Index

Show enters and exits. Hide enters and exits.

14:21:14qzis rubinius suitable for embedding in C/C++ program, and does it compile on windows with visual studio?
14:22:04scooprI don't think it compiles on windows quite yet
14:25:40cremesqz: and no one knows if it is suitable for embedding because (AFAICT) no one has done it yet; you could be the first!
14:27:36qzcremes, nice. i'll try if i'll be able to build it on linux ;)
14:27:58cremesexcellent....
14:28:03qzare rubinius developers visiting this channel so i can ask questions if something goes wrong?
14:28:10qz*build on windows ofc
14:29:11cremesyeah, they are on here all the time; wait about 3 more hours and they'll all start popping in...
14:30:16cremesi think a bunch of them live at the GMT - 7:00 timezone
14:30:25cremess/at/in/g
15:26:37dbussinkqz: building on windows won't succeed atm, maybe you'd be succesful with cygwin
15:27:10dbussinkqz: if you get it to build on windows that would be great though of course
15:27:19qzheh, i'll try to..
16:53:31jeranyone here on osx 10.5 having build problems with latest git? keep getting an error on redefinition of struct std::tr1::hash<long long unsigned int> ... it happens in a ifdef check for GNUC <= 4 && GNUC_MINOR <= 3
16:53:37jerusing gcc 4.2.1
16:58:00brixenjer: yeah that check needs to account for 4.2.1
16:58:43jerok, how?
16:59:05jeri mean, do i just need to exclude 4.2.1 from the check, or is there something else that needs to be done?
17:00:30brixenshould be able to change it to GNUC_MINOR <= 2
17:00:45jererr wait, it is < 3 not <= 3
17:00:56brixenhm
17:02:28brixenright, < 2
17:02:32brixentry that
17:02:40jerjust did
17:02:44jermoved past it now
17:02:47jerhopefully that's the end of it
17:02:49brixenk
17:03:05brixendid you use gcc_select to get 4.2.1 on os x?
17:03:34brixenhere's the issue, 4.2.1 has an issue compiling with -O2, so once it builds, rbx won't run
17:03:55brixenevan is looking into that, but you should build with 4.0 for now
17:04:12jerbrixen, hrmm ok
17:04:24jerwell, new stuff: http://pastebin.com/m25753866
17:04:53jerlooks like a missing include
17:06:29brixenjer: how are you configuring it to use 4.2?
17:08:19jeryes
17:08:38jeri'll switch to 4.0
17:27:37jerbrixen, same thing
17:27:42jeras in the pastebin link
17:28:09brixenjer: ok, let me try to build head
17:29:00jerfixed by including sys/types.h, sys/time.h and sys/resource.h
17:29:04jerin vm/vm.cpp
17:29:08brixenk
17:29:20brixenI'm not seeing the error, which is odd
17:29:43jerhrmm
17:29:43brixenyou're on os x 10.5.7?
17:29:45jerinteresting
17:29:51jer10.5.6, haven't upgraded yet
17:29:56brixenhm
17:30:09jeroh shoot, i was building on a freebsd box
17:30:10jersorry
17:30:36jervm/builtin/compilemethod.cpp line 95 also needed a return NULL; fall-through or errors occur
17:30:44jer(after the assert)
17:30:55brixenahh, ok
17:31:01brixenwell that makes more sense then :)
17:31:22brixencould you put together a patch for these issues?
17:31:42brixenalso, go ahead and build with 4.2.1 and see if rbx runs
17:31:49brixenmaybe it's just gcc on os x
17:32:14jeroh man, a lot of errors after that one
17:32:23jerbrixen, yeah i can make a patch, should just fix them all first =]
17:32:29brixenok
17:32:43brixenI don't think anyone has tried to build on freebsd in a long while
17:32:47jerhttp://pastebin.com/m53a076cc
17:32:48jergood times
17:33:46brixenprobably just clashing on std::tr1 again
17:33:46jeroh wait, g++ was still 4.2.1 and gcc was 4.3.4 hrmm, let's fix that and try
17:34:02jercleans
17:34:25brixenfolks are building on linux ok with 4.3.[234]
17:35:53jeryeah i think it was a problem with the mixed gcc version, that was my mistake
17:36:40jerfreebsd is more strict on missing includes than any other platform i know about
17:36:53brixenyeah, those you'll need I'm sure
17:37:12brixenos x adds a bunch by default so we miss them
17:41:08jergetting better
17:41:09jerhttp://pastebin.com/m933f06c
17:41:17jerjust a missing -pthread
17:42:16brixencool
17:42:37jeraha, adding a -lpthread in vm.make's ex_libs method under PLATFORM =~ /bsd/
17:42:39jerfixed that
17:42:55jerslowly getting there, now compiling all the .rb files
17:50:14jerhrmm looks like another missing include in one of the tests
17:50:58jernope, expects the compiler to define __WORDSIZE methinks, which freebsd's gcc apparently doesn't
17:51:08brixenerg
17:51:20brixenI thought that was standard gcc :(
17:51:46jerwell i'm just working around it for the moment
17:51:51jeri'll investigate more later
17:52:05jerneato, 724 tests run, no errors
17:52:11brixensweet
17:52:35jerjust need to figure out why __WORDSIZE isn't defined, and then i'll have a nice little patchset
17:52:45jeroh i spoke too soon, it started running more tests, see a few F's
17:53:09brixenthat should be the specs
17:56:00dbussinkfew F's could also point at platform differences on freebsd, dunno how regular rubyspec is ran on freebs
17:56:03dbussinkfreebsd
17:56:37brixenyeah, possibly
17:56:55brixenjust distinguishing the vm tests from the specs
18:00:58jerthat was fun, one of the tests killed the box heh
18:01:17brixenheh, oops
18:03:22yakischlobatiger does not have __WORDSIZE either iirc that was a problem for me a little hwile back
18:03:39brixenyakischloba: what did you use?
18:03:54yakischlobammm WORD BIT somethin or other? i assumed that was equiv
18:04:10brixenok
18:04:23yakischlobaya
18:04:23yakischloba./i386/limits.h:#define WORD_BIT 32
18:06:02jerthis crash for anyone else: def x(y) puts y if y == 2**10; x(y + 1); end; x(1) #=> 1024 then core dump
18:06:46jerjust wants to know if it is potentially related to some change i made (while not definitive any answer you give me, it might help point me into the slap myself on the wrist mode, or continue scratching my head, as the backtrace is very hard to interpret =])
18:06:57brixenjer: I get a StackError
18:07:05brixenexception backtrace
18:07:42jerah ok
18:24:21evanmorning.
18:24:37brixenmorning
18:24:51evangot my parents back off to Montana
18:25:07brixencool, did they have a fun time?
18:25:12evanyep
18:25:17brixennice
18:25:17evanwent to a Dodgers game
18:25:30evansunned at the pool yesterday (I managed to get my first minor sunburn of the summer)
18:26:31brixenheh
18:26:45brixenit's sunny there almost all the time!
18:27:27evanhah
18:27:29evanyeah, funny enough
18:27:31evanit's been overcast
18:27:37evanwe've got some june gloom going on
18:27:53brixenhere too
18:28:07brixendark clouds for 2 days but no rain
18:31:13evanhow was your weekend?
18:31:33brixenpretty good
18:31:40brixengot jenny some more seat time
18:31:51brixenshe's going for try 5 to get her endorsement
18:31:54brixenheh
18:31:56evanseat time == motorcycle?
18:31:58brixenyeah
18:32:03evanhehe
18:32:15evanpractice makes... frusteration!
18:32:19brixenindeed
18:32:27brixenshe does really good until test time
18:33:02evannerves I guess.
18:33:06brixenyeah
18:35:39brixenevan: so I'm trying to track this down http://gist.github.com/125952
18:36:11brixenbut I'm not following why @method is a Compiler::Node::Script in ZSuper#bytecode
18:36:25evanlets see...
18:36:26brixenat least I think that's where I was on sat
18:36:45evanah
18:36:50evanit's a script because there is no def
18:36:53evanfor it to track
18:37:04brixenok
18:37:27evanthis is a tricky one
18:37:38evanbecause zsuper is supposed to inject the args
18:37:59brixenyeah
18:38:10evanso what probably needs to happen is that the block args from the define_method need to be presented to zsuper
18:38:20evanand it will look at those to figure out what to push on the stack to call super with
18:39:21evanlikely needs to be a special case
18:39:35evanhm.
18:39:42evani see a problem with doing that though....
18:39:56evandef blah(a)
18:40:07evan 1.times { |b| super }
18:40:08evanend
18:40:16evansuper should send a up, not b.
18:43:01brixenI wonder how jruby does it, they added this spec
18:43:20evanI suspect their zsuper is done in the VM
18:43:29evanthats what MRI does
18:43:32brixenit seems like super can't know what to send until the method is defined
18:43:35brixenah ok
18:44:01evanif we have to move it into the VM
18:44:03evanthats fine
19:20:05brixenwin 22
19:20:11brixenum
20:13:03evanyay kirk!
20:13:09evanhe's doing 1.8.6 commits
20:21:16brixenyep, and specs too
21:11:41jaribevan, brixen: the patch attached to #774 should be good to go
21:12:12jaribi'm not sure what's the best way to implement #call though
21:12:47jaribi.e. class Foo; attr_reader :foo; end; Foo.new.method(:foo).call
21:15:50brixenjarib: ok, I was hoping to get to committing it this weekend but didn't finish the rubyspec stuff I was doing
21:16:01brixenjarib: I'm heading to lunch, but I'll take a look when I return
21:16:48jaribok, no hurry :)
21:17:05brixenheh, I'll hurry, promise :)
21:17:25jaribhehe
23:18:15evanstupid Array#[]= specs
23:18:22evanthey test like 20 cases in one it block
23:19:14brixen:(
23:21:49brixenevan: I was just about to merge the specs
23:21:58evango for it.
23:21:58brixenwhich ones are causing issues?
23:22:04brixenI can fix them first
23:22:18evani'm splitting them out
23:22:25brixenok, I'll wait
23:22:27evanthe last one tests the lhs being a Range
23:22:32evanbut with every kind of rhs in one it block
23:22:46evanthere needs to be one it block per kind of rhs
23:22:57brixendefinitely
23:23:39evanyou can merge
23:23:39evanthats fine
23:23:50evani'm a big boy, i can fix conflicts if need be
23:23:51evan:D
23:24:04brixensure, n/p
23:24:15brixenI'm releasing a gem and bumping the rubyspec version too
23:24:49evansounds good.