Index

Show enters and exits. Hide enters and exits.

00:00:36brianmarioevan: just read the rubyconf slides - pretty awesome
00:01:14brianmariooh and it says "yajl-json" in the slides, but the gem is actually "yajl-ruby" (not that it matters much)
00:01:16brixenevan: hrm, seeing random hangs even after clean build
00:01:18brixenhttp://gist.github.com/242349
00:01:26brixenevan: can I get more info on this?
00:01:53evani don't see it hung
00:01:59evanyour backtrace isn't stuck
00:02:05brixenyeah, I know
00:02:15brixenbut it stuck on that dot for like 3 minutes
00:02:29brixenand I didn't do anything but attach with gdb
00:03:26brixenand that's a different spot in the specs from the last hang
00:03:27evanwhat about CPU?
00:03:30evanis it burning up CPU?
00:03:35brixenyeah
00:03:39evanyou can't be attached to GDB to tell
00:03:39evanok
00:03:40brixenwell, about 50%
00:03:47evanwell, thats why your backtrace doesn't say
00:03:53evanyou need to start and stop it
00:03:55evaner.
00:03:57evanbreak in
00:03:59evanthen do
00:04:02evancontinue
00:04:08evanlet in run for a few seconds
00:04:11evanthen ^C
00:04:13evancheck it again
00:04:17evanyou need to get samples for where it's looping
00:04:21evanto figure out why it's looping
00:04:22brixenk
00:06:03evanhm, it's in the IO#gets spec
00:06:04evanit looks like
00:06:06evanI notice.
00:06:40brixenyeah
00:08:40brixenmaybe related to send_super?
00:09:02brixenI break in and do n(ext) but I end up in a jitted method I think
00:09:19brixenframes are ?? and I can't continue nexting
00:09:23evanyeah
00:09:26evanthats JIT'd frames.
00:09:30evantry running with -Xint
00:09:33evansee if you can get it
00:09:36brixenk
00:11:47brixenblast, it's totally random it seems
00:22:53evanARG.
00:35:16evanwell, i think i've got the rdoc/yaml bug isolated.
00:38:46brixensweet
00:40:17evanrdoc uses Struct.new
00:40:21evanto create some struct classes
00:40:33evanand yaml seems to have specific support for struct style classes
00:40:45evani don't yet get what exactly that means
00:49:23evansyck has it's own constant finder.
00:50:49brixenwhat the hell?
00:51:19evanand it's got a bug.
00:51:28evanwell, there are 2 bugs.
00:51:33evanwe each have one
00:51:41evanand they collaberated
00:51:44evanto cause the failure.
00:52:01evansyck prepends Struct:: to whatever name it gets and then ask for it to be looked up
00:52:03evanlike, for instance
00:52:11evanStruct::RDoc::Markup::Flow::H
00:52:19evanwell, there is no Struct::RDoc
00:52:31evanthats a bug in syck that it expects that.
00:52:45dwaiteman c++ sucks sometimes
00:52:52evanbut in 1.8, that bug is masked by the fact that rb_const_defined will pick up toplevel constants
00:52:53dwaitewhen's that go-based rewrite starting up again?
00:53:02evanso when it asks for RDoc in Struct
00:53:10evanit gets RDoc at the toplevel.
00:53:14brixenahh
00:53:17brianmariohaha
00:54:20evanbrianmario: yes, a fun little bug.
00:57:39evanujihisa: hi!
00:57:45evanujihisa: great to meet you finally at rubyconf
00:57:49ujihisaevan: hi!!!!!!!!!!!!!!!!
00:57:55ujihisayes!!!!
00:58:43evanare you back in canada now?
00:59:04ujihisayes. Yesterday I came back to Canada.
00:59:49ujihisayeah
00:59:59ujihisaRubyConf was a great opportunity
01:00:44evanyeah, lots of fun as always.
01:00:51evanit's always the highlight of my year.
01:03:42ujihisahttp://www.flickr.com/photos/kakutani/ http://www.flickr.com/photos/tmaedax/
01:03:54ujihisaabsolutely.
01:04:38evanRAWR
01:04:42evanfixed the bug.
01:05:12evanso, should I fix syck as well
01:05:12evanprobably.
01:06:59evanwoo
01:07:03evanrdoc gen out of gem works
01:09:32evanaah
01:09:33evani see.
01:09:37evanif you use
01:09:54evanStruct.new("Blah", :a, :b, :c)
01:09:57evanthen you get
01:10:00evanStruct::Blah
01:10:10evanbut if you use
01:10:21evanStruct.new(:a, :b, :c)
01:10:28evanthen there is no Struct::<name>
01:10:31evanbecause you don't pass in a name
01:10:38evanso it gets it's name when you assign it to a constant
01:12:57evanwhat a silly API.
01:22:23boyscoutFix rb_const_[defined|get]. Fixes syck/rdoc generation. - cb33a55 - Evan Phoenix
01:22:23boyscoutDelay loading yaml in rubygems, reduces startup time - 2ed9e2f - Evan Phoenix
01:22:27evanyay rdoc gen!
01:24:15brixensweet
01:24:29brixenjust installed yajl-ruby fine and ran it
01:24:42brixener sweet was the rdoc gen :)
01:25:08boyscoutCI: 2ed9e2f success. 3004 files, 11464 examples, 35591 expectations, 0 failures, 0 errors
01:25:20evansweet to both
02:07:00boyscoutGracefully fail from a JIT error. Fixes #88. - 96907cc - Evan Phoenix
02:09:43boyscoutCI: 96907cc success. 3004 files, 11464 examples, 35591 expectations, 0 failures, 0 errors
03:21:40boyscoutAdd another benchmark and a util script - 599d883 - Evan Phoenix
03:21:41boyscoutAdd rdoc, rake, and rake-compiler and preinstalled gems - 56edbca - Evan Phoenix
03:23:40boyscoutCI: 56edbca success. 3004 files, 11464 examples, 35591 expectations, 0 failures, 0 errors
05:10:59boyscoutKeep command stubs for preinstall gems - 3bb6ecd - Evan Phoenix
05:12:29boyscoutCI: 3bb6ecd success. 3004 files, 11464 examples, 35591 expectations, 0 failures, 0 errors
07:38:37brixenholy freakin marshmallows batman, I believe I have an install
07:38:48brixenrake is teh suck
07:38:54brixenI have to do everything
07:39:05brixenrbx runs but still some paths not right
07:39:10brixenhm, irb...
07:55:34brixenrake gem installs and runs
07:57:23brixenyajl-ruby installs (builds C ext) and runs
08:34:03dbussinkrdoc is succesfully churning my cpu cycles when installing some gems :P
09:06:56dbussinkhmm, it's really really slow on rdoc generation, it's still running half an hour later :s
09:09:10brixenit may not be anything to do with rdoc
09:09:20brixenI'm seeing random spinning during spec runs
09:09:31brixenit's not hung, it's just looping somewhere
09:09:41brixencouldn't track it down yet
09:12:14brixendbussink: try running with -Xint
09:12:31dbussinkno way to attach something to see what it's doing?
09:12:37brixensure
09:12:44brixengdb -p <pid>
09:12:49brixenknock yourself out :P
09:13:43brixentranslation: go to town with gdb and see if you can discover the cause :)
10:16:13dbussinkbrixen: that's really hardcore though ;)
10:16:19dbussinkbut i need to get some work done anyway :P
11:04:39ruedbussink: Is it a repeatable thing?
11:07:19dbussinkrue: i'm going to try if it is
11:07:24dbussinkbut doesn't look like it
12:46:49ruedbussink: You were just generating some lib's doc?
12:47:06dbussinkrue: i was gem installing some stuff
12:47:14dbussinkbut tried it again and there were no issues that run
12:47:23dbussinkso i guess it's the same as the issue brixen has seen
13:09:10rueYep...
13:09:17rueI was trying to reproduce without luck
13:10:01scooprtry with a girl
13:18:03rueClever :)
13:28:26scoopron a rare occasion ;)
13:28:53scooprand all you iphone/ipodtouch users should get our new game ;) http://stairdismount.com/
13:30:47dbussinkdoesn't have an iphone
14:25:36rueHa, nice, used to enjoy that
15:28:49Defilerevan: I put this in rubygems trunk: http://github.com/evanphx/rubinius/commit/2ed9e2f202e87ea880e3d3676d138e47a00b3739
15:29:16evanDefiler: oh. I sent a pull request to jb to pull it in
15:29:23Defileroh. heh.
15:29:28evaneither way's fine
15:29:29evanthanks!
15:29:48DefilerNo problem. Anything I can do to halt the spread of the YAML curse :)
15:30:15evanheh
16:48:26dwaitegood morning
16:51:40chadgood morning!
18:11:12rueMorning, morning
18:12:40rueevan: Nonblocking and blocking reads being incompatible is a known, expected thing
22:02:56boyscoutReworked configuring tasks. - 243a775 - Brian Ford
22:02:56boyscoutRemoved obsolete tasks from Rakefile. - 8bd6101 - Brian Ford
22:02:56boyscoutAdded new rake install tasks. - a632b48 - Brian Ford
22:02:56boyscoutMake error msg more descriptive when runtime is not found. - d7f2b97 - Brian Ford
22:02:56boyscoutUpdated install instructions. - 17ba357 - Brian Ford
22:02:57boyscoutMore robust check for not installing over source directory. - b53bba7 - Brian Ford
22:02:59boyscoutPermit rbx lib versions to install without conflict. - cd98be9 - Brian Ford
22:03:01boyscoutFixed running IRB from install. - b8b3505 - Brian Ford
22:03:03boyscoutBump version to magic one dot oh rc one. - fdd6869 - Brian Ford
22:03:05boyscoutFix configure host variable for fetching llvm. - a00c1aa - Brian Ford
22:03:07boyscoutFix release task. - a356153 - Brian Ford
22:06:05boyscoutCI: a356153 success. 3004 files, 11464 examples, 35591 expectations, 0 failures, 0 errors
22:06:28brixensweet jesus
22:06:37imajesnice
22:06:44brixenimajes: shhhhhh
22:06:48brixenI'm not done yet
22:06:54imajesheh
22:07:16brixenI wish I was a web guy..
22:07:29brixenwell, just for the next 10 minutes :)
22:57:21boyscoutUpdate download, build info for one oh rc one. - 3f90a9f - Brian Ford (gh-pages)
23:17:37slavabrixen: does it run on windows?
23:18:01ezmobiusyay! rbx is now installable
23:18:40brixenslava: no, it doesn't yet
23:18:44brixenslava: but soon I hope
23:19:05slavabrixen: congrats on 1.0rc1 in any case :)
23:19:05brixenI've been trying to make the build sane enough to cross-compile
23:19:14brixenslava: thank you!
23:19:17dwaiteI'm the last person at the office; I win
23:19:18imajesyay!
23:19:24brixendwaite: haha
23:19:34brixenimajes: :)
23:19:35dwaite1.0.0rc1?
23:19:36imajesi'm still in my office, dwaite
23:19:40dwaiteI guess its time for me to try rubinius
23:20:01brixendwaite: I didn't get a chance to work on the 1.9.1 build stuff
23:20:02dwaiteup to this point I've just been bluffing on the compilation stuff and sending in random crud as patches
23:20:18dwaitein fact, brixen saw my apple notebook; its a college ruled notebook with an apple sticker on it
23:20:24brixenhehe
23:20:26brixentrue dat
23:20:29brixenstylin'
23:20:39dwaiteI've been hand-rolling unified diffs as notes on my iphone
23:20:58imajesnice
23:21:32dwaiteimajes: I win for my office. I'm not trying to compete on any regional/national/international level
23:21:40imajes:P
23:21:43imajesmy office is a cafe
23:21:46imajesi can stay here all night
23:23:07dwaitewhew, thats rough
23:23:17dwaiteI'm glad I am at a place with a bunch of turkey-loving slackers
23:23:51slavabrixen: so what's next?
23:24:00brixenslava: oh, lots
23:24:08brixenmore jit, more compat, more everything
23:24:13brixenmore windows :)
23:24:26slavamore gc I hope too :)
23:24:26brixenslava: I'm sure we'll want some advice on windows building
23:24:32brixenheh, yeah, more GC
23:24:45brixenI might play with concurrent in immix
23:25:09slavacard marking!
23:27:29brixenslava: what's your gc time to total run time % on a typical app?
23:27:52slavahmm, I've never looked at the GC time on a web app that I run
23:28:13brixenI want to get stats/tuning via BERT solid soon
23:28:28slavawhen compiling code its about 10% I think
23:28:32slavaditto for most benchmarks
23:28:36brixenif a normal app can stay around 17% GC, we will be smoking MRI I believe
23:28:44brixenahh, that is great
23:29:14slavahow big is a typical heap for you and how long does a full GC take on it?
23:29:29dwaitebrixen: I have a patch to remove the GIL, but it keeps crashing
23:29:47dwaiteit was really like, removing three lines and adding a define
23:29:59dwaite;-)
23:30:02brixenyeah, that sounds like a hot patch :)
23:30:08brixenracy hot
23:30:35brixenslava: here's some gc #s http://gist.github.com/231803
23:30:50brixenslava: we don't have solid numbers on total heap size am
23:30:51brixenatm
23:31:04brixensince immix can get and release chunks
23:31:22brixenthe new stats will emit for each gc epoc, so we'll be able to track it over time