail out
18:15:38evanyeah
18:15:40evanprobably
18:16:08evanwe use longjmp in rb_raise, so we can't use C++ destructors
18:16:37brixenyeah, some explicit cleanup before we rb_raise
18:16:51boyscoutMore friendly error when a large number of variables is used - cf560ad - Dirkjan Bussink
18:22:21evandbussink: if you want to REALLY fix that
18:22:34evanvar_table should use an STL vector
18:22:38evanso it has no fixed size.
18:22:46dbussinkthat's an even nicer fix yeah
18:23:00dbussinkbut isn't melbourne c only?
18:23:18evanno
18:23:21evani don't think so...
18:23:30boyscoutCI: rubinius: cf560ad successful: 3037 files, 11971 examples, 36275 expectations, 0 failures, 0 errors
18:23:35evanit's C++
18:23:41evani see lots of .cpp files :D
18:23:41dbussinkwell, someone has rewritten it to be used in mri too
18:23:50evanbrian did
18:23:51evanthats fine
18:23:53dbussinki see extern C ;)
18:23:57evanyou can use C++ in MRI extensions
18:26:36brixenfun [ruby-core:28383]
18:28:37dbussinkbrixen: i see a quote that will thrill you again "According to the draft of ruby-std, it is implementation defined when
18:28:37dbussinkthe class variable scope is ambiguous."
18:30:35brixenexactly
18:30:53brixenbut now we have a mechanism to deal with this in rubyspec
18:31:52brixenI think the "implementation defined" stuff in the ISO standard is mostly to get 1.8 and 1.9 both under the same document
18:31:58brixenwhich sucks, but oh well
18:32:06brixenno one is going to run the ISO standard :)
18:32:35evanimplement the ISO spec and you'll pass nothing!
18:33:11brixenyou'll pass some hot air about the iso standard blah blah blah
18:33:12brixenheh
18:33:38brixenironruby has been so quiet on the rubyspecs lately, maybe they switched to the iso standard
18:33:41brixenhaha
18:34:04evanhah
20:06:48dbussinkis there any reason xfree() is used in var_table.cpp in melbourne?
20:07:13evani don't recall
20:07:15evanprobably not
20:07:43dbussinksince that breaks if i remove the extern C stuff
20:09:05evanyou can pretty safely change it
20:11:07antonioalegriahi peoplinius
20:11:22antonioalegriaanyone here recommends RVM for multiruby development?
20:11:31evanantonioalegria: wayneeseguin probably does.
20:11:34wayneeseguinlol
20:12:26antonioalegrialead developer, I see
20:12:28antonioalegria:-D
20:12:43antonioalegriabut c'mon, other people, does it "Just Work (tm)"?
20:13:03evana lot of people using rubinius use rvm to set it up
20:13:06evanso i'd say yes.
20:13:38antonioalegriaI'm trying to develop a new app with 1.9 but I need to come back to 1.8 to see if some crashes are because of incompatibilities and I would like to try Rubinius while I go along
20:13:39antonioalegriaso
20:13:54antonioalegriaI'm having problems with default Mac OS X ruby and macports ruby1.9
20:13:59antonioalegriabecause of gem binaries
20:15:45antonioalegriaok, 'll give it a try
20:18:23antonioalegriawayneeseguin: do you recommend installing via git instead of gem?
20:18:36wayneeseguinyes
20:18:53wayneeseguinantonioalegria: http://rvm.beginrescueend.com/rvm/install/
20:19:01wayneeseguinMethod 1
20:19:31antonioalegriaappreciate it
20:19:40wayneeseguinantonioalegria: Depending on the state of your system you might also end up needing these: http://rvm.beginrescueend.com/packages/
20:19:54wayneeseguinUsually people who have upgraded to SL instead of clean install need those things
20:20:13antonioalegriaright
20:20:23antonioalegriaI'm a clean install kind of guy ^_^
20:20:34wayneeseguinGOOD :)
20:20:42antonioalegriabut if I'll be sure to remember it if any errors occur
20:55:51dbussinkevan: https://gist.github.com/6bcf71ee83990fa55617
21:13:36dbussinkevan: that's with a vector :)
21:27:07evandbussink: perfect!
21:27:15evan12445 locals? no problem!
21:27:52dbussinki removed the deletion from var_table since it seemed that it wasn't used anywhere
21:28:56evancool
21:31:16brixenfalls out of chair, kinda
21:31:18brixenwow [ruby-core:28385]
21:31:34brixenthat's at least the second post where an MRI patch also has rubyspec patch
21:31:38brixenmame rocks
21:32:38brixenmame is Yusuke Endoh's irc nick for those following along at home
21:33:38boyscoutRework melbourne's var_table to use a std::vector - f6644e3 - Dirkjan Bussink
21:33:47dbussinki love commits like " 2 files changed, 16 insertions(+), 57 deletions(-)"
21:34:34evanhell yeah
21:40:15boyscoutCI: rubinius: f6644e3 successful: 3037 files, 11971 examples, 36275 expectations, 0 failures, 0 errors
21:41:09brixenDirkjan the Deleter
21:41:10brixen:)
21:45:51dbussinkbrixen: i can be much more rigourous ;)
21:45:59dbussinkbut i'm going to get some sleep :)
21:47:03brixendbussink: heh, of course, rest up to battle another day :)
21:47:17dbussinknite!
21:47:22brixennite!
21:47:29evandbussink: nite!
23:07:23evanhuzzah, got the specs fixed.
23:12:07brixensweet
23:29:49boyscoutHandle very long lines in the backtrace better - 3daf3fd - Evan Phoenix
23:29:49boyscoutLame rails ivar conflict fix - 9fd5185 - Evan Phoenix
23:29:49boyscoutLame rails ivar conflict fix - bfdfcb6 - Evan Phoenix
23:29:49boyscoutFix Regexp#names to return an array of Strings, not Symbols - aaa2cad - Evan Phoenix
23:29:49boyscoutChange asserts() in interpreter to raise InternalError - 3823dd2 - Evan Phoenix
23:29:50boyscoutFix exception stomping bug in exception condition check - 20ff335 - Evan Phoenix
23:29:50boyscoutLame rails ivar conflict fix - 1f21ac3 - Evan Phoenix
23:29:50boyscoutAdd spec Autoload for running in an exception condition - feb4fed - Evan Phoenix
23:30:24evanbadabing
23:30:30brixenbada boom
23:38:14boyscoutCI: rubinius: feb4fed successful: 3037 files, 11972 examples, 36276 expectations, 0 failures, 0 errors