Index

Show enters and exits. Hide enters and exits.

07:12:45tarcieriI assume you've all seen this already, but...
07:12:46tarcierihttp://www.sdtimes.com/blog/post/2010/01/30/Facebook-rewrites-PHP-runtime.aspx
07:15:48ZoxcI don't really care about PHP and I haven't seen that
07:16:04tarcieriheh
07:16:22tarcieriI don't really care about PHP either, but I'm curious what the underlying technology is
07:17:12ZoxcI'm going to guess: nothing new
07:18:23postmodernim guessing they added new bugs
07:18:34postmodernNuBug Technology (tm)
07:20:04tarcieriZoxc: yes, and Rubinius is nothing new :P
07:22:10ZoxcA) I care about Ruby, B) I'm betting Rubinius is more advanced than that PHP reimplementation
07:22:55tarcieriwell by "you all" perhaps I should've said "evan"
07:22:56tarcieriheh
10:28:17slavatarcieri: its kind of impossible to make a good php implementation
11:01:15postmodernand facebook is not known for their language designers
13:21:39Kerobug report in order for this? http://pastebin.com/m53057e37 debian unstable; I did git clone && ./configure && rake
13:22:43matthewdKero: Happen to have a 'ulimit -v' set?
13:23:07matthewdOr, I guess, something else that may cause rubinius to run out of available memory?
13:24:12matthewdSounds similar to http://github.com/evanphx/rubinius/issues/issue/178
13:24:43Kerono ulimit, and only 100k of swap in use (after test, but I'd expect that to be more if I ran out of memory)
13:26:49dbussinkKero: is it reproducable for each run?
13:27:01Kerothat does not look similar to me. but then I just tried to install rubinius, I've not done anything with it before.
13:28:10matthewdKero: Well, the similar bit is just that (with a ulimit), I saw the same backtrace (in saw_object)
13:28:28matthewdBut yeah... for me, removing my ulimit made the problem go away... so, may be unrelated
13:29:15Kerodbussink: lemme try. would be nice to know which rake task specifically it happens in.
13:29:29dbussinkKero: the command is given at the top
13:30:22Kerothat was just `rake`
13:31:00matthewdKero: The top line of your paste is the command rake was running
13:31:08dbussinkKero: that builds and subsequently runs all the specs with the command at the top there
13:31:38Keromm, wrong swap usage; was looking a the wrong machine. second run doe not crash
13:34:15Kerothat top line does not run. I used a prefix.
13:35:56Kerowithout swap, no crash. i.e. not reproducable. race condition?
13:36:41slavadbussink: what's the difference between rubyspec in the rubyspec repo, and the rubyspec in the rbx repo?
13:37:52dbussinkslava: rubyspec in the rbx is synced every once in a while
13:38:33dbussinkslava: main reason is that for ci purposes we don't want to have them in sync, since that would create a lot of noise when new specs that are added fail
13:38:58dbussinkslava: rubyspec actually started as part of rbx but it was split off
13:47:28slavaah, gotcha
14:10:38Kerocouple of reruns, no crashes. So the original q remains, do you want a bug report?
15:07:46goyox86Hi all, I'm interested in helping Rubinius i've read the getting started guide and it encourages to look here for directions
15:08:43goyox86So, anyone can give me a clue of where to start?
15:15:15kronos_vanoI rewrite Array#pack. bin/mspec spec/ruby/core/array/pack_spec.rb => 1 file, 303 examples, 657 expectations, 135 failures, 40 errors. Something wrong :)
15:16:05kronos_vanogoyox86, See spec/tags and make test passed :)
15:19:53kronos_vanogoyox86, use common chat pls. I think you need evan. He is maintainer. About your question: I dunno. I just select spec and make patch :) pretty easy
15:21:45Zoxcrewrite the GC
15:22:07kronos_vano)))
15:30:08tarcierislava: haha
15:30:39tarcierifunny how there was never a "PHP doesn't scale" argument
15:30:40tarcieriheh
15:31:24tarcieriI guess there was the guy showing how much electricity Facebook could save if they rewrote their entire PHP codebase in C++
18:54:46evanZoxc: stop telling people to rewrite the GC.
18:54:49evanyou're confusing them.
18:55:01evanfor the record, we do not need the GC rewritten.
18:56:24Zoxcthe point was to hopefully confuse them :)
19:06:51tonio6hoy maties!
19:07:20evanZoxc: :)
19:07:26evantonio6: allo allo.
19:07:29tonio6does Rubinius use a C++ testing framework?
19:07:40tonio6for the VM code
19:08:06evanyeah
19:08:07evancxxtest
19:08:10evanit's pretty meh.
19:08:18evanbut it sucks less that all the others.
19:08:26evanthat I tried.
19:09:21tonio6very well, I'm ill-accustumed with ruby but I'm starting a C++ project and would like to do things right... with testing and documentation
19:09:31tonio6you use oxygen for docs?
19:09:46tonio6*doxygen
19:09:46evandoxygen? we have
19:09:50evanwe don't run it often enough
19:09:54evanin time, we will use it more.
19:10:39evantonio6: i'm honored to be held up as the right way to write a C++ project, thanks!
19:10:47tonio6I have you guys in high-regard, wrt code quality and code good-taste so I decided to ask you
19:11:23tonio6well, I have been keeping up with the project and from what I hear and see you guys strive to do things "The Right Way"
19:12:10evanthanks!
19:12:15scooprI use my own testing framework for c++ stuff, http://github.com/scoopr/specific
19:12:26tonio6scoopr: thanks, will check it out
19:12:43scooprit's very small and simple =)
19:13:04evanfor me
19:13:07tonio6what matters is that you are doing tests
19:13:29evanthe big thing was tests running automatically just by writing them
19:13:42evansurprisingly, thats not a feature other frameworks had.
19:14:20kronos_vanoYAHOO
19:14:23kronos_vanoI did it
19:14:28dbussinkevan: frozen actually found the first issue: http://github.com/evanphx/rubinius/commit/9fb3828d98d9d49da176921ea2b9a27cf80210ed
19:15:00evandbussink: ug.
19:15:09evanyeah, there are a few other methods that mutate the arguments
19:15:15evanthey're really bad.
19:15:18scooprtonio6, with http://github.com/scoopr/autobuild you can sorta automate running the tests :P
19:15:20evanso i'm glad frozen is helping.
19:15:33dbussinkevan: was with running rails, i was able to run a webrick again after that fix
19:15:40dbussinkevan: generating a new rails app fails though
19:15:41evanoh good.
19:15:45evanhm. ok
19:16:03dbussinkevan: http://github.com/evanphx/rubinius/issues/issue/178
19:16:09dbussinkevan: looks like it's ObjectSpace related
19:16:12kronos_vanoevan, http://gist.github.com/291199 now Array#pack almost fast as MRI :)
19:16:24evandbussink: how so?
19:16:25evanack
19:16:26evanguys
19:16:27evansorry
19:16:29evani have to run.
19:16:33dbussinkevan: it's sunday yeah
19:16:40tonio6scoopr: specific looks really nice. I'm an RSpec fan so this seems like a nice and simple approach
19:16:44kronos_vanok
19:16:48dbussinkevan: well, it fails in code that matches stuff on object space, that's how far i got
19:16:48evanlets continue this on a work day!
19:17:00dbussinkevan: hehe, for other people weekend is rbx time ;)
19:17:35evankronos_vano: awesome!
19:17:44evangets a giant gold star ready for kronos_vano
19:17:51kronos_vanocool
19:18:06evandbussink: if you have any more info on that crash
19:18:09evanplease put it in the ticket
19:19:36evanok guys and gals, talk to ya later.
19:21:31dbussinkkronos_vano: hmm, i find it harder to read the code you wrote
19:23:56kronos_vanoJust patch my patch! if so :) I done the only one thing: remove parse by regexp (it is very slowly, because it can creates a lot of arrays)
19:24:30dbussinkkronos_vano: well, maybe the parsing needs to be extracted to a separate method
19:25:26dbussinkkronos_vano: or maybe some comments that explain what things do :)
19:25:48kronos_vanook.
23:04:49kronos_vanoah.... code in pack.rb suxx :(