From surrender_it at yahoo.it Tue Jan 4 07:13:25 2005 From: surrender_it at yahoo.it (gabriele renzi) Date: Tue Jan 4 07:11:23 2005 Subject: [Yarv-devel] segfault on mingw Message-ID: <20050104121325.8887.qmail@web50108.mail.yahoo.com> Hi, it seem that a strange bug appeared in latest revisions: c:/ruby/bin/ruby19 -I. ./test/runner.rb Loaded suite test Started .....*eval*:11: warning: already initialized constant C *eval*:11: warning: already initialized constant C *eval*:11: warning: already initialized constant C <*YARV eval*>:14: warning: already initialized constant C <*YARV eval*>:14: warning: already initialized constant C <*YARV eval*>:14: warning: already initialized constant C ...............E................................................................ ..... Finished in 1.803 seconds. c:/ruby/lib/ruby/1.9/test/unit/util/backtracefilter.rb:16: [BUG] Segmentation fa ult ruby 1.9.0 (2005-01-03) [i386-mingw32] This application has requested the Runtime to terminate it in an unusual way. Please contact the application's support team for more information. make: *** [test-all] Error 3 Interesting things: the bug does not appear if "ruby -I. ./test/runner.rb" is run from gdb, or from cmd.exe. It only appears when the command is run from make, or from sh.exe, so I think it relates to some environment stuff reflecting on ruby (i.e. different stack limit). I don't have a clue on how to get a core file on windows sorry :/ environment: win xp pro SP2 athlon 2000+ sh --version GNU bash, version 2.04.0(1)-release (i686-pc-msys) gcc --version gcc.exe (GCC) 3.4.2 (mingw-special) ruby19 -v ruby 1.9.0 (2005-01-03) [i386-mingw32] yarv rev:109 ===== icq #69488917 ___________________________________ Nuovo Yahoo! Messenger: E' molto pi? divertente: Audibles, Avatar, Webcam, Giochi, Rubrica Scaricalo ora! http://it.messenger.yahoo.it From ko1 at atdot.net Tue Jan 4 08:15:53 2005 From: ko1 at atdot.net (SASADA Koichi) Date: Tue Jan 4 08:15:01 2005 Subject: [Yarv-devel] segfault on mingw In-Reply-To: <20050104121325.8887.qmail@web50108.mail.yahoo.com> References: <20050104121325.8887.qmail@web50108.mail.yahoo.com> Message-ID: <41DA9709.1000404@atdot.net> Hi. gabriele renzi wrote: > Hi, > it seem that a strange bug appeared in latest > revisions: Thank you. Can you try again with option "-v" > c:/ruby/bin/ruby19 -I. ./test/runner.rb -> c:/ruby/bin/ruby19 -I. ./test/runner.rb -v > Interesting things: > the bug does not appear if "ruby -I. ./test/runner.rb" > is run from gdb, or from cmd.exe. It only appears when > the command is run from make, or from sh.exe, so I > think it relates to some environment stuff reflecting > on ruby (i.e. different stack limit). > mmm... From surrender_it at yahoo.it Tue Jan 4 10:55:23 2005 From: surrender_it at yahoo.it (gabriele renzi) Date: Tue Jan 4 10:53:21 2005 Subject: [Yarv-devel] segfault on mingw In-Reply-To: <41DA9709.1000404@atdot.net> Message-ID: <20050104155523.44940.qmail@web50104.mail.yahoo.com> --- SASADA Koichi ha scritto: > Hi. > > gabriele renzi wrote: > > Hi, > > it seem that a strange bug appeared in latest > > revisions: > > Thank you. > > Can you try again with option "-v" ah, sure, sorry, the test with error is: test_xstr(TestBIN): E full output from make test-all (it is the same if run from sh directly) attached. hope this helps. ===== icq #69488917 ___________________________________ Nuovo Yahoo! Messenger: E' molto pi? divertente: Audibles, Avatar, Webcam, Giochi, Rubrica Scaricalo ora! http://it.messenger.yahoo.it -------------- next part -------------- A non-text attachment was scrubbed... Name: test_out.log Type: application/octet-stream Size: 3240 bytes Desc: test_out.log Url : http://rubyforge.org/pipermail/yarv-devel/attachments/20050104/f98c43fb/test_out-0001.obj From ko1 at atdot.net Wed Jan 5 01:51:24 2005 From: ko1 at atdot.net (SASADA Koichi) Date: Wed Jan 5 01:51:31 2005 Subject: [Yarv-devel] segfault on mingw In-Reply-To: <20050104155523.44940.qmail@web50104.mail.yahoo.com> References: <20050104155523.44940.qmail@web50104.mail.yahoo.com> Message-ID: <41DB8E6C.1080307@atdot.net> Hi, > ah, sure, sorry, the test with error is: > test_xstr(TestBIN): E mmm... I have no idea about this problem. YARV has some problems about environment management. And I'll fix it. After that, problems will be vanished ... I hope :) -- // SASADA Koichi at atdot dot net // From surrender_it at yahoo.it Wed Jan 5 07:09:38 2005 From: surrender_it at yahoo.it (gabriele renzi) Date: Wed Jan 5 07:07:34 2005 Subject: [Yarv-devel] segfault on mingw In-Reply-To: <41DB8E6C.1080307@atdot.net> Message-ID: <20050105120938.39171.qmail@web50110.mail.yahoo.com> --- SASADA Koichi ha scritto: > Hi, > > > ah, sure, sorry, the test with error is: > > test_xstr(TestBIN): E > > mmm... I have no idea about this problem. I indagated this further, the problem seem to just be `ls`. If test_xstr was added recently the problem can be just that, otherwise it seem that the path is somewhat messed up. Just replacing 'ls' with the full path seem to work. (yes, I have ls.exe in my path :) Thanks for working on this. ===== icq #69488917 ___________________________________ Nuovo Yahoo! Messenger: E' molto pi? divertente: Audibles, Avatar, Webcam, Giochi, Rubrica Scaricalo ora! http://it.messenger.yahoo.it From ko1 at atdot.net Wed Jan 5 23:45:55 2005 From: ko1 at atdot.net (SASADA Koichi) Date: Wed Jan 5 23:46:01 2005 Subject: [Yarv-devel] segfault on mingw In-Reply-To: (Your message of "Wed, 5 Jan 2005 13:09:38 +0100 (CET) ") <20050105120938.39171.qmail@web50110.mail.yahoo.com> References: <20050105120938.39171.qmail@web50110.mail.yahoo.com> Message-ID: <20050106134555.5a10713%ko1@atdot.net> gabriele renzi wrote : [ Re: [Yarv-devel] segfault on mingw ] at Wed, 5 Jan 2005 13:09:38 +0100 (CET) Hi, > I indagated this further, the problem seem to just be > `ls`. > If test_xstr was added recently the problem can be > just that, otherwise it seem that the path is somewhat > messed up. > Just replacing 'ls' with the full path seem to work. > (yes, I have ls.exe in my path :) nice :) Would I replace "ls" to "ruby -e '1'" or something? -- // SASADA Koichi at atdot dot net // From surrender_it at yahoo.it Thu Jan 6 07:15:10 2005 From: surrender_it at yahoo.it (gabriele renzi) Date: Thu Jan 6 07:13:28 2005 Subject: [Yarv-devel] segfault on mingw In-Reply-To: <20050106134555.5a10713%ko1@atdot.net> Message-ID: <20050106121510.99550.qmail@web50109.mail.yahoo.com> --- SASADA Koichi ha scritto: > gabriele renzi wrote : > [ Re: [Yarv-devel] segfault on mingw ] > at Wed, 5 Jan 2005 13:09:38 +0100 (CET) > > Hi, > > > I indagated this further, the problem seem to just > be > > `ls`. > > If test_xstr was added recently the problem can > be > > just that, otherwise it seem that the path is > somewhat > > messed up. > > Just replacing 'ls' with the full path seem to > work. > > (yes, I have ls.exe in my path :) > > nice :) > > Would I replace "ls" to "ruby -e '1'" or something? I'd say "ruby -v", but since ENV['path'] is completely messed up anyway I won't mind about it :) In test_xstr ENV['path'] is something terrible like: ??" which would mean: 168.chr, 216.chr, 34.chr. This is always the same result, if that matters, and I think it is part of the "environment problems" of yarv, so I know I have to wait :) Onr more thing (sorry if I'm so pedantic :) extconf.rb --help says: others(dengerous)(default disable): * fake-inline-method-cache * inline-method-cache while -DINLINE_METHOD_CACHE is actually enabled. ===== icq #69488917 ___________________________________ Nuovo Yahoo! Messenger: E' molto pi? divertente: Audibles, Avatar, Webcam, Giochi, Rubrica Scaricalo ora! http://it.messenger.yahoo.it From ko1 at atdot.net Sun Jan 9 20:04:28 2005 From: ko1 at atdot.net (SASADA Koichi) Date: Sun Jan 9 20:03:27 2005 Subject: [Yarv-devel] [ANN] YARV: Yet Another RubyVM 0.1.0 Message-ID: <20050110100428.27329f4%ko1@atdot.net> Hi, I released YARV: Yet Another RubyVM 0.1.0. YARV Home page: http://www.atdot.net/yarv/ YARV 0.1.0 : http://www.atdot.net/yarv/yarv-0.1.0.tar.gz * What's this? YARV is Virtual Machine for Programming Language Ruby. The goal of this project is to develop the fastest Virtual Machine for Ruby in the world. You can see how fast YARV works. http://www.atdot.net/yarv/bench-20050110-rev120-cygwin.txt * YARV can: - Parse some ruby scripts - Execute some ruby scripts - AOT compiler (Ruby -> C) (very partial) * YARV can't: - Parse all ruby scripts - Execute all ruby scripts Especially, YARV has many incompatibility with current ruby interpreter. * How to install and use: See http://www.atdot.net/yarv/ * Remarks We have mailing list to discuss about YARV in English. http://rubyforge.org/mailman/listinfo/yarv-devel This project is supported by IPA (Information-technology Promotion Agency, Japan) "Exploratory Software Project (youth)". Thanks, -- // SASADA Koichi at atdot dot net // From g-stuff at bk.iij4u.or.jp Wed Jan 19 10:25:04 2005 From: g-stuff at bk.iij4u.or.jp (Ryuichi Maeno) Date: Wed Jan 19 10:22:34 2005 Subject: [Yarv-devel] (no subject) Message-ID: <41EE7BD0.4040406@bk.iij4u.or.jp> From mdg at absolute-performance.com Thu Jan 20 10:07:02 2005 From: mdg at absolute-performance.com (Michael Garriss) Date: Thu Jan 20 09:59:40 2005 Subject: [Yarv-devel] ping Message-ID: Is this mailing list active? Is the development of Yarv related to Rite in anyway? -- Michael Garriss Senior Software Engineer Absolute Performance, Inc. From ko1 at atdot.net Fri Jan 21 03:10:42 2005 From: ko1 at atdot.net (SASADA Koichi) Date: Fri Jan 21 03:10:27 2005 Subject: [Yarv-devel] ping In-Reply-To: (Your message of "Thu, 20 Jan 2005 08:07:02 -0700") References: Message-ID: <20050121171042.5e81533%ko1@atdot.net> Michael Garriss wrote : [ [Yarv-devel] ping ] at Thu, 20 Jan 2005 08:07:02 -0700 > Is this mailing list active? not active because I have always discuss on Japanese (yarv-dev). if you have any topic to talk, pls post. > Is the development of Yarv related to > Rite in anyway? I hope so. -- // SASADA Koichi at atdot dot net // From surrender_it at yahoo.it Sat Jan 22 06:46:00 2005 From: surrender_it at yahoo.it (gabriele renzi) Date: Sat Jan 22 06:43:20 2005 Subject: [Yarv-devel] benchmark crashing and register questions Message-ID: <20050122114600.4886.qmail@web50104.mail.yahoo.com> hi, I was playing with the latest yarv(rev123) with ruby 1.9.0 (2005-01-21) compiled with gcc 3.4.2 (mingw-special) on my windows box (XP pro SP2) over my Athlon cpu (AthlonXP 2000+) I noticed yarv crashes on its own benchmarks: user system total real ruby 29.162000 0.530000 29.692000 ( 30.233000) ./benchmark/bm_so_object.rb:57: [BUG] BUG: unknown node: NODE_SUPER ruby 1.9.0 (2005-01-21) [i386-mingw32] This application has requested the Runtime to terminate it in an unusual way. Please contact the application's support team for more information. ** benchmark failure: Invalid argument ./benchmark/run.rb:74:in `gets' ./benchmark/run.rb:74:in `benchmark' ./benchmark/run.rb:71:in `popen' ./benchmark/run.rb:71:in `benchmark' ./benchmark/run.rb:85:in `yarv_exec' ./benchmark/run.rb:37:in `bm' ./benchmark/run.rb:96 ./benchmark/run.rb:94:in `each' ./benchmark/run.rb:94 actually, it crashes like this if run from a benchmark as simple as: class T def initialize(bar) end end class NT< T def initialize super 'foo' end end If I recall correct SUPER/ZSUPER are broken at the moment, so this could be normal, but then, should'nt testcases for this appear in test/test_class.rb ? Another little thing: I noticed there are quite a bit of "register" variables, but I wonder if they are worthy in a register starved architecture like x86. Also, given that lots of stuff is in registers.. koichi, have you considered caching the top of the stack too? It seems a resonably simple optimization to me, just requiring tuning of instructions, and it could be useful in architectures with lots of registers such as x86-64/IA-64/ppc. As usual, I beg you pardon for being total lame in C, so if my questions seem really stupid try to consider answers as "yarv documentation for dummies" :) ===== icq #69488917 ___________________________________ Nuovo Yahoo! Messenger: E' molto pi? divertente: Audibles, Avatar, Webcam, Giochi, Rubrica Scaricalo ora! http://it.messenger.yahoo.it From joaopedrosa at gmail.com Sat Jan 22 07:02:08 2005 From: joaopedrosa at gmail.com (Joao Pedrosa) Date: Sat Jan 22 06:59:30 2005 Subject: [Yarv-devel] benchmark crashing and register questions In-Reply-To: <20050122114600.4886.qmail@web50104.mail.yahoo.com> References: <20050122114600.4886.qmail@web50104.mail.yahoo.com> Message-ID: Hi Gabriele, On Sat, 22 Jan 2005 12:46:00 +0100 (CET), gabriele renzi wrote: > hi, > > I was playing with the latest yarv(rev123) > with ruby 1.9.0 (2005-01-21) > compiled with gcc 3.4.2 (mingw-special) > on my windows box (XP pro SP2) > over my Athlon cpu (AthlonXP 2000+) > > I noticed yarv crashes on its own benchmarks: I haven't had good luck with gcc 3.4 and its tools. I don't know for sure, but my bad experience was with Ruby on Linux Ubuntu. Once I installed the common gcc 3.3 tools (the default), Ruby got stable for me... So I don't know if the problem that you are getting is a too early incompatibility to tackle or not. Maybe you could try gcc 3.3 or the default... I don't know. :-) I'm just checking out YARV, still. It's good to talk about it. I hope YARV becomes Ruby 2.0 (Rite). :-) Nice job SASADA. I appreciate it. Cheers, Joao From ko1 at atdot.net Sat Jan 22 07:54:43 2005 From: ko1 at atdot.net (SASADA Koichi) Date: Sat Jan 22 07:53:29 2005 Subject: [Yarv-devel] benchmark crashing and register questions In-Reply-To: (Your message of "Sat, 22 Jan 2005 12:46:00 +0100 (CET) ") <20050122114600.4886.qmail@web50104.mail.yahoo.com> References: <20050122114600.4886.qmail@web50104.mail.yahoo.com> Message-ID: <20050122215443.5fba428%ko1@atdot.net> gabriele renzi Sat, 22 Jan 2005 12:46:00 +0100 (CET) / [Yarv-devel] benchmark crashing and register questions Hi, >I was playing with the latest yarv(rev123) >with ruby 1.9.0 (2005-01-21) >compiled with gcc 3.4.2 (mingw-special) >on my windows box (XP pro SP2) >over my Athlon cpu (AthlonXP 2000+) > >I noticed yarv crashes on its own benchmarks: > > > user system total real >ruby 29.162000 0.530000 29.692000 ( 30.233000) >./benchmark/bm_so_object.rb:57: [BUG] BUG: unknown >node: NODE_SUPER >ruby 1.9.0 (2005-01-21) [i386-mingw32] > > >This application has requested the Runtime to >terminate it in an unusual way. >Please contact the application's support team for more >information. > >** benchmark failure: Invalid argument >./benchmark/run.rb:74:in `gets' >./benchmark/run.rb:74:in `benchmark' >./benchmark/run.rb:71:in `popen' >./benchmark/run.rb:71:in `benchmark' >./benchmark/run.rb:85:in `yarv_exec' >./benchmark/run.rb:37:in `bm' >./benchmark/run.rb:96 >./benchmark/run.rb:94:in `each' >./benchmark/run.rb:94 > It's known bug. This problem occurs when scripts use "rb_call_super" YARV doesn't support it. >Another little thing: I noticed there are quite a bit >of "register" variables, but I wonder if they are >worthy in a register starved architecture like x86. > >Also, given that lots of stuff is in registers.. >koichi, have you considered caching the top of the >stack too? It seems a resonably simple optimization to >me, just requiring tuning of instructions, and it >could be useful in architectures with lots of >registers such as x86-64/IA-64/ppc. Yes, some "VM registers" saved on machine stack (not on machine registers). On PPC, I tried but not saved on registers. I must optimize for each architecture. GCC has some schemes to choose registers to C variable, so I try it. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Local-Reg-Vars.html#Local-Reg-Vars > > >As usual, I beg you pardon for being total lame in C, >so if my questions seem really stupid try to consider >answers as "yarv documentation for dummies" :) I will write documents, but on Japanese :-P -- // SASADA Koichi at atdot dot net //