Index

Show enters and exits. Hide enters and exits.

00:44:13rueWheeeen will people stop (suggesting) trying to fix MRI by adding stuff that will just not fit in the arch..
00:45:54yakischlobalike what
00:47:52evanthe GC discussion
00:47:57evanon ruby-core
00:48:10rueMost recently, yeah
02:05:35tarcieriheh
02:05:40tarcierilol @ Matz
02:05:42tarcieri" I am happy to be proven wrong (that means
02:05:42tarcieripatches are welcome)."
02:57:52benblackis pkcs7 in rubinius a non-starter or do i just need to install something extra?
03:04:41rueWhich?
03:04:50rueWhat is pkcs7?
17:27:34brixenmorning
17:28:27evanmorning!
17:44:41wayneeseguinOhio!
17:49:15brixenMichigan!
18:03:08rueHi
18:09:03brixenhey rue
18:12:30brixenwayneeseguin: have you run the benches since evan's immix fix yesterday?
18:12:41brixenwayneeseguin: I'm curious if it affects the memory load any on thase
18:12:45brixener those
18:12:48wayneeseguinbrixen: hrm
18:12:54evanit probably won't
18:13:02wayneeseguinI didn't run benches yesterday
18:13:03evanthat was more of a pathalogical fix
18:13:03wayneeseguinI ran them Wed night
18:13:11evanI am working on another memory condition right now though
18:13:11wayneeseguinhere's the last: http://rvm.beginrescueend.com/benchmarks/2010-01-06/
18:13:15brixenevan: ok
18:13:17wayneeseguinso 155e42c4
18:13:29wayneeseguinWhenever you guys want I'll kickoff another run
18:13:32evani've got 27M allocated for objects
18:13:33wayneeseguinI just ssh in and start it
18:13:35wayneeseguindedicated machine
18:13:38evanbut AM says i've got 95M wired
18:13:43evani'm going to figure out where it is all going.
18:13:49brixenwayneeseguin: ok, cool, whenever
18:14:10wayneeseguinbrixen: so should I wait till evan is done with what he's doing or kickoff now?
18:14:23brixenwayneeseguin: you can wait
18:14:27brixenI was just curious
18:14:41wayneeseguinkk
18:14:52evanyeah, wait on it.
18:34:55boyscoutThrow a proper exception if a thread can't be created - 0bec6e5 - Dirkjan Bussink
18:42:36boyscoutCI: rubinius: 0bec6e5 successful: 3022 files, 11697 examples, 35898 expectations, 0 failures, 0 errors
18:45:38rue"Can we add Rubinius to MRI?" :P
18:46:16Defileralias ruby='rbx'
18:48:10dbussinkrue: i read it yeah, kinda hilarious
18:53:55dbussinkha, people who create thousands of threads now don't get crashes anymore :p
18:54:09dbussinkand yeah, there was actually an issue reported on that
18:54:15evani saw
18:54:19evanis that what the error was
18:54:22evanpthread_create was erroring out
18:54:45dbussinkevan: and i think some gc issue might be going on too
18:54:50dbussinkbut this was definitely part of it
18:55:21dbussinki couldn't reproduce now what that guy reported, so i asked him to check again with the last few fixes
18:55:35evank
18:58:01dbussinkevan: any luck on the vmmethod thing?
18:58:07evanworking on it now.
19:01:04evanI wish the C++ STL mantained a method exist that would report the amount of memory a container was using
19:04:30scoopruse a custom allocator that does statistics? ;)
19:06:32rueOh dear
19:06:46rue"Don't know who Jimi Hendrix was? Click here!"
19:06:58ruescoopr: ^ is.fi :/
19:07:28rueAnd yeah, an allocator would be the hook-in for that
19:07:43evanhm.
19:21:07evanhm
19:21:15evanmalloc trace causes segfaults
19:21:19evanthanks shark.
19:38:58scooprrue, hm? =)
20:13:26mistergibson1Anyone have a suggestion for the following: a non-software-project (groupware) project manager app for rails, etc?
20:19:08brixenmistergibson1: tried Basecamp?
20:19:22brixenI've used it for non-software projects
20:19:35rueAlso #ruby-lang or #ror
20:20:11brixenalso, lunch, bbiab.. :)
20:21:29DefilerI can't remember.. who is the rvm guy?
20:21:41wayneeseguinhe went that way =>
20:21:52evanpoints to wayneeseguin
20:22:04wayneeseguin<= or was it that way
20:22:12evan^^^^ that way!
20:22:19dbussinki'm sure he went ^^ way
20:22:29wayneeseguinlooks around quickly
20:22:45brixenwayneeseguin: run!
20:22:46dbussinkhe might pop up here vvvv
20:22:58wayneeseguinruns away quickly
20:23:04evandbussink: we've got him now!
20:23:10evanthrows a blanket over wayneeseguin
20:23:21wayneeseguinlets out a muffled scream
20:23:58evanpushes the wayne+blanket into the truck
20:24:11dbussinkhits the gas
20:24:21brixenhears sirens, it's CSI LA!
20:24:40evanhehe
20:25:10brixenDefiler: yeah, so um, you looking for the rvm guy? hehe
20:25:38wayneeseguinthinks of a better disguise for next time
20:25:40dbussinkwe have to travel from LA to NY with our truck then i guess
20:25:42dbussinkroad trip!
20:25:59evancan we stop at 7-11 first?
20:26:01evanwe need supplies!
20:26:06wayneeseguinrealizes foomanchu with the mullet wasn't cutting it
20:26:19Defilerwayneeseguin: :)
20:26:32DefilerI was going to ask you a dumb question, but I figured it out from your delectable website
20:26:47evanwell, it's going to be run valgrind day here at the phoenix office.
20:27:03dbussinkvalgrind doesn't compile without hassle on SL :(
20:27:11evani'm running it on linux
20:27:21evanvalgrind is the kind of you think you don't want to be running an experimental version of.
20:27:28dbussinkyeah
20:27:30evans/think/thing/.
20:27:34wayneeseguinDefiler: My 'delectible website' that needs updates badly? ;) ::grin::
20:28:08evanok, leak hunting after lunch I think.
20:28:23evani wonder if this would be easier in an ubuntu vbox...
20:28:41dbussinkcompared to? a remote linux machine?
20:28:45evaneyah
20:28:47evanyeah
20:29:04dbussinki always end up using ssh, i find the integration annoying
20:29:05evanprobably, i can at least use gvim then.
20:29:23dbussinkeasiest is probably some file sharing
20:29:35dbussinklike a shared nfs folder or something
20:29:51dbussinki actually like that if i need windows in parallels, so i can use the same files etc.
20:29:56dbussinkwithout any hassle
20:30:10evani think i'll give my existing ubuntu vbox a shot.
20:30:15evani've fixed other linux bugs in it before
20:30:23evanso it's all setup dev mode wise.
20:30:58evanmaybe i can get kcachegrind to work finally too.
20:31:00evanthat would be cool.
20:31:35mistergibson1sorry, wrong chat window ... hehe
20:32:01mistergibson1... and I'm seeking an open source one - no fees as ... I'm broke :P
20:32:03evanwell, post lunch.
20:32:29dbussinkevan: looks like you can even install a vboxsf for sharing stuff
20:32:40dbussinkso you can use the editor etc. like usual
20:32:59mistergibson1thanks rue
21:05:06Defilerwayneeseguin: I want to send a donation; does Pledgie send you the money immediately, or does it wait until you hit the target?
21:06:01wayneeseguinDefiler: as far as I am aware the pledgie account sends it right away
21:06:02wayneeseguinhttp://pledgie.com/campaigns/7067
21:06:13wayneeseguinDefiler: I received one last night for example
21:06:18wayneeseguinand it's not hitting the target
21:08:13wayneeseguinevan: back from lunch?
21:08:38DefilerGod damn pledgie is buggy
21:08:50yakischlobamabye pledgie needs a pledgie :P
21:08:54dbussinkugh, have a nice socket / thread interaction issue
21:09:08evanwayneeseguin: just got back
21:09:35wayneeseguinevan: ok I need to discuss benchmarks with you for a sec if you can :)
21:09:42evank
21:10:57dbussinkevan: i also might have an interesting issue, seems like a TcpServer#accept isn't interupted when an exception is raised on the thread
21:11:20dbussinkevan: is that a known issue already?
21:12:32evani think i've seen it before.
21:13:18dbussinkhappens during the mongrel tests
21:14:59dbussinkevan: http://github.com/evanphx/rubinius/issues/#issue/159
21:15:24dbussinkevan: is it even possible to abort an FFI call?
21:15:42evannot atm I don't think
21:16:15dbussinksince i assume that's where the accept is waiting now
21:16:38evanprobably, yeah.
21:17:01evanfor accept, which is a blocking but interruptable syscall
21:17:07evanwe should probably go a different route than FFI
21:17:14evanrather than make all FFI calls interruptable.
21:17:34evani suppose the route taken could be a special FFI mode
21:17:41evanthat it tagged per function
21:17:46evanto say if it's interruptable
21:18:17yakischlobaand what? send signal to interrupt it if raise from another thread like in that example?
21:18:19evani think that the only problem is that i don't think FFI releases the GIL
21:20:26wayneeseguinDefiler: THANK YOU!!!!!!!!!!!!!!!!!!!
21:20:28wayneeseguinwow
21:20:31wayneeseguinThank you so much
21:22:10dbussinkevan: do you know how mri does it? just some sort of busy waiting?
21:22:34evanthe same way it does all blocking calls
21:22:58evanit uses it's thread scheduler to way until the socket is ready
21:23:01evanthen calls accept()
21:23:37evanwe can easily implement it the same as how we do read as a prim
21:24:19evanwtf
21:24:27evanwhy is the refresh on vbox totally fucked.
21:24:52evanit's not redrawing
21:27:48dbussinkevan: should i look at IO::sysread?
21:27:54evanprobably.
21:28:39evanwtf vbox!
21:28:45evanthis worked fine not long ago.
21:28:49evannow the network doesn't work
21:28:56evanand the drawing is fucked.
21:29:17dbussinkevan: how does the WaitingForSignal work?
21:29:27ruePrecariously
21:29:38evani'm trying to get this other stuff working
21:29:43dbussinkok, np
21:29:50evantrack it around
21:29:51evanyou'll see.
21:29:51Defilerwayneeseguin: No problem. Thanks for all the work you've done on rvm + rbx
21:30:18wayneeseguinDefiler: my sadistic pleaure! :)
21:30:24wayneeseguin*pleasure
21:31:09yakischlobaevan: it would be nice if there was such a mode as you suggested to make ffi calls interruptible from other threads, if someday permitted by gil stuff
21:33:02evanwtf
21:33:05evanvbox is busted.
21:33:06evan:?
21:33:07evan:/
21:33:57brixenevan: what version are you running?
21:34:06evanI just upgraded to the latest
21:34:37brixenhm, I had refresh issues after the upgrade last time
21:34:55brixenI think I reinstalled the display stuff
21:35:11brixenor upgraded ubuntu, can't remember which actually heh
21:35:30evanwell
21:35:33brixenbut it would not redraw the screen
21:35:38evani was going to make a new ubuntu image
21:35:51evanand the redraw problem happened on the console mode of the installer
21:35:58evanso it's nothing to do with ubuntu.
21:36:22brixenhrm odd
21:37:42evanhttp://www.virtualbox.org/ticket/4653
21:37:50evanat the bottom
21:37:58evanmy exact setup is reported as not redrawing
21:39:49yakischlobawayneeseguin: how do I install rbx head with llvm again?..
21:40:00wayneeseguinyakischloba: rvm update --head
21:40:05wayneeseguinthen
21:40:15wayneeseguinrvm remove rbx-head ; rvm install rbx-head
21:40:17wayneeseguinenjoy :)
21:40:42yakischlobaso the default is llvm now?
21:47:30evanyes.
21:48:13yakischlobacool.
22:01:10wayneeseguinllvm is the new llvm!
22:01:22wayneeseguinthinks that was funny on a meta level ::grin::
22:01:49brixenheh
22:02:50evani'll say this though, vmware fusion has much more polished UI than vbox.
22:03:25evanwayneeseguin: I meant to reply-to-all.
22:03:39wayneeseguinevan: so did I
22:03:44wayneeseguinwhich is what I failed at, twice :)
22:03:46evanso why did you tell me not to?
22:03:54evanoh
22:03:55evannm
22:03:56wayneeseguinI didn't?
22:03:58evani gotcha.
22:03:59wayneeseguinlol
22:04:37wayneeseguinHeading home for family time, bbl.
22:13:18brixenhmm delightful
22:13:30brixenthese require specs for relative paths were totally misleading
22:13:39brixen1.8 does not care what the cwd is
22:13:52brixenit only resolves relative to a $: entry
22:14:08brixen1.9 will resolve a relative path against cwd
22:30:29rue1.8 does not care about much
22:30:59brixenheh, so true
22:59:00evanoh wow.
22:59:15evanunity mode works on vmware fusion in ubuntu
22:59:26evangnome-terminal mingling with the rest of things
22:59:27evanrad.
23:00:06BrianRice-workvery cool
23:06:50ZoxcI'd find compiz unity mode on virtualbox more impressive if it didn't hide the window borders :(
23:07:15evanheh
23:45:37evanUmmmmm
23:45:37evanhttp://gemcutter.org/gems/melbourne
23:47:08evanUMMM
23:47:09evanwtf.
23:47:10evanhttp://github.com/marcoow/melbourne/commits/master/
23:50:25rueGithub forks and pull requests are an AWESOME feature.
23:52:18brixenheh
23:52:22rueAlmost as awesome as releasing a library that is not yours as a Gem
23:52:22brixenum, hmm
23:52:26rueSeriously?
23:55:58ruePolite kinda guy :P
23:56:20evanit's mainly just that we've never heard about it thats a bit wtf.
23:57:24rueWell, yes, it completely breaks all the possible rules, but aside from that...
23:57:57ruePerhaps just someone a bit overenthusiastic.