Index

Show enters and exits. Hide enters and exits.

00:00:20brixenahh, not surprised actually
00:00:51jakedouglasthe test i was looking at actually has to do with whether "constant is not defined directly in the module"
00:01:03jakedouglasneed to check that behavior..
00:04:02jakedouglasgonna go swimming while its still warm, bbl
03:19:37jakedouglassup
04:18:48jakedouglasbrixen: around?
04:39:28boyscoutAdd a spec to assert Module#const_missing is not called during Module#remove_const - 296fbfd - Jake Douglas
04:39:28boyscoutFix Module#remove_const to not call Module#const_missing - e26a196 - Jake Douglas
04:47:49boyscoutCI: rubinius: e26a196 successful: 3500 files, 14323 examples, 42087 expectations, 0 failures, 0 errors
04:53:04halorgiumzzzzomg, crazy complex merb app running-ish on rbx :)
04:53:22jakedouglasworking on this rails one :/
04:54:11halorgiumthis one runs ey-cloud :P
04:54:32jakedouglasthis one runs….other clouds
04:54:41halorgiumcirrus?
04:54:59jakedouglasuh huh
04:55:52halorgiumjakedouglas: how is your experiment going?
04:56:03jakedouglasno. the rightscale app.
04:56:05halorgiumi have found stupid edge cases which are uses
04:56:07jakedouglasuhh
04:56:12halorgiumlolz
04:56:13halorgiumnice
04:56:27halorgiumsay hi to max :P
04:56:35jakedouglasmax R. ?
04:56:36halorgiumyou working there now?
04:56:45jakedouglasyea…have been since december
04:56:50halorgiumnice
04:56:56halorgiumglad you found a place
04:57:02halorgiumtele-commute?
04:57:05jakedouglasuh huh
04:57:31jakedouglasim just trying to get the test suite running…after disabling 4-5 gems, it got as far as loading the migrations and ran into a bug that i just finally fixed
04:57:42jakedouglasnow i upgraded rvm and im trying to reinstall rbx-head but the new rvm fucked everything up
04:57:57halorgiumrvm == the fail
04:58:01halorgiumusing bundler?
04:58:10jakedouglasuh huh
04:58:34halorgiumi've actually had to use bundler 1.0 + BUNDLE_PATH env var + rbx dev install
04:58:36halorgiumno rvm
05:00:20jakedouglashmm
05:00:48jakedouglasi didnt have bundler problems
05:02:14halorgiumjakedouglas: many native extensions?
05:03:13jakedouglasi disabled hpricot, libxml-ruby, ruby-debug, rcov
05:03:43jakedouglasi think we have 5 or so others that built fine anyway
05:04:28halorgiumyer, i hit issues where bundler + rvm would reuse the same shared libs as mri :/
05:05:07jakedouglashmm. i just had to scrap my ~/.rightscale_bundle directory or whatever and it just rebuilt them all
05:05:32jakedouglasim not sure if our bundler setup is really standard though, we have a bunch of scripts that control it and i dont know much about them
05:05:46halorgiumah, i see
05:06:39jakedouglaswell..darn, my rvm is busted now and wont install
05:07:13halorgiumthink there are some bugs with the git part :/
05:07:34halorgiumnot exactly easy to debug either :(
05:08:45jakedouglasthis is just some nonsense about that it cant find mri to build it or something
05:09:32halorgiumoh, didn't rvm use system|1.8.7 ?
05:10:11jakedouglas /Users/jake/.rvm/scripts/utility: line 135: /Users/jake/.rvm/wrappers/ruby-1.8.7-p249/rake: No such file or directory
05:10:25halorgiumusing a diff gemset?
05:11:25jakedouglas /Users/jake/.rvm/scripts/log: line 25: printf: `r': invalid format character
05:11:29jakedouglasi keep getting stuff like that
05:11:38jakedouglasnot sure if its causing the trouble
05:11:58halorgiumlol
05:12:33jakedouglasyea i duno why i upgraded, i always screw myself like that
05:12:53halorgiumhehe
05:14:08halorgiumhrm, not having ruby-debug == tough
05:16:18jakedouglasheh
05:19:17halorgiumw00t
05:19:18halorgiumirb(main):001:0>
05:23:05jakedouglasarg.
05:34:40jakedouglashalorgium: max's mind is blown that we are acquainted :p
05:34:58halorgiumhaha
05:36:59jakedouglasrvm implodes
05:44:36evanhalorgium: i'm working on the debugger now
05:44:51evanhalorgium: how do you typically launch it?
05:44:57evanby putting "debugger" into your code
05:45:24evan?
05:46:23halorgiumevan: yip, exactly
05:46:37halorgiumbtw, got some spec fails on rbx but the code is running
05:46:38halorgiummostly
05:46:40evanyou can do that in rubinius
05:46:46evanwe don't have control flow working yet
05:46:56evanbut you can set explicit breakpoints
05:47:00evanrequire 'debugger'
05:47:02evanDebugger.here
05:47:15evani'll have a bunch more features fleshed out for it tomorrow.
05:47:27evannamely next
05:47:29halorgiumah, variable inspection are stuff?
05:47:35evanyep
05:47:40evaninspection works great
05:47:50evannext/step are the big missing parts
05:47:52halorgiumany docs?
05:48:02halorgiumtries "help" :P
05:48:02evanit's all in one file atm :)
05:48:04evanlib/debugger.rb
05:48:07halorgiumk
05:48:17evantomorrow though, i'm going to do a little refactoring
05:48:21evanand i'll write some docs.
05:49:25jakedouglasnow im trying to reinstall rvm stuff from scratch and rubygems wont download :(
05:49:59evanjakedouglas: ack.
05:50:15evanhello katz.
05:50:17jakedouglasi feel stranded!
05:56:17jakedouglaswhew. working now.
05:57:45jakedouglashalorgium: did you have a lot of spec failures?
05:58:40halorgiumjakedouglas: on rbx?
05:58:46halorgiumor the app?
05:58:59jakedouglasapp
05:59:52halorgiumhaven't tried
05:59:56halorgiumstill ironing out issues
06:00:01jakedouglasoh. you meant you had spec rails of rbx
06:00:05jakedouglaserr
06:00:07jakedouglasspec fails
06:00:09halorgiumyip
06:00:21jakedouglasgotcha
06:00:40jakedouglasim rebuilding rbx now. we'll see how much further i can get..
06:01:32brixenhalorgium: can I see the spec failures? and what platform are you on?
06:01:47halorgiumbrixen: darwin
06:01:55halorgiumbrixen: they are gone :/
06:02:04brixenexcellent
06:02:14brixenL or SL?
06:02:36halorgiumSL
06:02:43brixenk
06:02:53brixenwell, if they reappear, gist it up
06:03:31halorgiumsure
06:07:49halorgiumbrixen: is there a way to see which files are loaded from where?
06:07:56halorgiumlike expand_path on Kernel.requrie
06:10:23brixenhalorgium: #require is implemented by CodeLoader
06:10:46brixenyou could put a puts in the method that loads the file
06:10:55halorgiumbrixen: yer, i know :P
06:11:00halorgiumi had good fun haxing in there
06:11:04halorgiumnothing default?
06:11:09brixenwe used to have a switch to do that
06:11:13brixenjust need to add it again
06:11:17evan--dl doesn't work anymore?
06:11:35evanprobably got unwired during the #require rewrite
06:11:39brixenyeah
06:11:45brixenneeds to be a -X opt anyway
06:12:00brixen-Xloader.paths or something
06:12:21brixenevan: holy shit, just ran fib 40 in 14 sec
06:12:26brixenused to be 17
06:12:30brixenwhat have you been up to
06:12:36evanWEEEE
06:12:39evani'm not telling.
06:12:44evanwhistles
06:12:58brixenmy cpu is at 40% too with firefox running
06:13:48brixenfor those at home, 1.9.2 is 44sec on my system
06:13:56brixenfor fib(40)
06:17:35evani really want a rubinius status window
06:17:38evani should work on that.
06:18:36brixenI want a rubinius status window on my iphone
06:18:45brixenI should work on that
06:19:08evantotally.
06:19:53halorgiumstatus of builds and perf?
06:20:44evanah! nope, but that would be cool too
06:20:53evani'm thinking one that runs in a seperate OS X window
06:21:03evanmonitors any rbx running and gives me starts about it
06:21:10brixenyeah
06:21:11evanfor instance, scrolling by what files it's requiring
06:21:17evanhow busy the jit is
06:21:18evanetc.
06:21:39brixenand on the iphone you could just point it at an ip:port
06:22:17brixenevan: we need Bonjour support in rbx!
06:22:31evanwe can certainly do that
06:22:37evani basically did that with the agent
06:22:39evanvia files.
06:22:43evanthats how console works.
06:22:58evanI considering using bonjour
06:22:58brixenyeah, but for another device
06:23:07evanwell, my thought was you run a forwarder
06:23:13evanthat can talk for all rbx's running on a machine
06:23:30evanso file/agent discovery locally, bonjour on the network
06:23:40evanthat way, they all don't have to be running bonjour
06:23:46brixencool
06:25:15evanok, going to read in bed
06:25:16evannite!
06:26:24brixennite!
06:32:33jakedouglasyay, i got it to successfully not run the specs because i have pending migrations
06:33:14jakedouglaswow. looks like it actually ran the migrations too
06:33:36brixensweet
06:33:52jakedouglasmaybe th test suite will start to run now :p
07:10:02jakedouglasspecs are executing!
07:12:32brixenawesome!
07:13:52jakedouglasreeeeeeeal slow :p
07:13:58brixenhaha
07:14:18brixenwell, that means there's lots of room for improvement
07:15:14jakedouglasfor some reason mysqld is at 30% cpu, i could be wrong but i dont think its usually that bad. maybe something funny going on there
07:15:24brixencould be, yeah
07:16:01brixenwe need some solid profiling data from a real app
07:16:46jakedouglasours does everything the least efficient way possible so you should get lots and lots of data :p
07:16:59jakedouglasnice, i seg faulted it.
07:17:50brixenI so love our backtraces
07:18:01brixencompare: Pegarus::Grammar#S= (method_missing) at lib/pegarus/grammar.rb:30
07:18:12brixenwith: /Users/brian/devel/pegarus/lib/pegarus/grammar.rb:30:in `initialize'
07:18:51jakedouglas:p
07:20:04dbussinkjakedouglas: reliable segfault?
07:20:43jakedouglasi don't know. i'd have to run another 10 minutes of specs to find out. will do later.
07:22:10jakedouglasit looks like machinist or AR or something is hosed, causing most of the failures
07:23:04jakedouglasbed time. night
07:30:32dbussinkbrixen: still there?
07:43:28brixendbussink: sup?
07:45:25dbussinkbrixen: i was running those benchmarks which heavily use packing but looks like it's not using the new stuff yet
07:45:55dbussinkany easy way to see if stuff is already implemented or not? this is using a lot of N stuff, are all variants there added?
07:46:18dbussinkbrixen: i can also show you the steps to run those benches yourself
07:46:39dbussinkit's a lot of pack and also fixnum math, so with your changes it should scream on rbx :)
07:46:46dbussinkalgorithmic stuff etc.
07:50:38brixenshow me the benchmarks
07:51:14dbussinkthe benchmarks are at http://github.com/wvanbergen/chunky_png_benchmarks and they execute benchmarks against http://github.com/wvanbergen/chunky_png
07:51:32dbussinkit benchmarks the png encoding / decoding for that library
07:51:49dbussinkand the chunky_png library uses pack() for a lot of the file writing / reading
07:52:17brixenit doesn't likely use pack for reading ;)
07:52:34dbussinkunpack, sorry :P
07:52:37brixenonly #pack has the integer directives implemented
07:52:48brixenthat's probably one issue
07:52:58brixenit's a little early to run these
07:53:09brixenbut I'll look into them
07:53:27dbussinkthey do represent a real life usecase, that's a positive thing :)
07:54:06brixenyeah, a bunch of this would fail the primitives right now
07:54:12brixenAaZ
07:54:42dbussinkwell, i see N on top of the profile
07:54:43dbussinknet_long
07:54:50dbussinkbut it uses NXNXNX
07:54:55dbussinkor xNXNX
07:54:56brixenwell, that's why
07:55:09brixenanything not implemented causes the primitive to bail
07:55:17dbussinkbecause i did see N being mentioned in the logs, but i guess X isn't supported yet right?
07:55:20brixenyou don't get partial stuff in the primitive
07:55:25brixenright
07:55:35dbussinkthat explains then yeah :)
07:55:39brixenyep
08:05:15dbussinkbrixen: is that on the planned list or not for now?
08:07:39brixenyes, it is
08:15:18halorgiumbrixen: can i override Hash#initialize ?
08:15:54brixenwhy?
08:16:54brixenlike Hash#initialize has no useful purpose, we just added it for kicks? :P
08:19:17halorgiumbrixen: merb inherits from Mash and overrides #initialize
08:19:34brixenmerb can kiss my ass
08:19:36brixen:)
08:20:03brixenHash in MRI is not Hash in rbx
08:20:17brixenrbx is, as much as possible, an OO implementation
08:20:20brixenMRI is not
08:20:28brixenso, caveat emptor
08:20:36brixenoverride away :)
08:21:17halorgiumso i definitely am screwed?
08:21:31brixenwell, merb is stupid for doing that
08:21:36brixenat the least, it should super
08:21:41halorgiumhrm, any idea why things are not working?
08:21:56brixenI don't know what "things" are, sorry
08:23:18halorgiumi think it does super
08:23:41halorgiumhaha, sorry
08:24:30brixenhttp://gist.github.com/476660
08:24:46brixenHash in rbx should be more inheritable than in MRI
08:24:55brixenas long as OO principles are respected
08:25:18brixenwe've had to repeatedly guard Hash methods from terrible Ruby code though
08:25:43halorgiumwow, my irc conn is terrible
08:26:41halorgiumv6 internet ftl
08:27:01brixenwell, sure, because we have a #setup method defined I'm guessing
08:27:15brixenwhich is a pretty important method for us
08:27:25brixenI suppose we need to guard that too
08:27:59halorgiumah, i see
08:28:02brixenhalorgium: go into kernel/common/hash.rb and change def and use of setup to __setup__
08:28:43halorgiumhttps://gist.github.com/6188d75c6fea5a8e43e2
08:28:43halorgiumthat is the code
08:29:05halorgiumk
08:29:06brixenyeah
08:29:44dbussinkString#delete is weird
08:40:38halorgiumbrixen: hrm, interesting
08:41:46halorgiumsegv with debugger :)
08:41:56brixenoops :)
08:42:03halorgiumalso, 5 mins to startup our app :(
08:42:16brixenbut it starts?
08:42:21halorgiumyer
08:42:25brixencool
08:42:26halorgiumwas just getting these issues
08:42:30brixenprofile it
08:42:36halorgiuma reasonable diff compared to master :P
08:43:05brixenI bet the problem is, Merb can't scale
08:43:20halorgiumwe have ~2k files loaded on startup
08:43:28brixenhaha
08:43:30brixenoh lord
08:43:38brixendeliver me from my enemies
08:43:47brixen:)
08:43:52halorgiumyes, good stuff :P
08:44:18brixenhow is that even possible
08:44:27halorgiumwhich?
08:44:32halorgiumthat many files?
08:44:34brixen2k files to load
08:44:35brixenyeah
08:44:47halorgium100 gem deps ...
08:44:58brixenum...
08:45:14halorgiumand vendored code
08:45:17brixenso, yeah, there are some rubygems issues
08:45:30brixenspecifically eval'ing a lot of gemspecs I guess
08:45:35brixenevan has the deets
08:45:45halorgiumyer, i wanted to get this far
08:45:47brixenwe're trying to come up with a solution
08:45:50halorgiumhe said to profile
08:45:53brixenyeah
08:46:18halorgiumshould i commit back the __setup__ thing?
08:46:23brixenprofile, baby, profile
08:46:38brixenhmm, doesn't have the ring of drill, baby, drill
08:46:46brixenbut then, I'm no SP
08:46:57brixenyeah, commit the __setup__
08:47:35brixenI suppose we could make it __internal_rbx_hash_setup_override_and_we_ll_hunt_you_down__
08:47:35halorgiumdo you have branch-based ci?
08:47:48brixenCI runs on master
08:48:07halorgiumwe have it so you push a branch and it builds ;)
08:48:15halorgiummakes our machines not explode!
08:48:36brixenwell, we keep master sane
08:49:24halorgiumhehe, everyone i talk to has the benefit of working on sane software ;)
08:51:19halorgiumoh, no parse tree == no action args!
08:53:02brixenyeah
08:54:14halorgiumw00t, it runs though
08:56:02brixenwow, sweet
08:56:44brixenI gotta crash though
10:32:13wayneeseguinhalorgium: w00t :)
10:32:43halorgiumwayneeseguin: almost w00t
10:45:19wayneeseguin:)
10:45:23wayneeseguinsemi-w00t
16:38:01evanmorning
16:41:01jakedouglasmorning
16:42:59brixenmorning
16:44:37evani'm working on "next" in the debugger
16:44:44evanit's actually kind of a pain
16:44:52evanwell, it's more complicated than I realized.
16:45:31brixenstopping, finding where to stop, other?
16:45:40evanfinding where to stop
16:45:43evanthe stopping is pretty easy now.
16:45:48brixencool
16:45:59evanI was like "oh, just add 1 to the line, figure out the ip, stop there."
16:46:10brixenheh
16:46:12evanwell, when you do "next" on the last line of the then half of a condition...
16:46:16evanO_o
16:46:34brixenyeah
16:47:48BrianRice-workI've thought about this for Slate, but haven't implemented it: we keep a map from bytecode offset to source location in a resident data structure, and more or less would need to increment the IP and run up to the next offset.
16:52:26evanBrianRice-work: i've got a ip => line mapping
16:52:37evanit's more than next doesn't always go to the next line
16:52:45evans/than/tha/
16:52:55evans/than/that/
16:53:11evanand in the presence of a goto_if_true
16:53:19BrianRice-workyeah I don't know if there's an easy "next line" arrangement
16:53:20evanyou have to set 2 breakpoints that are associated with eachother
16:53:32evanunless the debugger hand evaluates conditions
16:56:14brixenseems like the 2 possible is pretty clean way to go
16:56:35khaaseso, there is this guy that doesn't like me
16:56:37brixensince you hit one or the other and hitting either would remove both
16:56:47khaaselet's call him json
16:56:51brixenheh
16:56:51evanbrixen: yep.
16:57:00brixenkhaase: are you trying to install the json gem?
16:57:01khaase"rror: field ‘basic’ has incomplete type"
16:57:04khaasebrixen: yes
16:57:12brixenkhaase: you need to install evan's fork
16:57:12evankhaase: use github.com/evanphx/json
16:57:20evanthe current gem's build is broken
16:57:24evanand the json maintainer is missing.
16:57:54khaaseah, ok, cool, thanks
16:58:07brixennot that the topic is probably read much
16:58:28brixenbut, it's like 3 days, 3 json install compliants I think :)
16:58:31evan:D
16:58:42evani'm hoping my twitter plee works
16:58:46brixenme too
17:47:28jakedouglasdealing with some bullshit this morning, then i will try to run my specs again :p
17:47:52evanbullshit be gone!
17:48:35jakedouglasagreed
17:49:53halorgiumevan: you'll be happy to know UI of awsm is running
17:50:00evanWOO
17:50:03evan^5
17:50:10halorgiumissues with resque workers
17:53:55halorgiumevan: http://github.com/evanphx/rubinius/issues/issue/410
17:54:02halorgiumneeded that to get things started
17:54:32evanFUCK YOU MERB.
17:54:36evanseriously.
17:54:37evanwtf.
17:54:46halorgiumlol
17:54:56evananyway
17:54:59evanyeah, just change the method name
17:55:03evani'll give you commit privs
17:55:05evanyou can push it.
17:55:06halorgiumoh, there is a branch
17:55:12evanoh, you did.
17:55:12evanok
17:55:22halorgiumi did need it :)
17:55:27halorgiumw00t
17:55:33halorgiumlooks like worker started now
17:55:35evani'll pull that.
17:55:55halorgiumi've been having local testing issues
17:56:55evanok
17:59:00halorgiumdo you know if there is an equiv way to do action-args?
17:59:10evani don't know what action-args is.
17:59:17halorgiumdynamic methods at runtime
17:59:21evanhuh?
17:59:26evanuse more words.
17:59:35halorgiumhttp://technicalpickles.com/posts/meet-merb-action-methods-taking-arguments/
17:59:37dbussinkhalorgium: action-args should support 1.9 with what that provides in Method objects
17:59:39halorgiumi'll let him
17:59:56halorgiumdbussink: 1.9 in rbx?
18:00:10dbussinkhalorgium: we have some stuff like source_location
18:00:30dbussinksome small 1.9 api things that were added because someone wanted to :P
18:00:34evanthere is Method#parameters
18:00:40evanjust needs to be massaged into the format you want.
18:00:41dbussinkevan: ah, yeah, that's the one
18:00:47dbussinki think i might even have added that :S
18:00:48dbussink:P
18:01:09halorgiumah, cool
18:01:45halorgiumi'll see if i can work it
18:02:38halorgiumi don't think we use defaults
18:02:52dbussinkyeah, that's not (yet) available in parameters
18:03:25halorgiumevan: ah, got a fun one last night
18:03:46halorgiumour salesforce integration caused what looked like an infinite loop in the compiler
18:03:50evanthe defaults stuff in that yehuda hack was a major hack.
18:03:57evanwe'll never support it.
18:04:00evannor will 1.9
18:04:02evanas I recall.
18:04:11evanhalorgium: rad!
18:04:13evanrepro?
18:04:20halorgiumi can give you the file
18:04:28halorgiumit is kinda sensitive though
18:04:33halorgiumluckily we work at the same place :)
18:04:44dbussinkevan: hack as in the api was a hack or how it was determined?
18:04:57evandetermined.
18:05:18dbussinkah ok, well, we could add it in a nice way :P
18:05:21evanit handled stuff like numbers and strings only
18:05:23evanas I recall
18:05:51evannot arbitrary expressions
18:06:26halorgiumcan i turn off term colors for rbx?
18:06:35halorgiumfor exceptions
18:06:56evanexport RBX_NOCOLOR=1
18:07:02halorgiumah, k
18:07:10evanor
18:07:18evan-Xrbx.colorize_backtraces=no
18:07:42brixenwhich we need to change to -Xbacktrace.colorize=no
18:08:00evanyep
18:08:26evanbrixen: do it now while you're thinking about it :D
18:08:33brixenheh
18:08:52brixenI need to do the config work so we have -Xhelp output and docs
18:09:27brixenbut since we've had one request for no colors since ryan, I think I'll put this on the later list
18:10:40evan:)
18:10:47halorgiumlol, can't upload to github :(
18:10:48halorgium413 Request Entity Too Large
18:10:53halorgiumperhaps the problem!
18:11:29brixenhalorgium: you should do this on an instance and give us account access
18:11:33brixenhalorgium: is that possible?
18:11:44brixenthat way we can log in and run/debug
18:13:40halorgiumbrixen: if only an instance could build rbx :P
18:14:17brixenhum, you can't install it directly?
18:15:36evanbrixen: see IM.
19:19:29jakedouglasok. starting to run specs again
19:21:37jakedouglasseg fault again. let me see if its a particular spec.
19:23:15jakedouglasyea, looks like i can repro
19:25:56jakedouglashttp://gist.github.com/477398
19:26:08jakedouglasi included 3 different ones. looks like they are a little different
19:26:27jakedouglasthey seem to occur at the same place though
19:26:34jakedouglasdoes that help or do you need more information?
19:26:56dbussinkjakedouglas: you probably should catch it with gdb
19:27:01jakedouglask
19:28:52jakedouglasupdated the gist with gdb bt
19:28:55dbussinkjakedouglas: and ask evan if he can help out ;)
19:29:03dbussinkjakedouglas: are you able to repro with a smaller script?
19:29:53jakedouglashmm. i dont know exactly where in the spec it's exploding.
19:30:46jakedouglasis there anything i can call from gdb to see what the last file/line it was executing was?
19:32:07halorgiumjakedouglas: evan said:
19:32:08halorgiumframe 1
19:32:10halorgiump call_frame->print_backtrace(state)
19:32:20jakedouglasthx
19:32:43jakedouglasNo symbol "call_frame" in current context.
19:33:07dbussinkjakedouglas: for a ruby backtrace you need to up until you reach a point where you do have a callframe
19:33:26jakedouglaslooks like frame 0 worked
19:33:47dbussinkjakedouglas: you can see it in the frame description
19:35:15jakedouglasthe particular line is pretty benign looking. im not really sure where to go from there as far as trying to rebuild a more simple case
19:39:23jakedouglasit looks like this occurs reliably every single time though so im happy to let someone do a remote gdb if they want
19:46:41goyox86brixen: hi bro!, http://flori.github.com/json/ is unmantained now?
20:14:22jakedouglasgot another seg fault. this one looks potentially approachable :p
20:14:27jakedouglas(by me)
20:16:35jakedouglashmm. what were you guys discussing about finalizers recently?
20:18:38evanthat i fixed them.
20:19:05jakedouglasReason: KERN_INVALID_ADDRESS at address: 0x0000000000000018
20:19:05jakedouglasossl_x509stctx_free (ctx=0x0) at ossl_x509store.c:345
20:19:05jakedouglas345 if(ctx->untrusted)
20:19:29evanfun.
20:19:32evan:/
20:21:13jakedouglasevan: do you need more information for the one i posted earlier? would you prefer i filed an issue on GH?
20:21:55evana repro
20:21:58evanwould be good.
20:22:40jakedouglasis there some good strategy for simplifying a repro? its kind of in the middle of a bunch of code and the last line doesnt look fishy at all
20:22:42evanjakedouglas: ok, i see a bug
20:22:55evanMRI's use of the free functions is haphazard
20:23:05evanmmm, maybe a bug..
20:23:25evanjakedouglas: looks like openssl
20:23:33evantry and write a script that does whatever you're doing with openssl standalone
20:24:09evanok, yeah
20:24:20jakedouglasi mean, its getting called from the gc
20:24:20evanseems that MRI only calls the dfree if data is not 0
20:24:31jakedouglasok
20:24:51evanwhere as I always call it.
20:25:08jakedouglasgotcha. is the openssl stuff here just copied from mri more or less?
20:25:19evanyep
20:25:47evanjakedouglas: i'll commit a fix for that
20:25:48evanone sec.
20:25:55jakedouglasthanks
20:28:25jakedouglasany specs yet for the finalizers?
20:28:58evanC finalizers? no.
20:29:19jakedouglasok
20:38:11jakedouglasi get a TypeError in a script/console but the console continues to run. how can I get a backtrace?
20:40:51jarib$! maybe?
20:41:01jakedouglasnegative
20:43:54evanhuh?
20:44:01evangist the output in script/console
20:44:16evanof, rather.
20:44:23jakedouglashttps://gist.github.com/99869395d016f920f322
20:44:36evanyou just put in that ticket
20:44:36boyscoutOnly run ->dfree if ->data isn't NULL - 3a09f66 - Evan Phoenix
20:44:38evanyes?
20:44:43jakedouglasyes
20:44:53evanit's possible the exception has no backtrace
20:45:05evanthere are still some cases where a TypeError is raised with no backtrace.
20:45:45jakedouglashmm. so where would i go next
20:46:05evanhow do you know thats the offending coed?
20:46:09evancode.
20:46:19jakedouglaswell
20:46:38jakedouglasthe TypeError says that's where it is. maybe thats not where the real problem is, though
20:46:45evanrepro it
20:46:48evanoutside of rails
20:46:49evanin a script
20:46:50evanthats step 1
20:47:30jakedouglasok. it would help if i knew what was calling it or what it was being called on
20:48:22jakedouglasill try to use my brain
20:49:43evanyou'll have to
20:50:36dbussinkusing one's brain is highly overrated :P
20:50:45dbussinkbut slam at it like a monkey until it works
20:50:54jakedouglasany idea about narrowing down the first seg fault? any strategies? i dont know what to try
20:51:05evanwhat seg fault?
20:51:08evani don't see an issue for one.
20:51:22jakedouglasheh, the one i posted earlier. ill file it, hold on
20:53:03boyscoutCI: rubinius: 3a09f66 successful: 3500 files, 14323 examples, 42087 expectations, 0 failures, 0 errors
20:54:47evanbrb.
20:59:07jakedouglashttp://github.com/evanphx/rubinius/issues/issue/412
21:30:07evanjakedouglas: you need to include a bunch more details about the segfault
21:30:13evanwhat code was running, etc.
21:30:25evanas is, i'll just close it
21:30:30evanbecause there is no way to test that it's fixed.
21:30:48evanoh, you put stuff at the very bottom of the gisst
21:30:52evanplease move that to the issue and update it.
21:31:13jakedouglaswould you prefer i didnt use gists?
21:31:56jakedouglaswhich piece do you want me to put in the issue?
21:32:28evana gist shouldn't hide anything
21:32:34evanatm, the ruby backtrace and the info about the line are hidden
21:32:38evanput those in the ticket itself
21:32:43evantickets should not contain a link
21:32:45evanthats an invalid ticket
21:32:54evanthey need to contain an description of the problem
21:32:54jakedouglasok
21:33:02jakedouglasill put everything in the ticket?
21:33:17evanleave the gdb backtrace in the issu
21:33:19evaner.
21:33:20evanthe gist
21:33:24evanmove everything else out.
21:33:30evanthats the general rule.
21:33:38jakedouglasyou mean "tickets should not contain only a link" ?
21:34:40evanyes.
21:34:42brixenwell, the ticket should describe the issue, with a gist link for patch, or a bunch of data
21:34:51evansorry
21:34:52dbussinkevan: http://github.com/evanphx/rubinius/issues#issue/374 can be closed right?
21:35:05evani don't think so
21:35:11evani didn't work on it
21:35:15evanis it fixed?
21:35:27dbussinkoh wait, i'm confusing it with another lambda / proc / block thingy i guess
21:37:56evanjakedouglas: also, please edit the issue and change it
21:38:01evanrather than adding comments to it
21:38:08jakedouglasok
21:40:24jakedouglashttp://github.com/evanphx/rubinius/issues#issue/412
21:40:26jakedouglasis that better?
21:40:55dbussinkjakedouglas: could you use <pre> tags?
21:41:01jakedouglasyep
21:41:04dbussinkand it's more about the context of the issue
21:41:29jakedouglasright
21:41:40brixenjakedouglas: I just added pre tags for you :P
21:42:26jakedouglasso my problem is that it's from a private codebase and i don't know how to narrow the issue down to the point of being able to provide a non-private reproduction
21:42:43evanand i've changed your title
21:42:58evanas an example of a more suitable title.
21:43:00jakedouglasthanks
21:43:10evana title of "seg fault" is just asking me to close it without looking :)
21:44:00jakedouglasshrug. it could be "running ActiveRecord code", "running RSpec code", "running Secret code"
21:44:06jakedouglasi guess ill just pick one at random next time :p
21:44:31evanjust help us out and be verbose
21:44:53evanno one ever got in trouble for being overly verbose in a bug ticket.
21:45:02jakedouglask
21:45:16jakedouglascleans up his other ticket now
21:53:45jakedouglashttp://github.com/evanphx/rubinius/issues#issue/411
21:53:47jakedouglasis that better?
21:54:43evanmuch!
21:54:48evangives jakedouglas 3 cookies.
21:54:48jakedouglasgreat
21:54:53jakedouglasnom
22:01:51evanjakedouglas: silly question
22:01:57jakedouglassup
22:02:02evanbut I suspect there is a reason you're not on rails 2.3
22:02:03evan?
22:02:28jakedouglasheh
22:02:49Zoxchow's rubinius memory usage?
22:03:07jakedouglasbecause the world isn't a perfect place and things don't always get to be done the way they should
22:03:14evanjakedouglas: ok, i figured as much.
22:03:30evanZoxc: you've been running it under 1.8.7 up til now?
22:03:31evaner.
22:03:33evanjakedouglas: ^^
22:03:46evanZoxc: bigger than MRI at the start, because we've got a lot more runtime data
22:03:53jakedouglasyea. i think i use mri locally, it runs on ree in production
22:04:00evanbut we should manage that memory better.
22:04:05evank
22:04:42jakedouglasevan: yea, upgrading is just not a business priority to the powers that be
22:05:18evank
22:06:08Zoxcthe only numbers I've seen is MRI with around ~6, rubinius with ~50 and JRuby with ~100 or something
22:06:20evanseems reasonable
22:06:31evanif you're talking about something that has just started up.
22:06:40evanand not loaded any code
22:07:20Zoxcyeah, that's quite alot, especially if you run mulitple processes
22:08:32evanZoxc: Activity Monitor says 26M private for 'rbx -e "sleep"'
22:08:48evanMRI is 9.8M
22:09:06evanactually, rbx is 50M virtual
22:09:24evanZoxc: we haven't done much optimization to get that down.
22:09:29evani'm open to suggestions
22:09:42evanMRI keeps all it's methods as C functions in shared memory
22:09:55evanrbx keeps all 4000 core methods as ruby in the heap
22:09:58evanthats the difference.
22:10:17evani'm not aware of a way around that.
22:10:52evanI could probably make InstructionSequence a bit smaller in memory again
22:11:31Zoxcyou could design it to be copy on write friendly atleast
22:11:37evanno
22:11:38evani can't.
22:11:45evanwe use a copy collector
22:11:55evanand with generations
22:12:05evanwhich is by definition not CoW friendly.
22:12:11brixenevan: how much memory is in the iseqs according to heap dump?
22:12:18Zoxcthat's used for code too?
22:12:19evanlets see...
22:12:23evanZoxc: course
22:12:40evani could probably move the mature mark bits out of the object headers
22:12:57evanso that just the young memory was in copied pages
22:13:17evanbrixen: i need to add some retained size calculation
22:13:21evanperhaps i'll do that in a bit.
22:13:25brixenk
22:13:28evani'm working on "next" in the debugger atm.
22:13:33evangoing well
22:13:45brixencool
22:13:50evanstill not sure what i'm going to do about step though.
22:14:11brixenyou were right about #visit, the method is huge and you get a very long list of BBs
22:14:28brixenI'm trying a little opt for storing the BB successors
22:14:46brixenand converting #visit to iteration, of course
22:14:55jakedouglasmy memory usage for script/console is actually not bad compared to mri.
22:15:00evanbrixen: cool.
22:15:06jakedouglas237mb vs. 288mb
22:15:10evanjakedouglas: nice.
22:15:14evanthats what we hope for
22:15:22evanthat rbx evens itself out
22:15:23brixenmy opinion is: we need hard data on real runtime mem usage before worrying about COW
22:15:32evanbrixen: agreed.
22:16:00brixenfortunatley, rbx dev is not buzzword/buzzconcept driven
22:16:07brixenthank lebron zeus
22:16:40evanonly zeus.
22:16:43evanlebron has defected.
22:16:51brixenheh, poor lebron
22:16:57brixennow he knows
22:17:02brixenyou can never make everyone happy
22:18:03evanthis is true.
22:18:09jakedouglasbrixen: this is similar to the issue i was talking about last night http://github.com/evanphx/rubinius/issues#issue/379
22:18:12evancourse, he didn't have to be a dick about it.
22:18:31evanjakedouglas: i can take a look at that
22:18:34evanthat should be easy.
22:18:34brixenjakedouglas: ah, ok
22:18:42jakedouglask
22:18:51jakedouglasjust a popen problem?
22:19:22evanyeah
22:19:27evanprobably.
22:20:11jakedouglascurious that #read doesn't have a problem but stuff like #stat or #sync= do?
22:21:27evan*shrug*
22:21:29evani'll have to see.
22:21:49evanor you can look
22:21:50evan:)
22:22:28jakedouglasheh. i see at least that read/write methods have are actuated differently than others.
22:23:50jakedouglasoh. i kind of get it.
22:24:58nerdsharkhey guys, i'm getting a file missing error when trying to run a program that uses the chingu gem, can anyone give me any advice on how to fix it? 16:45:19 nerdshark | if you use gosu without ruby-opengl, it's very fast │
22:25:03nerdsharker, sorry
22:25:08nerdsharkhttp://pastebin.com/njQJmHEU here's the paste
22:25:49nerdsharki also made sure that the file existed, which it does
22:26:22brixennerdshark: you have some require loading a glob of files
22:26:31brixenyou should not be loading .rbc files directly
22:26:48brixenI'm guessing RequireAll
22:26:52nerdsharkhm
22:26:57nerdsharklet me take a look
22:27:17brixendon't require things like "#{path}/*"
22:27:35nerdsharkyeah, that's what's going on
22:27:40nerdsharki'll let the module author know :p
22:28:31evanthats called a great way to write brittle software
22:28:41evanA+ brittle software
22:28:46nerdsharkhaha
22:28:51evanand not the good kind of brittle like peanut brittle
22:28:53nerdsharkyeah
22:28:56brixenI think people do not get that #require can only load .rb ruby files
22:28:57evanmmmm peanut brittle...
22:28:58brixenever
22:29:36brixenunless you have a native ext mixed in, pass #require names with .rb on them
22:29:40brixenit likes that
22:30:20nerdsharkyeah, he's about to have it fixed
22:30:40brixenalso, why ever just slurp up a bunch of .rb files
22:30:49brixenit's *so* hard to name the files that should be required
22:32:51jakedouglasevan: i guess i dont understand how to have similar semantics as popen(3)
22:33:05jakedouglaswhich is what mri uses
22:35:57jakedouglaswhat does #stat mean for a pipe object composed of 2 pipes
22:37:07evanhonestly no clue.
22:37:13evani'll have to dig into MRI to see.
22:39:24jakedouglasoh whoops, im mistaken. it looks like they only use popen on some weird platforms
22:51:06jakedouglasin MRI, it looks like #stat just uses the result from the first pipe.
22:54:13evanok, lets have a look...
22:54:20boyscoutA bit more debugger work, 'next' mainly. - f9cfd66 - Evan Phoenix
22:54:49jakedouglasrb_io_stat in file.c
22:55:20evank
22:55:35evanthis is because of our BidirectionalPipe kludge
22:55:41evanI think it's time to fix it.
22:55:58evanputs on his kludge unraveling hat
22:56:02jakedouglaswell…i think MRi is two pipes under the hood anyway
22:56:11evanit is
22:56:16evani'm going to change it so we have that too
22:56:27jakedouglasoh, deeper down? what's wrong with it being in ruby
22:56:28evanIO can reference 2 descriptors
22:56:35evanit breaks pretty often.
22:56:53jakedouglaswhy is that a result of being in ruby
22:57:01evancheck out the implementation.
22:57:11evanof BidirectionalPipe
22:57:13jakedouglaswell i saw the implementation heh
22:57:22evanit's a kludge.
22:57:39evani'd rather do it right.
22:57:51jakedouglasshrug
22:58:22evandoing it right doesn't seem important to you?
22:58:49jakedouglasi don't know why it's more right to have it in C/C++ instead of ruby
22:59:01jakedouglasi thought writing shit in ruby was your big goal heh
22:59:07evanit is
22:59:14evanbut not at the expense of having it work well.
22:59:32evanactually, maybe i can fix BidirectionalPipe
22:59:37evanit's doing something kinda dumb
22:59:43evanit should be treating itself as the read half
22:59:44jakedouglasjust because the current implementation of BidirectionalPipe sucks doesn't mean it can't work just as well in ruby
22:59:49jakedouglasi honestly don't care where it goes
22:59:51evanand redifining the write methods to dispatch to another IO
23:00:18evanby lower, i don't necessarily mean not in ruby.
23:01:07jakedouglasi see
23:01:25evani'm going through all the places that MRI uses the ->f2
23:01:29evanthe write descriptor
23:01:38jakedouglasmmhmm
23:04:07evanhm, yes, i can do this
23:04:11evani've got some new tricks in my bag
23:04:14evansince we wrote this.
23:04:17jakedouglas:p
23:04:28evannamely, Rubinius::Unsafe.set_class
23:04:30evan:D
23:04:46evanI should really name it .transmogrify
23:04:50jakedouglasheh
23:06:08boyscoutCI: rubinius: f9cfd66 successful: 3500 files, 14323 examples, 42087 expectations, 0 failures, 0 errors
23:06:39jakedouglasi dont think i saw you push a fix for the free func when data is 0 issue. do you want me to file a ticket for that one?
23:06:45evani pushedit.
23:06:49jakedouglasoh, ok
23:06:51jakedouglasi missed it.
23:12:29rb2kwird, gem is crashing for me on master
23:12:29slavahi evan
23:12:36evanhi doc!
23:12:49rb2khttps://gist.github.com/5b40ef525627d24acd33
23:13:11evanwhen did you last rebuild?
23:13:15evani were seeing that a couple days ago
23:13:21evanwe were, rather.
23:13:27rb2k5 minutes ago
23:13:34evanare you using rvm?
23:13:46rb2krubinius 1.0.1 (1.8.7 3854f9df 2010-06-03 JI) [x86_64-apple-darwin10.4.0]
23:13:47rb2kyes
23:14:07rb2klooks at "2010-06-03"
23:14:12evan3854f9df is from the 8th
23:14:17evanrvm isn't updating properly
23:14:20rb2kuhm...
23:14:23rb2kyeah, seems like it
23:14:30evanwayneeseguin: poke
23:15:01rb2krvm remove rbx-head && rvm install rbx-head
23:16:52evanrb2k: you're doing that now?
23:18:15rb2kyes
23:18:18evank
23:18:21evanlet me know how it goes.
23:23:15brixenblast, this guy will be speaking at portland state on monday http://alarmingdevelopment.org/
23:23:29brixenI wonder if I'll be able to walk, or sit...
23:24:31brixenI will just have to haul myself there somehow
23:25:30evanwhen is the ride?
23:25:35evanthis weekend?
23:25:55jakedouglasstp?
23:26:18brixenyeah
23:26:24brixenI'm driving up tomorrow
23:26:26jakedouglascool.
23:26:45jakedouglaswhere ya sleepin
23:26:53evanbrixen: how is your car getting back?
23:26:55brixenwhere driving == 3 people in a 14' uhaul because there was a slight screw up on train ticket purchase
23:27:00evanor are you driving with someone
23:27:03evanaah
23:27:09evani was going to ask about the train actually!
23:27:13brixenand it costs a bazillion $ to rent a normal vehicle
23:27:28brixenjakedouglas: ace
23:27:41brixenjakedouglas: fri night, centralia in a tent on sat night :)
23:27:48jakedouglasprobably too many bikes on the train during stp. they only have space for like 10 bikes or something last time i checked
23:28:07brixenwe had tickets with bikes for today :P
23:28:16jakedouglasheh
23:28:22jakedouglasoh yea you can reserve it
23:28:25jakedouglasi remember now
23:28:26brixenyeah
23:28:39jakedouglaswhere is 'ace'? dont know it
23:28:48evanbrixen: ack! why did you not go today? not everyone ready?
23:29:38brixenevan: chandra has a test, quiz, and project due today
23:29:47evanack.
23:29:51slavaemerging langs in 7 days!
23:30:02brixenslava: don't remind me! :-/
23:30:17slavaevan: started preparing your talk yet?
23:30:20brixenjakedouglas: I can't describe places in seattle, gmaps for ace hotel
23:30:22jakedouglasi think i went to sign up for emerging langs and it was long since full or something
23:30:26evanbrixen: i probably forgot to mention it, but we'll be in portland in august
23:30:28evanabby and I
23:30:29evanfor a wedding
23:30:36evanperhaps the 4 of us should get dinner
23:30:42brixenevan: cool! let's
23:30:42evanif you're ready for that.
23:31:05brixenevan: early or late aug?
23:31:09evanlate
23:31:18brixenk, aug can be very hot
23:31:21evan20-22
23:31:22slavabrixen: the factor crew will be in portland next week, we should all get beer
23:31:25jakedouglasbrixen: i see. thats uhm…north belltown
23:31:30evanjakedouglas: yeah
23:31:30brixenslava: yes, definitely
23:31:32evan1st and bell
23:31:40evanabove cyclops bar.
23:31:40brixenslava: multiple beers on multiple nights :)
23:31:44slavahehe
23:31:53slavaO(m*n)
23:31:56brixenheh
23:32:00evanMxN beers
23:32:06slavaevan's beer input is exponential
23:32:20evanhah
23:32:21slavaevan: I added instructions with more than one output to my IR
23:32:24jakedouglasevan: nah its a couple blocks away
23:32:32brixenwhere M is the set of good brew pubs in pdx and N is the set of good beers at each
23:32:35slavaevan: it cleans up FFI calls, where value struct and long long returns are more naturally modelled this way
23:32:35evanonly if you want the contents of my stomach on your shoes
23:33:09evanjakedouglas: yer, yes 1st and wall.
23:33:18jakedouglas:p
23:33:20evanjakedouglas: I used to live at Western and Bell
23:33:33jakedouglasah cool
23:33:34evanslava: nice.
23:34:02slavaevan: also callbacks have an initial instruction that outputs all the inputs to the callback
23:34:20slavaevan: I had this split up into individual "load param" instructions but then other passes kept on inserting stuff in between and clobbering registers, etc
23:35:07slavaevan: how does LLVM deal with multiple ouptuts? are they allowed?
23:35:23evanoutputs in what way?
23:35:33slavacan an IR instruction define more than one virtual register?
23:35:52evanno
23:35:58evanbecause it's tree SSA
23:36:03evanso the instruction is the value
23:36:36evanI don't know if tree SSA is the right term
23:36:44evanbut in LLVM, instructions are values
23:36:49slavaI see
23:41:47slavaI have this very real fear that my laptop will die before oscon
23:41:51slavaits been terribly flakly lately
23:41:57evaneeks
23:41:59evanbackups!
23:42:09slavaall my life's work is on the internet
23:42:28evanMona Slava
23:42:40evanThe Factorine Chapel
23:42:56rb2kevan: the rvm reinstall worked fine
23:43:02evanrb2k: bonus!
23:43:04rb2krunning my project perfectly
23:43:05rb2kwheeee
23:43:24bakkdoorslava, brixen: since you're talking about beer again ;) are you going to be around in portland all week? i'm coming monday (aroung noon) but don't have a general oscon pass, so i'll be around and have lots of time :P
23:43:42slavabakkdoor: I'm in town 19th to 24th
23:43:53bakkdoorslava: ah cool me too :)
23:44:07bakkdoorslava: where exactly are you from?
23:44:10slavaI'll check IRC don't worry so just ping me any time :)
23:44:17bakkdoorok cool :)
23:44:18slavaI'm from the internet
23:44:21bakkdoorhaha :D
23:44:38brixenheh
23:44:46slavaBrianRice will be around too
23:44:48brixenbakkdoor: yah, I'll be around
23:44:57rb2kwow, rbx is doing pretty ok and all of the ?10? gems seem to be working
23:44:57slavalang implementor beer fest
23:45:04bakkdoorbrixen: alright cool :)
23:45:08rb2kme gusta!
23:45:09brixenrb2k: sweet!
23:46:39evanrb2k: weeeee
23:50:20bakkdoorbtw, would it be ok if i ported rubinius' compiler & codegen code to fancy? i could of course use the existing but i'd rather write it in fancy itself for bootstrapping etc :)
23:50:52evanfancy?
23:51:01brixenbakkdoor: sure, you can port it
23:51:14bakkdoormy language i'm working on. i want to run it on rubinius
23:51:18brixenbakkdoor: I'm planning on doing the same for Poison
23:51:24bakkdoorbrixen: ok cool
23:51:31brixenalas, Poison compiling Poison will not be done for oscon
23:51:34rb2kI don't have a language, now I feel bad :(
23:51:35rb2kdamn you!
23:51:54evanbakkdoor: go right ahead
23:51:59evanbakkdoor: do whatever you'd like
23:52:02evanit's all BSD licensed anyway
23:52:04evani can't stop you.
23:52:05evan:D
23:52:36bakkdoorbrixen: yeah.. i won't get fancy running on rubinius by next week as well. but i've managed to get the s-expression output for fancy code in fancy-compatible syntax. all i need now is the codegen :)
23:52:47brixensweet!
23:52:56slavaso will fancy be at next year's emerging langs?
23:53:03bakkdoorevan: hah yeah i know just thought it would be nice to ask ;) but thanks, i'll port it then ;)
23:53:15bakkdoorslava: don't know if they'll invite me again :P
23:53:23slavaor are you talking about it this year?
23:53:28bakkdoorslava: yeah :)
23:53:31slavaI see
23:53:40evanoh awesome!
23:53:44bakkdoorthats why i'm coming..
23:53:48bakkdoor:D
23:53:49evan2 languages on top of rubinius at emerging lang!
23:53:53bakkdooryay :D
23:53:54evani'm like a proud father!
23:53:57evanno no
23:54:00evani'm a proud father!
23:54:07bakkdoorevan: congratulations :P
23:54:11evanholds bakkdoor close to his bossum.
23:54:12slavaevan: just wait, I'm rewriting the jvm in ruby so you'll have 3 languages by next week, because clojure will run on rbx
23:54:27evanbosom
23:54:35evanslava: WEEEEE
23:54:39bakkdoori was thinking about using parrot for fancy but rubinius just seems much closer to fancy semantics etc.
23:54:54bakkdooralso, i like the project and support it in some way
23:55:00slavaspeaking of which, are there any parrot people at the summit?
23:55:10bakkdoorslava: allison randal is there i think
23:55:10brixenslava: yeah
23:55:24brixenslava: well, I'm assuming chromatic will be there
23:55:26bakkdoorslava: she organizes oscon afaik. i think she'll speak at emerginglangs as well
23:55:36slavaI see
23:55:37bakkdoorbut maybe i'm wrong with that
23:55:48evanyou guys should talk to here about parrot using LLVM
23:55:51evansince I can't.
23:55:56slavaI'm worried if I meet chromatic I'll just end up trolling him
23:55:59evani talked with her and Larry about it at the last LLVM dev meeting
23:56:06evanslava: BE NICE.
23:56:09slavabut Alison is cool, she was at the vm summit
23:56:25brixenI think work is underway to use llvm
23:56:37brixenI talked to chromatic about it a bit at osbridge
23:56:58bakkdoorslava: seriously, porting jvm to rubinius? sounds fun :D
23:58:23nerdsharkbakkdoor: what is fancy?
23:58:39evannerdshark: do you hunt down nerds that swim?
23:58:48evanie, are you shark for nerds?
23:58:54nerdsharkfactually, i am a shark that is a nerd
23:58:56nerdsharkactually*
23:58:59evanor are you shark that reads computer programming books.
23:59:02bakkdoornerdshark: a smalltalk-like, ruby & erlang inspired dynamic pure object-oriented scripting language or so :)
23:59:04evanah good!
23:59:04nerdsharkevan: that one
23:59:06evani was worried.
23:59:08nerdsharkhaha
23:59:11evani'm glad it's the latter.
23:59:23brixenhttp://github.com/kazzmir/Pegs
23:59:26brixenwoot
23:59:29bakkdoornerdshark: www.fancy-lang.org (or: http://github.com/bakkdoor/fancy)
23:59:32brixenjon rafkind put it on github
23:59:40nerdsharkbakkdoor: ah, just found it on github
23:59:47bakkdoornerdshark: alright ):
23:59:50bakkdoor:)