Index

Show enters and exits. Hide enters and exits.

00:01:19boyscoutCI: rubinius: 92c6a15 successful: 3460 files, 13517 examples, 41094 expectations, 0 failures, 0 errors
00:52:58boyscoutFixed Ping specs. - 448642d - Brian Ford
00:52:59boyscoutAdded ping.rb from MRI standard library. - deeea85 - Brian Ford
00:52:59boyscoutRemove obsolete CompiledMethod specs. - e98fd0d - Brian Ford
00:52:59boyscoutRemoved old failing Module rbx specs. - 7368527 - Brian Ford
00:52:59boyscoutSome cleanups for default.mspec. - b5df30a - Brian Ford
00:53:26brixenevan: I think that's it for the spec cleanups
00:53:37brixenshall we let it simmer for morning?
00:53:44evanyeah
00:54:23brixenwe only have ~540 tags now
00:54:58brixenoh wait, 856... hmm I wonder
00:55:43brixensome of those are incomplete tags I think
00:56:41brixen754 from --list-all with the ci_files
00:57:33brixen530 for fails, critical, unstable
00:58:35brixenso difference are incomplete
00:58:59brixenso yeah, about 530 failing out of 14k
00:59:54evanpretty good.
01:03:53brixenhmm, some of these tags look stale
01:04:00slavahi fellas
01:04:43brixenhi slava
01:17:29boyscoutCI: Commit b5df30a failed. http://github.com/evanphx/rubinius/commit/b5df30a0620837d6882e7a2ef984a836e9285c53
01:41:06boyscoutCleaning up some specs and CI tags. - f976513 - Brian Ford
01:41:07boyscoutQuarantine Ping spec. - 437efdd - Brian Ford
01:42:57evani'm checking out the frozen weirdness on elle
01:44:05brixenok
02:05:38boyscoutCI: Commit 437efdd failed. http://github.com/evanphx/rubinius/commit/437efdd12e24c443e1be07bbb345498a9e88700d
07:00:28boyscoutHave the JIT check frozen status of self before setting ivars - b9662ac - Evan Phoenix
07:00:48slavayo evan pushing JIT patches at midnight?
07:10:59boyscoutCI: rubinius: b9662ac successful: 3457 files, 13581 examples, 41201 expectations, 0 failures, 0 errors
07:23:37dbussinkevan: the spurious frozen error?
15:08:33khaasebrixen: http://ci.finn.de/builds/rbx_redmine-trunk_mysql
15:14:52kronos_vanokhaase, awesome
15:17:18khaasekronos_vano: no, not that awesome
15:17:24khaasekronos_vano: it does not actually pass
15:17:31khaasekronos_vano: just the wrong return value
15:17:47kronos_vanoWhy it is green then?
15:17:55khaasewrong return value
15:18:26khaaseall failures seem IO/FileSystem related
15:18:39khaases/FileSystem/file system
16:17:53brixenkhaase: ah ok, I didn't have any of the scm's installed
16:18:22brixenkhaase: that ci formatted output is terrible! the red bt is hard to read and it strips the failure #
16:18:35brixen:(
16:18:39khaasebrixen: blame schmidtwisser
16:18:50khaasedon't have access to the ci
16:18:58brixenkhaase: I don't know who that is, could you blame him for me? :)
16:19:00khaasei mean, like admin/root/whatever
16:19:10khaasei'll do that. he just left
16:19:18brixenok :)
16:19:50brixenlike, I already know it's a failure, the whole bt needs to be easy to read, not red!
16:20:55kronos_vanobrixen, can you give me your mail?
16:22:58brixenkronos_vano: bford at engineyard
16:23:12evanthe IO failure is easy to fix
16:23:18evani can do that now.
16:26:24brixenkhaase: also, you will need a rake distclean; rake to rebuild oniguruma
16:26:58brixenkhaase: or you can cd vm/external_libs/onig; make clean; cd <root> rake
16:28:20khaaseok
16:28:39khaaseany idea what causes the io stream issues?
16:28:55boyscoutToggle binmode on the read stream of a pipe - 1be0177 - Evan Phoenix
16:28:56evankhaase: that ^^
16:29:15khaaseevan: w00t
16:34:37evanok, so no open issues
16:34:51evanand i don't want to make any changes after rc5 if we can help it.
16:35:06evanso are we ok with redmine not running green?
16:35:22evanactually, i need to check rails 3
16:37:22boyscoutCI: rubinius: 1be0177 successful: 3457 files, 13581 examples, 41201 expectations, 0 failures, 0 errors
16:38:21brixenI'm running redmine here
16:38:47brixenI don't have all the scm's and don't much feel like installing all that stuff, so I'm not going to test those atm
16:39:17brixenrails 3 is the more important one
16:39:34brixenI wouldn't worry about redmine tests
16:39:58brixenwe need to draw a line somewhere or we'll be on a perpetual treadmill
16:40:09brixengetting asymptotically close to 1.0 :)
16:41:52evank
16:42:07evandamnit.
16:42:14evanrails 3 keeps adding gems.
16:42:29evanlike the fucking added the json gem.
16:42:32evanwhich we don't run
16:42:35evanbecause it's terrible.
16:43:26Zoxcit uses callcc? :D
16:45:19brixenevan: the gem itself, or that script that doesn't compile?
16:45:42evanthe json gem itself seems to use RBASIC()
16:46:07brixenugh
16:46:32evani got it to install with --without mri
16:46:37evanthey have those gems grouped
16:55:17khaaseevan: 1.0 not running all redmine tests is fine with me (actually, it seems like all tests might pass now)
16:55:35evanok
16:55:45evanlooks like there are a couple of broken rails 3 tests now.
16:55:55evanhopefully they'll be easy.
16:56:47khaaseyeah, just implement callcc... piece a' cake ;)
16:57:21evan?
16:57:27evanit's nothing to do with callcc.
16:57:37khaase"18:43 Zoxc: it uses callcc? :D"
16:57:43khaasealso, just kidding
17:00:34evanarg. i'm getting a "No current exception" bug.
17:00:36evani hate these.
17:03:28evanhm, i thought I fixed this...
17:32:08kronos_vanobrixen, tnx!
17:35:12brixenkronos_vano: yw!
19:07:40wayneeseguinDefiler: ello!
19:07:47Defileryarr
19:07:49wayneeseguinIs there anything you can think of that would cause rubygems extensions to NOT load ?
19:08:09Defilernot load, no error?
19:08:14wayneeseguinhrm
19:12:14wayneeseguinDefiler: http://gist.github.com/392576
19:12:36wayneeseguinespecially the hpricot one
19:13:37DefilerOh, sorry, I misunderstood which kind of extension you meant
19:13:41Defilernot extensions to rubygems itself
19:14:19Defilerwayneeseguin: put :verbose: really in your .gemrc
19:14:28Defilerand then try installing hpricot again
19:14:41Defileralso, hpricot works on 1.9?
19:15:42Defilerhopefully gem_make.out has some useful output in it too
19:16:10Defilersad that there is spam here http://isitruby19.com/hpricot
19:24:13wayneeseguinDefiler: OK
19:24:35wayneeseguinugh yeah :( spam sux
19:26:37dbussinkalso the author names don't always make sense
19:26:46dbussinkwayneeseguin: any idea where it gets those from?
19:32:10wayneeseguindbussink: no?
19:32:37wayneeseguinlikely from the gravitar
19:36:52wayneeseguinDefiler: You sir, rock. That :verbose: worked awesomely.
19:36:57wayneeseguinDefiler: http://gist.github.com/392597
19:37:03wayneeseguinI've never seen anything like that :)
19:37:14wayneeseguinCan you guess what the issue was?
19:37:16wayneeseguin::grin::
19:40:48dbussinkwayneeseguin: 32 / 64 bit linking to each other?
19:42:42wayneeseguindbussink: Nope :) The guy had set somehow PATH="~/.rvm/scripts/:...."
19:42:46wayneeseguinwhich makes no sense
19:43:00wayneeseguinso 'make install' was finding the ~/.rvm/scripts/install instead of /usr/bin/install
19:43:06wayneeseguinWEIRD
20:09:55brixenkhaase: redmine should use RedCloth or RedCloth3 should be patched
20:09:57brixenkhaase: http://redmine.ruby-lang.org/issues/show/3217
20:11:16khaasebrixen: yeah, there are plans for replacing redcloth3 with redcloth in 1.0, but i think those are not high priority.
20:11:30brixenok
20:12:02brixenI guess that means there are maybe plans for supporting rbx in 1.0 :)
20:12:44evanrubinius and 1.9
20:12:51evanneither will match <
20:12:53evanin that case
20:13:36khaaseisn't it fixed in http://github.com/finnlabs/redmine/commit/b2c0885a509a517b?
20:13:53khaasethat commit already made it to redmine trunk, afaik
20:14:26evanoh, probably.
20:14:46brixenkhaase: oh, I'll try that, but I applied that patch before and it did not fix all issues that appeared to be RE related
20:15:24evanarg. my super fix broke rails
20:15:27brixenin fact, yes, I have that patch applied
20:15:34evanbecause rails tests have, ready for it?
20:15:37brixendamn rails
20:15:47khaaseyeah, but switching to redcloth 4 will be painfull afaik
20:15:59evandefine_method(...) { class << obj; def method; end; end; }
20:16:26brixenwtf
20:16:50evanand the way I fixed super was to keep the name on a CompiledMethod accurate
20:17:07evanwhich, in the case of a define_method, means walking down into any sub CompiledMethods and changing the name
20:17:16evanbecause I assumed i'd be changing the name of blocks
20:17:34evanbut the CMs for the sclass body and method clearly shouldn't be changed.
20:17:48brixenyeah
20:18:21evanso now I need to tag CMs that are for blocks
20:18:25evanso that i only change those.
20:19:26brixenwe have @for_block in the generator...
20:19:43evani need it at runtime
20:19:49brixenah
20:19:55brixenI mean, put it into the CM
20:20:12evanright
20:20:40evanI think I stuck the metadata slot into CM
20:20:43evani can put it into there.
20:20:55evanjust need to make sure that slot is roundtriping fine
20:21:46evanhm
20:22:02evanivars themselves seem to be roundtripping it some cases...
20:22:39evanhm, yeah, we've been using __ivars__ for that
20:22:44evanwhich i think is wrong, isn't it....
20:23:18evanyeah, ack.
20:23:24evanwe have a __ivars__ accessor
20:23:27evanon CompiledMethod
20:23:46evanwell, i can just change it's named to metadata
20:23:50evanand use it.
20:23:54evanwithout breaking any formatting
20:24:24brixenheh, oops, we've been dragging that __ivars__ around for a while
20:24:39brixen2 little lines in compiled_file.rb
20:25:01evanyep.
20:25:17evanbetter to have it be explicitly metadata
20:25:23evansince it's form is highly restricted
20:25:35evansince it has to be able to be put into a .rbc
20:25:40brixenyeah
20:26:09brixenstepping through redmine is so fun/painful/idk, it hurts
20:26:14brixenI want a debugger :/
20:27:47brixenI wonder if this is an issue with quoted printable
20:33:49brixenevan: poke
20:54:01brixenevan: comparing these two TMail outputs http://gist.github.com/392664 would you guess a mime encoding issue?
20:55:34evanyes
20:55:42brixenk
20:55:49evanquoted-printable
20:55:50evanyou dog you.
21:23:25josiahhey, I have a question about code changes
21:23:39evanok
21:25:15josiahi was mistakenly working with an old version, and fixed a problem that someone else had fixed as well. The changes I made were a few lines of ruby, the other changes added the "vm_check_super_callable" native method
21:25:50evanok
21:25:58josiahI assume the native one is faster, but I also know that one of the goals is to implement as much as possible in ruby
21:26:48josiahshould I make up a patch that replaces the native method with the ruby one, or just leave it alone?
21:26:58evanit's much more difficult (maybe not even possible) to do it ruby for that
21:27:15evanyou'll see that vm_check_super_callable uses the same logic as normal method dispatch to answer the question
21:27:20evanyou must leave it alon.
21:27:21evanalone.
21:27:29evanyou're code is probably not right.
21:27:42evani'm curious to see it though
21:27:44brixenjosiah_: you could show us your code
21:27:49brixenevan: jinx
21:27:50brixenheh
21:28:53josiahin "variable_scope.rb" around line 83 i had "found_method = sup.find_method_in_hierarchy(@method.name)
21:28:53josiah return found_method != nil && found_method.visibility != :undef"
21:29:10josiahbasically just adding a check for the method visibility
21:30:25evanas I recall, there is another case that that doesn't work for
21:30:30evanbut I don't recall it off hand
21:31:55josiahfrom what I could see the current version doesn't use the "super_method_defined?" function for anything anymore
21:33:18evanthats correct
21:33:20evanit doesn't
21:33:47evanthe reason it was moved into the VM is because we were trying to duplicate the logic the VM was doing anyway
21:33:50evanand duplicating it badly
21:34:06evanit's much less buggy in the current form
21:37:22josiahcool, I hadn't noticed that it was so similar to the dispatch function. thanks
22:30:47boyscoutTrack the current exception better and more often - 4aac89d - Evan Phoenix
22:30:47boyscoutTrack if a CM is for a block properly finally - f653ab9 - Evan Phoenix
22:31:03evanwell, still a few rails 3 bugs
22:31:13evanstill a weird exception one i have to work on.
22:32:54brixenso, "#".inspect != "#".dump :(
22:33:11brixenor something
22:33:15brixenthis is so confusing
22:33:46evanin MRI it's the same
22:33:55brixenthe String#inspect specs suck for one
22:34:05brixenlook at the String#inspect specs
22:34:07evanlooks like we're escaping # in both though
22:34:08evanwe shouldn't be.
22:34:19brixenwe are matching that spec
22:34:37brixen(34.chr..36.chr).to_a.to_s => "\"\#$"
22:35:08brixen"#".dump => "\"#\""
22:36:20brixenhm, so what does Array#to_s actually use...
22:37:33brixenrb_output_fs
22:40:00evanwell, i'm headed out
22:40:04evanbbiab.
22:44:31boyscoutCI: rubinius: f653ab9 successful: 3457 files, 13581 examples, 41201 expectations, 0 failures, 0 errors
22:50:33brixenso, if you "#" << "$" or "@", you get "\#"
22:50:42brixenbrilliant
22:52:44brixenno, I'm wrong, that's only inspecting the string
22:52:49brixenwhew