Index

Show enters and exits. Hide enters and exits.

00:01:46shinta43 enters the room.
00:05:38jtoy enters the room.
00:07:06boyscout1 commit by Eric Hodel
00:07:07boyscout * Update to RubyGems r1761.; 1b17478
00:08:56michalw leaves the room.
00:11:57anteaya_ leaves the room.
00:12:57anteaya_ enters the room.
00:15:09m7d enters the room.
00:18:00m7d leaves the room.
00:22:24rubuildius_ppcEric Hodel: 1b1747876; 2377 files, 7769 examples, 26855 expectations, 0 failures, 0 errors
00:24:56pauldix leaves the room.
00:28:11sodaking enters the room.
00:29:13AndrewO enters the room.
00:34:09elight leaves the room.
00:34:24elight enters the room.
00:34:49lopex leaves the room.
00:39:40rubuildius_ppc leaves the room.
00:40:32shinta43 leaves the room.
00:41:36rubuildius_ppc enters the room.
00:45:58lstoll_ enters the room.
00:50:24chris2_ enters the room.
00:54:31lstoll leaves the room.
00:56:19chris2_ leaves the room.
00:58:28chris2 leaves the room.
01:02:31jbarnette leaves the room.
01:02:45binary42 enters the room.
01:03:20boyscout1 commit by Evan Phoenix
01:03:21boyscout * Fix GC bug and scheduler bug.; 729460e
01:04:07vincentluevan: hi, is the external_ip field in struct rubinius_state no longer used? I see only use is m->c->ip_ptr = &m->s->external_ip.
01:04:33jtoy leaves the room.
01:05:46evanhm
01:05:48evanwhy do you ask?
01:06:15evanlooks like it's no longer in use, yes.
01:06:40vincentlujust trying to understand all the fields in the structs :)
01:06:57evanah, ok.
01:07:16rueI thought someone added those docs already--please jot as you go if not
01:12:32pfarley enters the room.
01:13:01Yurik leaves the room.
01:13:44shame leaves the room.
01:17:10bbyars enters the room.
01:18:16t-mac enters the room.
01:19:03qxjit enters the room.
01:19:31akshay enters the room.
01:19:44akshayhey guys
01:20:16vit______ enters the room.
01:20:41zenspiderhey akshay
01:21:07akshaywhat are you guys upto today ?
01:21:22vit______ leaves the room.
01:21:51vit______ enters the room.
01:21:57zenspiderakshay: I was just fixing warn so that miniunit would pass tests again
01:22:05zenspiderbut I can't push them because of perm issues
01:22:49wmoxam leaves the room.
01:25:42evanzenspider: could ya pastie the perm issues?
01:30:05wycats leaves the room.
01:30:57qxjit leaves the room.
01:32:35zenspiderevan: this sums it up minially I think: http://rafb.net/p/quiqWh26.html
01:32:51zenspiderwhich I think means that when you switched to rsyncy stuff, mine didn't flip?
01:32:55evanzenspider: run
01:33:01evanrake spec:committer
01:33:10evanand it will disappear
01:33:25zenspidertada
01:35:00ezmobius leaves the room.
01:35:13blakewatters enters the room.
01:35:16akshay_ enters the room.
01:35:34zenspiderok... since i did this w/o the rake recipes... I do git rebase master?
01:36:53evanyeah
01:38:19zenspideris it ok if git log says I merged master above my other commits? doesn't that imply rebase didn't quite work?
01:38:26zenspideroh. I'm ON master... doh
01:38:40fbuilesvzenspider: git fetch && git rebase origin should do the trick
01:38:41rueIt does show that you are a lesser man but go ahead
01:38:42evanjust commit it
01:38:43evanthats fine
01:38:49evanyou did a pull into master with new commits
01:38:53evanso it generated a merge commit.
01:38:58evanthats fine
01:39:05wmorganimpure!
01:39:16evani've decided that it's not a golden rule
01:39:20evanjust something to aim for
01:39:23evanif you can.
01:39:34rueYou can do a soft reset into a branch, then fetch && rebase and rebase the branch
01:39:36zenspiderrue: if using a POS like git is the measure of being a man, I'd rather be a girl
01:39:37evanthere are times where I do it.
01:39:45zenspiderfbuilesv: that worked. thanks
01:39:49rueYou could do it on master too but it is juggly
01:39:58tarcieri*boggle*
01:40:29fbuilesvzenspider: out of curiosity, what would you recommend then, svn?
01:40:40tarcieriwould say SCM of choice isn't a measure of a man, but I look at people who still use CVS as being weirdoes
01:41:01rueEnhanced male p4ce
01:41:12tarcieriGNU ARCH 4EVR
01:41:22zenspiderfbuilesv: I use perforce for all my personal stuff. I use whatever I'm forced to use on various projects/clients
01:41:39zenspiderpersonal stuff -> all my projects
01:41:51fbuilesvzenspider: is that free for personal use?
01:42:01zenspiderbasically I just want my tools to stay out of my way so I can get stuff done
01:42:09boyscout7 commits by Ryan Davis
01:42:10boyscout * Ugh. I suck at saving files before commit these days; 4d10a50
01:42:10zenspiderfbuilesv: it is free for open source work, so yeah
01:42:11boyscout * Added Rubinius.warn; d78d4f1
01:42:12boyscout * Fixed output to match MRI. Adding Rubinius.warn for internal calls; 541a32d
01:42:13boyscout * Exception#backtrace wasn't returning result of set_backtrace; 21b7816
01:42:14boyscout * Updated to latest miniunit; 311ed2d
01:42:14fbuilesvcool
01:42:15boyscout ...
01:42:35zenspiderdoes boyscout not report spec/ruby pushes?
01:42:35tarcieriI mean, half of what makes git cool is github
01:42:46zenspiderI'd like to know they went up.
01:42:46tarcieriand I don't know of anything like github for any other scm
01:42:54tarcierithere's already like 40+ people following Reia
01:42:56tarcierithat's crazy
01:42:59fbuilesvzenspider: they did, check #rubyspec
01:43:03tarcieriI just talk about it on IRC and twitter and nothing else
01:43:04zenspiderit looks like it pushed... but I dunno
01:43:09zenspiderah. ok
01:49:03akshay leaves the room.
01:51:15mahargso much of git's complexity comes from checkout-local branches. Of all the things it could borrow from cvs, I really don't understand why that's what linus thought was a good idea
01:51:51mahargdir-per-branch is what perforce and subversion did *right*
01:52:19tarcieriI don't think git really "borrowed" that from CVS
01:52:35mahargcvs is the only other vcs I know of that has it
01:52:38Yurik enters the room.
01:53:03tarcierithat doesn't mean git borrowed it from cvs
01:53:13mahargdoesn't really matter in the end
01:53:15mahargit has it
01:53:17mahargit's obnoxious
01:53:37zenspidermaharg: agreed
01:53:40mahargthe thing I enjoyed most about getting away from cvs years and years ago was getting away from insane branching mechanisms
01:53:41tarcierihow would you prefer it worked yet still retained decentralization?
01:54:10mahargthe other decentralized vcs do just fine. To branch you clone the repo
01:54:20vit_______ enters the room.
01:54:38tarcieriso you want it to copy everything when you branch
01:54:56maharghaving sections of your fs morph semi-randomly from one thing to another is just obnoxious
01:55:27tarcierifinds massive amounts of copying obnoxious, personally
01:55:38tarcieriymmv
01:56:40zenspidertarcieri: works great. uses an incredibly simple mechanism and it makes sense.
01:56:47mahargin hg, perforce, and svn you are only copying the head files. It's not at all that bad. Hell, the fact that it's listing you all the files its copying is probably the main thing making that op slow
01:56:57mahargand the directory is *always* what it says it is
01:57:10zenspiderperforce has a really good whitepaper published on branching schemas and it makes a lot of sense
01:57:32mahargand it drastically reduces the complexity of the user interface
01:57:46mahargall three of those vcs' are far simpler to learn and get way less in your way than git
01:57:53tarcierihaving not worked with them (beyond svn) I can't understand what complexity is being masked
01:58:05tarcierihowever I don't find git offensively complex
01:58:28mahargI expect my vcs to Just Work. The learning curve for git is much much higher than any other I've ever used
01:58:53mahargand it's pretty much all because of local branches
01:59:11maharganyways, I'm gonna digress :P
01:59:45tarcieriI'm not sure how branches will be any less "local" in other distributed VCSes
01:59:53tarcieriit sounds like you just want branches to be local copies
01:59:56zenspiderhttp://www.perforce.com/perforce/bestpractices.pdf
01:59:59pauldix enters the room.
02:00:14zenspiderhugs maharg
02:00:57tarcieriwould like to understand your arguments but it's hard when I don't share your problems :/
02:01:30zenspidertarcieri: you share the problems... you use git. you just don't realize they're problems. :P
02:01:40zenspiderwhat's the avg time for tests to come back on rubuildius these days?
02:01:44tarcieriif I don't realize they're problems, how is it a problem?
02:01:46zenspiderI want to wait for the clear flag
02:01:51zenspiderthen I go eat
02:02:11zenspiderthen I release ZenTest, miniunit, RubyInline, and ParseTree
02:02:49zenspidertarcieri: red headed stepchild
02:03:00tarcieri*boggle*
02:03:01tarcieriok
02:03:02tarcieriheh
02:03:31jtoy enters the room.
02:03:52zenspidermeaning: being beaten every day of your life might make you not realize there is a problem, but that doesn't mean there isn't one
02:04:04tarcieriall I know is git-svn has completely fixed my workflow, and furthermore we have crappy sporadic internet and in that regard doing everything locally helps immensely
02:06:20tarcierifar more concerned with problems that do bite
02:06:30rubuildius_ppcRyan Davis: 4d10a5082; 2377 files, 7769 examples, 26855 expectations, 0 failures, 0 errors
02:06:31rubuildius_ppcEvan Phoenix: 729460ec0; 2377 files, 7769 examples, 26855 expectations, 0 failures, 0 errors
02:10:02benburkert enters the room.
02:10:43vit______ leaves the room.
02:11:13zenspideryay... I can go eat
02:11:27tarcierifood rulez
02:11:47cremeszenspider: rubuildius_ppc is taking around 20 minutes per run
02:11:56zenspiderthen I can release a shitload of software
02:12:01zenspiderbye
02:12:02akshayhey - i`m getting 7 failings specs in core
02:12:06cremeszenspider: sometimes it goes faster or slower... depends on what i'm doing on the machine at the time
02:12:11akshayany idea if these are valid failures ?
02:13:11evanakshay_: could you pastie your failures?
02:13:43cremesevan: will be upgrading rubuildius_ppc box to osx 10.5.3 this week; okay or leave at 10.5.2?
02:13:53evanshould be fine.
02:14:10cremescoolio
02:14:30akshayevan: http://pastie.org/211914
02:14:48Yurik leaves the room.
02:15:30pfarleylooking for some "intro" bugs to chew on. any suggestions?
02:15:51xhanjian leaves the room.
02:16:02evanakshay_: no idea why you'd get those failures.
02:16:05evandid they just show up?
02:16:43akshayevan: not sure, i`m running the tests after a long time
02:16:57evanok
02:17:03akshayevan: this is on osx, could this be a platform issue ?
02:17:05evanyou should probably do a 'rake distclean' first
02:17:06ruivaldo enters the room.
02:17:12evanif you haven't updated in a long time
02:17:15evanto clean out any cruft.
02:17:24akshayevan:thanks man
02:18:16akshayevan: the frozen folder looks new, what is it ?
02:18:26rubuildius_amd64Ryan Davis: 4d10a5082; 2377 files, 7770 examples, 26834 expectations, 1 failure, 1 error; http://rafb.net/p/mXoPw071.html
02:21:35rueakshay_: The specs for CI
02:22:17akshayrue: then all of those specs are supposed to pass ?
02:22:18bitsweat enters the room.
02:22:27rueBroadly, yes
02:23:46ruemaharg: Editing the "same file" has its benefits
02:25:31jtoy leaves the room.
02:26:01jtoy enters the room.
02:27:05xhanjian enters the room.
02:28:40ruivaldo leaves the room.
02:29:32fbuilesvpfarley: http://rubinius.lighthouseapp.com/projects/5089/tickets/601 shouldn't be too hard
02:30:09fbuilesvor http://rubinius.lighthouseapp.com/projects/5089/tickets/561
02:31:06pfarleyfbuilesv: thanks will check out
02:31:24fbuilesvpfarley: np, feel free to ask around
02:31:28rubuildius_amd64Evan Phoenix: 729460ec0; 2377 files, 7770 examples, 26834 expectations, 1 failure, 1 error; http://rafb.net/p/CJvoOj19.html
02:31:29rubuildius_amd64Eric Hodel: 1b1747876; 2377 files, 7770 examples, 26834 expectations, 1 failure, 1 error; http://rafb.net/p/vhBsYu47.html
02:31:30rubuildius_amd64Evan Phoenix: a109f3f22; 2377 files, 7770 examples, 26834 expectations, 1 failure, 1 error; http://rafb.net/p/2CLnSg75.html
02:33:11fbuilesvevan: read a ticket for the yaml situation, did you ever decide on what library to use?
02:38:00VVSiz_ enters the room.
02:38:51tcochran enters the room.
02:39:56evanwe're still using rbyaml
02:40:35evanbut it's got some hairy corners.
02:40:53tarciericould syck work through subtend now?
02:41:06evanperhaps
02:41:11evandrbrain and I were just discussing that.
02:41:23boyscout1 commit by Evan Phoenix
02:41:24boyscout * Fix symbol emitter; 62f01d1
02:41:27dysinger leaves the room.
02:42:45bbyars_ enters the room.
02:42:49fbuilesvshould I hold on the YAML specs then?
02:42:52MenTaLguYhow does subtend manage static storage with extensions loaded in multiple VMs?
02:42:53evanlooks like syck only uses RHASH one place.
02:42:57evanwe could change that one place.
02:43:08evanit uses RARRAY and RSTRING otherwise, which we support now.
02:43:18evanMenTaLguY: manage? :)
02:43:25evanMenTaLguY: it doesn't.
02:43:38tarcieriyeah that was my primary motivation with the Ragel parser
02:43:39evanthe only solution i've come up with to solve that problem involves LLVM.
02:43:58tarcieriget the C extensions out of the picture so I could actually work on multi-VM Mongrel
02:44:06tarcieriwhich became a bigger bitch than I ever anticipated
02:44:45akshay enters the room.
02:44:55evanthe only worry with syck (and I'm looking now) is VALUE escape
02:45:05evanthe same way mongrel stashed a VALUE away
02:46:15tarcieriurgh C
02:46:44tarcieridealing with Ruby crashing due to buggy C extensions elsewhere atm :/
02:47:18tarcierithe idiomatic Erlang approach of running C extensions in separate processes is really starting to sound like a good idea
02:47:39tarcieridealing with a crash in zlib which is probably unrelated to zlib
02:49:55cschneid leaves the room.
02:50:09cschneid enters the room.
02:51:34t-mac leaves the room.
02:54:15atduskgreg enters the room.
02:55:54VVSiz leaves the room.
02:58:29ShayArnett enters the room.
03:00:52pauldix leaves the room.
03:01:08ezmobius enters the room.
03:02:53akshay_ leaves the room.
03:03:13bbyars leaves the room.
03:07:11rubuildius_ppcEvan Phoenix: 62f01d118; 2377 files, 7769 examples, 26855 expectations, 0 failures, 0 errors
03:15:00rubuildius_amd64Evan Phoenix: 62f01d118; 2377 files, 7770 examples, 26834 expectations, 1 failure, 1 error; http://rafb.net/p/PoT6MV59.html
03:16:00djwhittdo other people get clean runs on Gentoo these days?
03:16:30djwhittI can't see how rubuildius_amd64 could be running the wrong specs ...
03:20:45loincloth leaves the room.
03:21:46mernendjwhitt: I'll have a look at those specs
03:22:04merneniconv is one treacherous library
03:24:43vit_______ leaves the room.
03:24:48tcochran leaves the room.
03:24:58akshay leaves the room.
03:25:53mernenfunny how it's supposedly the same GNU iconv library from ubuntu, yet the behavior is different
03:26:14djwhittheh
03:26:32djwhittyou have a Gentoo machine you can test on or you just going by the rubuildius output?
03:28:48mernenI'm on ubuntu... anyway, I've got 3 iconv errors here
03:29:09djwhittah, so not so much better as different ...
03:29:10djwhitthehe
03:29:39edwardam leaves the room.
03:32:39nicksieger leaves the room.
03:45:51jtoy leaves the room.
03:46:00jtoy enters the room.
03:49:35gnufied leaves the room.
03:52:37tcochran enters the room.
03:56:47ezmobius leaves the room.
03:59:01tcochran leaves the room.
03:59:32vit enters the room.
04:02:34dysinger enters the room.
04:04:26wycats enters the room.
04:04:32sambo82 enters the room.
04:06:06wycats_ enters the room.
04:06:24headius enters the room.
04:09:05atduskgreg leaves the room.
04:10:21wycatsevan: w00t :) I ran merb on rbx simply and easily
04:10:22wycatskudos!
04:10:26wycats leaves the room.
04:13:52tarcieriwycats: with which parser? the C one?
04:14:07headius leaves the room.
04:26:15wycatstarcieri: I just installed, ran rbx gem install merb
04:26:21wycats(after making a fake fastthread)
04:26:25wycatsand rbx merb -a webrick
04:26:26wycatstada
04:28:21zenspidertarcieri: only one parser so far
04:28:48zenspiderYAY! I just hit commit 4000 in my repo! http://blog.zenspider.com/
04:28:58zenspideralso, released zentest, miniunit, rubyinline and parsetree
04:29:45djwhittzenspider: I think he was talking about the mongrel parser, but I could be wrong
04:30:45shame enters the room.
04:30:47zenspiderdoh. I'll shut up now. :P
04:32:35lchin enters the room.
04:39:21pth leaves the room.
04:46:41lchin leaves the room.
04:47:31lchin enters the room.
04:54:07blakewatters leaves the room.
04:54:19gnufied enters the room.
04:55:46ShayArnett leaves the room.
05:06:54benburkert leaves the room.
05:09:12benburkert enters the room.
05:11:38edwardam enters the room.
05:11:59wycatsdjwhitt: the C parser
05:11:59wycatsregular gem install mongrel
05:11:59wycatsthat was for tarcieri as well
05:16:07benny leaves the room.
05:17:25hoopy enters the room.
05:25:38bitsweatzenspider: cool, was meaning to ask you about miniunit
05:25:53bitsweatI add setup and teardown callback chains to test/unit
05:26:14benburkert leaves the room.
05:26:22bitsweatany suggestion on how to wrap the equivalent of TestCase#run in miniunit? it's all wrapped up in a big run_test_suites method
05:26:49AndrewO leaves the room.
05:28:05benburkert enters the room.
05:29:12jzj enters the room.
05:32:06benburkert_ enters the room.
05:32:56ezmobius enters the room.
05:36:06mediogre enters the room.
05:37:14gnufied leaves the room.
05:41:44zenspiderbitsweat: hop in #seattle.rb and clarify pls
05:42:16bitsweatk
05:48:00benburkert leaves the room.
05:51:42dysinger leaves the room.
05:51:52loincloth enters the room.
05:56:44edwardam leaves the room.
05:57:06edwardam enters the room.
05:58:40freels leaves the room.
06:03:33wycats enters the room.
06:05:14anteaya_ leaves the room.
06:06:37benburkert_ leaves the room.
06:06:52benburkert enters the room.
06:09:58pfarleycan someone tell me how to run a new spec with mspec and MRI to verify behavior
06:20:11wycats_ leaves the room.
06:21:35zenspiderpfarley: ./bin/mspec run spec/path/blah_spec.rb
06:21:45zenspiderpfarley: ./bin/mspec run -tr spec/path/blah_spec.rb for mri
06:23:47JonathanT enters the room.
06:34:03thehcdreamer enters the room.
06:37:42Fullmoon enters the room.
06:39:08Fullmoon leaves the room.
06:39:12Fullmoon enters the room.
06:39:49Fullmoon leaves the room.
06:39:53loincloth leaves the room.
06:39:54Fullmoon enters the room.
06:40:12Fullmoon leaves the room.
06:40:15Fullmoon enters the room.
06:41:26JonathanT leaves the room.
06:49:18pfarleyzenspider: thanks. posted a failing spec to http://rubinius.lighthouseapp.com/projects/5089/tickets/561 . Not sure if you would want a failing spec as a patch?
06:50:12zenspiderabsolutely
06:50:16zenspiderthanks!
06:50:59pfarleyok. will change to a patch so it's not all gobbed up in the comments.
06:51:10zenspiderpfarley: tho, it'd help if you attached that as a real patch file
06:51:14zenspideryeah
06:52:37wycatsso why isn't Passenger's fork of Ruby a bigger deal?
06:57:12mitchellvriley enters the room.
06:57:19dfg59 enters the room.
06:57:22thehcdreamer leaves the room.
06:58:46bitsweatbecause leading bloggers have yet to rant or rave about it, so nobody knows how to feel yet
06:59:17bitsweatthey're trying to get their work merged and matz has expressed interested, so it may not exist for long
06:59:23bitsweatexpressed interest
07:02:04wycatsexcellent
07:02:12wycatsit looks like a big enough deal to care about it
07:02:20wycatscopy-on-write GC makes a HUGE difference
07:02:24wycatsmakes threadsafety matter
07:02:55bitsweatit was a slowdown at first, though, which is why matz waited
07:03:02bitsweatapparently they're faster than mri now though
07:03:42bitsweatmost ruby processes will never be forked, so cow doesn't matter :)
07:03:48benburkert leaves the room.
07:06:17wycatsbitsweat: it matters for Merb
07:06:23wycatsand Rails, if Rails is ever threadsafe
07:06:51wycatsit would be fucking great to be able to boot up a merb and cavalierly fork a thread for every request
07:07:01tarcieriCOW matters for processes too
07:07:05tarcieriand I guess you're going over that
07:07:20tarcieribut you could load Rails and prefork
07:07:34bitsweatfor MRI green threads, or native threads?
07:07:41wycatspresumably natives
07:07:44tarcierithreads don't matter... I'm talking about fork()
07:07:45bitsweatya
07:07:53bitsweattarcieri: well yeah, that's the point of cow :)
07:08:01bitsweatthat's how mod_rails gets its memory savings
07:08:05tarcieriif you pulled in the AST for your entire program, and forked, you could save a ton of memory
07:08:06tarcierinice
07:08:10bitsweatload base code in parent, fork children
07:09:04pfarleyzenspider: all set. thanks again for the help. please bash up on me a bit if I'm breaking convention on the spec.
07:09:33wycatsbitsweat: is there a reason that doesn't already "just work" ?
07:09:42zenspiderpfarley: that's fine... thanks for the contribution!
07:09:44wycatsis obviously somewhat ignorant on the inner workings
07:10:47zenspiderwycats: I think the main problem is the way the memory mgmt is set up requires copying a lot of pages when you fork because everything is always marked dirty all the time
07:11:00zenspiderwith COW, you don't mark nearly as much stuff, so less pages get copied
07:11:24wycatszenspider: is that an issue with OS fork()s though?
07:11:26zenspiderat least, that was an issue brought up a few years back, that was batted around and then ignored
07:11:53tarcieriNow, if Rubinius::VM.fork could be COW safe
07:11:54tarcieriheh
07:11:58tarcieri(or exist)
07:12:00zenspiderwycats: yeah, as I understand it.... tho I'm sketchy on the details
07:12:04wycatszenspider: hmmmm
07:12:18wycatsdoes a CPU fork() require threadsafety?
07:12:34tarcierino, but there's all other "safety" issues
07:12:35bitsweatit's because objects are marked directly in MRI
07:12:40bitsweatcow happens already
07:12:46bitsweatbut it's useless in child on first mark pass
07:12:50bitsweatbecause that's a write
07:12:59wycatsso presumably all mod_rails does is handle the fork()
07:13:00bitsweatso they keep marks in a separate table
07:13:07wycatsand add cow
07:13:23bitsweatthey don't add cow, they change GC marking to be cow-friendly
07:14:07wycatsbitsweat: I will endeavor to be more precise in the future ;)
07:14:30bitsweatforking kills all threads btw
07:14:34bitsweatexcept the current one
07:14:42bitsweatnot sure about 1.9
07:14:45wycatsbitsweat: hmmm...
07:15:00zenspiderwycats: the cheap and easy optimization is to record all mark/sweep data in a separate array, instead of in the objects
07:15:06bitsweatthe thread calling fork is the only one created in the child
07:15:09wycatszenspider: right
07:15:21wycatszenspider: which presumably slows down other operations
07:15:38zenspiderI think it is mainly just "harder". :/
07:17:19wycatszenspider: it looks like they've already done it and released as OSS
07:17:31wycatsnot sure why it hasn't been merged upstream already :P
07:17:43wycatswhile they're busy backporting 1.9 they can have COW :)
07:17:45zenspider"harder"
07:18:50bitsweatwycats: some background: http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-core/15740
07:19:08bitsweatnot upstream because of performance hit
07:19:28wycatsbitsweat: they claim it's faster now
07:21:25bitsweatyeah, and it can be toggled
07:21:33bitsweatruby --enterprisey or something
07:22:31zenspiderhaha
07:24:53jzjhi, zenspider
07:25:06zenspiderjzj: hello
07:25:56jzjwill u merge the ruby_parser project to Rubinius?
07:26:57wycatszenspider: how do I add an #include to RubyInline?
07:26:58wycatsprolog?
07:27:33Fullmoon leaves the room.
07:28:50tarcieriProlog?
07:29:08tarcieriProlog.
07:29:12wycatsI think there's an inline.prolog
07:29:14wycatsor somesuch
07:29:23wycatsor prefix
07:29:46wycatsbuilder.include, it seems
07:32:48zenspiderfor the C builder? yeah builder.include is fine. you can also do builder.prefix, but that is more meant for struct decls and the like
07:35:06sambo82 leaves the room.
07:35:41wycatszenspider: I don't need to provide VALUE self in the standard C builder, right?
07:36:01jzj leaves the room.
07:36:09tmm1wycats: what're you working on?
07:37:08wycatstmm1: some experiments for an icu binding
07:37:18wycatsthere already is one but I'm trying some stuff that's not already in there
07:37:31tmm1icu?
07:38:06zenspiderwycats: not for builder.c, it adds it
07:38:20zenspiderfor any of the _raw methods, you have to do your own thing
07:38:42wycatsokey :)
07:38:51wycatstmm1: unicode library used by couchdb and a bunch of other important peeps
07:38:59tmm1ah
07:39:01tmm1nice
07:39:07tmm1sweet
07:40:31wycats:)
07:42:24bitsweatsmall thing: rbx -v interprets -v as a file, not an option
07:42:32zenspiderheh
07:42:40zenspiderwe'll fix that
07:42:45zenspider(eventually)
07:42:48bitsweatdoh it's both actually
07:42:49zenspider:P
07:42:52bitsweatspits version also ;)
07:42:56zenspiderhaha
07:43:00zenspidereven better
07:43:07zenspidercan you file a bug pls
07:43:08wycatssweeeet
07:44:17tarcieriboth means it's half working, right?
07:44:23bitsweatsure
07:44:31tarcieriawesome
07:45:10bitsweathttp://rubinius.lighthouseapp.com/projects/5089/tickets/603-vm-option-v-causes-exception
07:45:14bitsweatalready there
07:45:26bitsweatwith tiny patch
07:49:15zenspiderbitsweat: oh... that thing with the rails tests I was bitching about before... it is mostly caused because rails opens classes w/o requiring them first. from my gem dir: `find act*/lib rails*/lib -name \*.rb | xargs grep -l "module Unit" | xargs grep -L "require.*test/unit"`
07:49:44zenspiderbitsweat: http://rafb.net/p/G9GroK12.html
07:50:44ezmobius leaves the room.
07:51:13stepheneb enters the room.
07:51:19bitsweatthose are going away very shortly
07:51:36zenspiderrad
07:52:15zenspiderbitsweat: can you please ensure that the replacements are well behaved citizens? miniunit IS taking over for 1.9
07:52:16bitsweatstarted to push everyone to use ActiveRecord::TestCase and friends
07:52:28zenspiderand I'd like to pare back some of the horrible code I was forced to write for rails testing
07:52:34zenspiderit was way way icky
07:52:52bitsweatyeah, can remove it probably
07:53:15bitsweatTestCase mangling will be gone in 2.2
07:53:21bitsweatit's in 2.1 only for compatibility
07:53:31zenspidercool
07:53:37bitsweatand 2.1 is the first to support 1.9 anyway
07:53:41bitsweat'support'
07:54:20wycats leaves the room.
07:54:23zenspiderwell, if I had my druthers, I'd replace test/unit in 1.8.8. :)
07:55:27bitsweat1.8.7 broke all Rails < 2.1 also, so..
07:57:08tarcieriheh awesome
07:57:27bitsweattoo many incompatible backports
07:57:33bitsweatit breaks tons of 1.8 code
07:57:40tarcieriyeah I run into that crap
07:58:11tarcieriI have a box with Ruby 1.8.4 and want to use sqlite3 on it but the latest sqlite3 requires Ruby 1.8.6 (but doesn't tell you that)
07:58:47olabini enters the room.
07:59:38bitsweatzenspider: ha, you add a Module#deprecate that steps on mine
07:59:53bitsweatin test/unit/assertions
07:59:53qwert666 enters the room.
08:06:49bitsweatmade those singletons on Assertions module since they aren't used elsewhere
08:08:07NoKarma enters the room.
08:18:26jzj enters the room.
08:18:26lchin leaves the room.
08:18:26TheVoic1 enters the room.
08:18:26stepheneb leaves the room.
08:18:26thehcdreamer enters the room.
08:18:26wycats enters the room.
08:18:26mass leaves the room.
08:18:26mass enters the room.
08:18:26dfg59 leaves the room.
08:18:26wycatszenspider: I'm getting a nested function error for: http://pastie.org/private/sn4qtdgoyhhootdiimkrka
08:18:29wycatsfrom the builder.c part
08:18:31wycatsthis is the generated C: http://pastie.org/private/gctwgqapwjtkasybibxhqq
08:18:33wycatsI don't see any nested functions
08:23:31headius enters the room.
08:24:40headius_ enters the room.
08:25:03headius leaves the room.
08:27:10zenspiderwycats: looking
08:27:45wycatszenspider: :)
08:28:14qwert666_ enters the room.
08:28:46dbussink leaves the room.
08:28:58wycatszenspider: I think it had to do with a bad include
08:28:59dbussink enters the room.
08:29:09zenspiderwycats: can you paste me the compiler output running with ruby -d ?
08:29:13zenspiderok
08:29:29TheVoice leaves the room.
08:29:54wycatsicu has a rename file that has a bunch of #defines to rename stuff to _4_0
08:29:56wycatsweird
08:31:35zenspiderwanders off
08:33:02wycatspeace
08:34:38bitsweatMini::Test::TestCase.test_suites and .test_methods order by name
08:34:55bitsweatas long as it's arbitrary, why not just rely on the Hash insertion order?
08:35:32dbussinkmorning :)
08:35:48dbussinklive from amsterdam :)
08:37:13dbussink leaves the room.
08:37:31dbussink enters the room.
08:39:01JonathanT enters the room.
08:39:11bitsweathola
08:40:14dbussinklet's see what headius is going to talk about this time
08:40:20headius:)
08:41:03dbussinkhehe
08:45:29qwert666 leaves the room.
08:50:48dbussinkheadius: sitting with zed here too :P
08:50:52dbussink?
08:51:25tarcierinice
08:51:29tarcieritell zed hi
08:52:02dbussinki mean his talk here
08:52:13tarcieriaww
08:55:22dbussinkfibonacci is the ultimate benchmark!
08:55:42tarcieriso it was on earing I take it
08:55:43tarcieriheh
08:56:11dbussinkyeah
08:58:09mutle enters the room.
08:58:26tarcierithing sounds crazy
08:58:26tarcieriheh
09:02:59jzj leaves the room.
09:03:50jzj enters the room.
09:06:14jzj leaves the room.
09:06:57mutle leaves the room.
09:07:29mutle enters the room.
09:12:06mutle_ enters the room.
09:19:47dbussinktarcieri: he just mentioned your project :)
09:23:03jzj enters the room.
09:23:35mutle leaves the room.
09:25:28lypanovtarcieri is mr reia?
09:25:35dbussinkyeah
09:25:39lypanovcool
09:26:00lypanovwonders if zed knows anything about data locality and grids
09:27:11bitsweatzenspider: disagree with removal of assert_nothing_raised also
09:27:46bitsweatyou can provide an error message with it
09:27:57bitsweatso you're explicit about the failure you're testing
09:28:56bitsweatanyway, got a bunch of Active Support tests passing with rbx and miniunit
09:29:17bitsweathave one that blows up, eating 100% cpu and mega memory churn
09:32:01Yurik enters the room.
09:32:51bitsweatdo miniunit patches go to rubinius or elsewhere?
09:33:44bitsweatI see it's released at rubyforge under bfts project, but that's pretty dead otherwise
09:33:47Yurik leaves the room.
09:36:42lchin enters the room.
09:39:31headius leaves the room.
09:39:34VVSizsomething very bad happened to the rubyspec with recent commits
09:39:44VVSizrubinius specific tests were added
09:40:01VVSizand they cause lots of failures/hangups in other - non-related tests
09:40:06VVSiz(with both MRI and JRuby at least)
09:40:34VVSizzenspider: could you please take a look at 2f9102ea99e4d33ba50f101e251e3a697fb5a9ae in rubyspec repo
09:40:50VVSiz"Rubinius.warn added" -- causes all the problems described above
09:42:05TheVoic1 leaves the room.
09:46:46Fullmoon enters the room.
09:51:30Yurik enters the room.
09:51:47VVSizzenspider: I'll revert the change for now then, so that the mspec ci runs are operational again...
09:53:51bitsweatzenspider: require 'test/unit/testcase' just gives a warning now
09:53:59bitsweatyou have to require 'test/unit' to get the TestCase constant
09:54:03bitsweatbut that runs all tests
09:56:51lstoll leaves the room.
10:01:39lchin leaves the room.
10:02:49bitsweatrbx -rfoo is like -v
10:02:59gnufied enters the room.
10:03:03bitsweatrequires foo then tries to run script -rfoo
10:04:25qwert666_ leaves the room.
10:07:37headius enters the room.
10:09:18imajes leaves the room.
10:12:41edwardam leaves the room.
10:14:04lchin enters the room.
10:15:23lchin leaves the room.
10:15:49lchin enters the room.
10:18:49mediogre leaves the room.
10:19:28michalw enters the room.
10:21:08lchin leaves the room.
10:21:16lchin enters the room.
10:23:59chris2 enters the room.
10:42:53bitsweatEnumerable#group_by in Active Support clashes with Rubinius'
10:43:19bitsweatreturns an ordered hash, which breaks Enumerable::Sort's quicksort
10:47:09bitsweat%w(a bc def).sort_by(&:size) # => []
10:47:43bitsweat%w(a bc def).sort{|x,y|x.size<=>y.size} #=> ["a", "bc", "def"]
10:48:03bitsweatso it's something else :P
10:48:14headius leaves the room.
10:48:32jzj leaves the room.
11:04:01headius enters the room.
11:12:42jtoy leaves the room.
11:16:39jtoy enters the room.
11:16:57bitsweatweird bug
11:17:34jtoy leaves the room.
11:18:48bitsweatbehavior of []= method with ||=
11:23:03bitsweathash[k] ||= v should be #[](k) || #[]=(k,v)
11:26:52jp_tixbitsweat: http://dablog.rubypal.com/2008/3/25/a-short-circuit-edge-case
11:27:18jp_tixif you haven't read it
11:28:06bitsweatrbx is doing the right thing above
11:28:15bitsweatbut ruby does something extra for []=
11:28:22bitsweatreturns the rhs instead of the method result
11:29:12bitsweathe shows the same expansion as above :)
11:29:27bitsweatx || x=y
11:29:31jp_tixrhs?
11:29:36bitsweatright hand side
11:29:49bitsweatrbx returns the result of the []= method
11:30:01bitsweatbut ruby returns the value assigned
11:30:15bitsweatseems a special case for ||= assignment
11:30:42jp_tixisn't the result of []= the value assigned?
11:30:52bitsweatnot necessarily
11:30:57bitsweatyou can define your own []=
11:31:04bitsweatthat's the bug I hit above ^^^
11:31:16bitsweatthis group_by implementation returns an OrderedHash
11:31:22bitsweatwhich is based on Array#assoc lookups
11:31:24jp_tixah, i see
11:31:38bitsweat[]= adds [key,value] pairs to the array
11:31:45bitsweatand returns the pair
11:31:54bitsweatbut ruby returns just the value
11:32:16bitsweatso this one thing breaks all enumerable sorts, because quicksort relies on it :)
11:41:01bitsweatrubinius mimicks ruby for the h[k] = v case though
11:41:12bitsweatreturns v, not #[]= result
11:41:29headius leaves the room.
11:49:18bitsweatzenspider: don't see the need for @@test_suites hash either, since it's set by self.inherited only so you'll have unique classes
11:49:30bitsweatcan just use an array
11:59:19mitchellvriley enters the room.
12:12:43BlackEdder enters the room.
12:14:34bitsweat leaves the room.
12:18:37w1rele55 enters the room.
12:19:09blakewatters enters the room.
12:22:43dctanner enters the room.
12:23:00anteaya enters the room.
12:31:29ch0wda enters the room.
12:31:35lchin leaves the room.
12:36:00blakewatters leaves the room.
12:48:06stepheneb enters the room.
12:49:28acfoeken enters the room.
12:55:15Fullmoon leaves the room.
12:55:15chris2 leaves the room.
12:56:37stepheneb leaves the room.
12:58:42qwert666 enters the room.
13:04:36Fullmoon enters the room.
13:05:08yugui enters the room.
13:05:16pfarley leaves the room.
13:08:04blakewatters enters the room.
13:21:39acfoeken leaves the room.
13:22:48blakewatters leaves the room.
13:23:04blakewatters enters the room.
13:27:45blakewatters_ enters the room.
13:35:00mutle leaves the room.
13:40:03blakewatters leaves the room.
13:47:13blakewatters_ leaves the room.
13:48:26mutle enters the room.
13:59:22pauldix enters the room.
14:04:26fabiokung enters the room.
14:04:32anteaya leaves the room.
14:09:36stepheneb enters the room.
14:21:33blakewatters enters the room.
14:22:29anteaya enters the room.
14:37:45moofbong enters the room.
14:46:52pauldix leaves the room.
14:55:05wmoxam enters the room.
14:55:20ShayArnett enters the room.
14:55:44pauldix enters the room.
14:55:53marcoow enters the room.
14:59:36stepheneb leaves the room.
15:04:50AndrewO enters the room.
15:08:45marcoow leaves the room.
15:10:12blakewatters leaves the room.
15:12:03pauldix leaves the room.
15:12:15pauldix enters the room.
15:37:03yasuhito enters the room.
15:46:44ch0wda leaves the room.
15:55:43naeu enters the room.
15:59:05fbuilesvThis looks fine for me, anyone else wants to +1 it to apply? http://rubinius.lighthouseapp.com/projects/5089/tickets/567
16:03:17anteaya leaves the room.
16:03:17fbuilesv leaves the room.
16:05:23anteaya enters the room.
16:05:23fbuilesv enters the room.
16:05:54gnufied leaves the room.
16:12:48krsh enters the room.
16:14:43dgtizedfbuilesv: the only thing is ri doesn't seem to report anything on BasickSocket for recv or send
16:17:31blakewatters enters the room.
16:20:12stepheneb enters the room.
16:20:32fbuilesvdgtized: yeah, that's why the specs use the TCPSocket, the clases that < BasicSocket use it
16:22:38ch0wda enters the room.
16:25:34bitsweat enters the room.
16:28:39headius enters the room.
16:31:47macournoyer enters the room.
16:32:33tomy enters the room.
16:32:42fabiokung leaves the room.
16:33:19fabiokung enters the room.
16:37:12dysinger enters the room.
16:39:00JonathanT leaves the room.
16:39:25JonathanT enters the room.
16:40:12tomy leaves the room.
16:41:06fabiokung leaves the room.
16:41:10fabiokung enters the room.
16:46:21NoKarma leaves the room.
16:54:07yasuhito leaves the room.
16:59:26olabini leaves the room.
17:00:25boyscout1 commit by Christopher Thompson
17:00:26boyscout * Clean out lib/rbconfig.rb.; b6b7f70
17:08:02headius leaves the room.
17:14:16fabiokung leaves the room.
17:14:27fabiokung enters the room.
17:14:49yugui leaves the room.
17:15:38mernenevan: you there? I'd like commit access
17:15:53evando you have a patch committed?
17:16:00mernenyeah
17:16:02mernenhold on
17:16:12binary42 leaves the room.
17:16:29Defilerdgtized: The RI output for the Socket classes is totally untrustworthy
17:16:38Defilerdgtized: Because almost everything comes down from IO
17:16:48binary42 enters the room.
17:16:54mernenhere: http://github.com/evanphx/rubinius/commit/31a3a966a681e46a4918a12641b9a2bb4b59cb0b
17:18:34evansure
17:18:38evani just need your ssh key.
17:20:29enebo enters the room.
17:22:01wvdschel enters the room.
17:22:54evanmernen: ok, added!
17:23:01mernenthanks!
17:23:25boyscout1 commit by Wilson Bilkovich
17:23:25boyscout * Don't exit with an error when 'rake clean' is run in a clean checkout; 8a7d4ad
17:23:42mutle leaves the room.
17:24:59w1rele55 leaves the room.
17:26:02seydar enters the room.
17:26:31fbuilesvDefiler: you have more experience in these things, that sockets patch looks fine for you?
17:26:41DefilerOh, point me at it?
17:26:53Defileroh, I see it
17:27:26fbuilesvtrying to get that and and that old popen patch fixed to apply
17:28:15DefilerHrm. It seems odd to need to re-"attach_function" connect, etc, when they are already done in Socket::Foreign
17:28:27DefilerPresumably we could just move them up to BasicSocket and use them there in both cases?
17:31:10fbuilesvall of them?
17:31:20DefilerThe ones they both call, I should think
17:32:27fbuilesvI see getsockpot, setsockopt, recv and send
17:32:59DefilerWe could either move the whole foreign module up so that they can both reach it..
17:33:08Defiler..or we could just move those up to BasicSocket::Foreign
17:33:16Defilerbut I think we should avoid repeating them
17:35:40fbuilesvactually, where's it reattaching? I only see them being defined in Socket::Foreign and then calling
17:36:19DefilerOh, I guess I misread the patch
17:36:31DefilerI assumed we already had send and recv.. how did we get along without those?
17:37:16fbuilesvthat's a good question
17:38:53fbuilesvwith no specs it seems?
17:40:38naeu leaves the room.
17:41:42anteaya leaves the room.
17:41:42seydarIs it wrong to make use of metaclasses and neato ruby tricks when writing a patch/replacement file?
17:44:43rubuildius_ppcWilson Bilkovich: 8a7d4ad02; 2377 files, 7769 examples, 26855 expectations, 0 failures, 0 errors
17:44:44rubuildius_ppcChristopher Thompson: b6b7f70e1; 2377 files, 7769 examples, 26855 expectations, 0 failures, 0 errors
17:45:43seydarok well it doesn't matter anymore because it doesn't work. now i have to do it the long way
17:46:31dalee leaves the room.
17:46:38dalee enters the room.
17:47:04seydar leaves the room.
17:47:18thehcdreamer leaves the room.
17:48:55DefilerMan, we need a noteserv on this crapular network
17:51:18manverumemoserv?
17:52:14nike234 enters the room.
17:52:19fbuilesvget a rbot running? :P
17:52:24nike234can someone explain to me how rubinius and rails can be integrated
17:52:42nike234and what the advantages are
17:52:53nike234well i sort of understand that last part
17:53:06nike234less ram right?
17:53:11djwhittnike234: Rubinius is an implementation of Ruby, so it should be able to run Rails
17:53:22nike234thread-safe ruby?
17:53:47Defilercheck out the README file and let me know what it fails to explain or point to
17:53:53DefilerThis is a good opportunity to test it out. Heh
17:54:50rubuildius_amd64Wilson Bilkovich: 8a7d4ad02; 2377 files, 7770 examples, 26834 expectations, 1 failure, 1 error; http://rafb.net/p/umZI2R52.html
17:54:51rubuildius_amd64Christopher Thompson: b6b7f70e1; 2377 files, 7770 examples, 26834 expectations, 1 failure, 1 error; http://rafb.net/p/X99bDP62.html
17:55:22Defilerhttp://git.rubini.us/?p=code;a=blob_plain;f=README;hb=HEAD
17:55:29nike234so i install rubinius then rails and mongrel and it should just run?
17:55:43DefilerYes
17:55:47bitsweatnike234: give it a shot
17:56:10bitsweatrails still has some issues with rubinius but they're being fixed up quickly
17:56:17nike234whats the advantages if ur only running a couple of mongrels?
17:56:27nike234is there much of a difference in terms of speed or ram
17:56:39djwhittnike234: hehe, yes right now it will be slower and consume more ram
17:56:49bitsweatnot a production stack yet, nike234
17:56:52nike234i see so its a high end solution right now?
17:56:56bitsweatlol
17:57:20nike234so its not production ready
17:57:20Misha leaves the room.
17:57:29djwhittnike234: correct. not production ready yet
17:57:31ch0wda leaves the room.
17:57:34nexcastellannike234, not really production ready at the moment, but hopefully eventually.
17:57:50djwhittnike234: it just recently got to the point where it could actually run Rails
17:58:18bitsweatturns out that's an over-broad statement :)
17:58:30bitsweatit can serve a request
17:58:30nike234so the stats here are not entirely true ;-) http://www.slideshare.net/evanphx/rubinius-improving-the-rails-ecosystem/
17:59:50nexcastellanAre stats ever *entirely* true? :)
17:59:50nike234isnt ruby 2.0 going to be more threadsafe
18:00:17bitsweat...
18:00:26bitsweathow do you be more threadsafe
18:00:41bitsweatit's not like pouring more chocolate syrup on your ice cream sundae
18:00:45nike234lol
18:00:48nike234u know what i mean
18:00:58bitsweatI don't actually
18:01:13nike234well isnt it going to support more concurrent processing
18:01:14djwhittnike234: the main problem at the moment is that the frameworks aren't thread safe
18:01:25Arjen_ enters the room.
18:01:34bitsweatalso a persistent myth
18:01:41bitsweatthread safety has become a throwaway term
18:02:04djwhittwell, there's a huge lock in Rails once it starts handling a request
18:02:14bitsweathence it's thread safe
18:02:23djwhittok, yeah, I see your point there
18:02:25nike234i had a look at merb recently
18:02:30bitsweatcan you process requests concurrently using green threads? no
18:02:41nike234sucks a bit
18:03:08djwhittyeah, but at least you can do other things while waiting for IO
18:03:11bitsweattroll much?
18:03:21djwhittmerb sucks a bit?
18:03:37bitsweatyou should know it's only socially acceptable in Ruby community to bash Rails ;)
18:03:44djwhitthehe
18:03:45nike234yeah
18:03:47wycatsnike234: ??
18:03:50nike234the stripped away the magic
18:04:04djwhittthat was kind of the point
18:04:05wycatsoh noes... the magic is gone
18:04:08wycats:(
18:04:10nike234thats why rubinius appeal to me
18:04:17wycatsbecause it's more magical?
18:04:19nike234u keep the magic of rails
18:04:27nike234with the concurrency
18:04:33nike234is that going to be possible
18:04:47djwhittheh, you're talking in pretty broad terms
18:04:48nike234or is rails going to still suck for that
18:04:55nike234with rubinius
18:05:15nike234the stuff ive read doesnt say much about rubinius making rails run faster
18:05:31nike234well use less ram
18:05:47bitsweatit's faster than a 3 legged dog
18:05:51wycatsnike234: Rubinius is not yet to the point where such claims make sense
18:05:55bitsweatthe numbers don't lie
18:06:07wycatsbitsweat: but slower than...
18:06:08wycats;)
18:06:40djwhitta single Rubinius process will likely consume more RAM than a single MRI process for quite some time... perhaps always
18:07:06shoenike234: rubinius is not primarily about making rails faster. If you need to make rails faster, you need to get a better handle on exactly what "make rails faster" means for your app.
18:07:09nike234ive worked on projects using 30+ mongrels that kept falling over
18:07:12michalw leaves the room.
18:07:14nike234theres definitely a market there
18:07:24djwhittit can make up ground with forked processes
18:07:35djwhittbecause it lets the OS do a lot more copy on write magic
18:08:23bitsweatnike234: if you want some of that stuff now, look at phusion's passenger module for apache
18:08:34djwhittnike234: or JRuby
18:08:46bitsweatya jruby has similar deployment sweetness
18:09:22nike234ive heard good stuff about evented mongrels but never really looked into it
18:09:30djwhittheadius send me a buck everytime I mention JRuby ;)
18:09:37djwhitt(jk of of course)
18:10:04nike234u guys looked at them?
18:10:09bitsweatnike234: evented mongrel doesn't use threads for each request
18:10:33djwhittevented mongrels will give you faster response time at the expense of concurrency
18:10:50bitsweatwhich sucks in MRI anyway
18:10:53JonathanT leaves the room.
18:11:08djwhittMRI can IO wait with the best of them
18:11:18bitsweatyou're pretty much insane to rely on green threads for concurrency
18:11:25bitsweatbecause you just aren't IO waiting much
18:11:33djwhittif you're handling uploads you are
18:11:53wycatsdjwhitt: JRuby
18:11:54wyhainesThere are specific use cases where a regular mongrel with green threads will give you better throughput that evented_mongrel.
18:12:06bitsweatdjwhitt: yup
18:12:14wycatswyhaines: limited though
18:12:32wyhainesBut outside of those (like uploads), the evented_mongrel is a win over threaded.
18:13:36wycatswyhaines: and couldn't you use defers to make that even more an edge-case?
18:13:47wyhaineswycats, yeah, you could.
18:14:12wycats:)
18:14:13wycatsw00t
18:14:21benburkert enters the room.
18:14:23wycatswe added support for that in Merb proper for people using evented
18:14:58wyhaines(*nod*) deferrables are cool.
18:16:07nike234 leaves the room.
18:17:13djwhittaw man, the conversation was just getting good
18:17:14djwhitthehe
18:20:09boyscout1 commit by Daniel Luz
18:20:10boyscout * Perform a full loop on rest arguments for Iconv.iconv.; 27db60c
18:21:56dctanner leaves the room.
18:23:46wvdschel leaves the room.
18:28:25tarcieriStop using Windows
18:29:38DefilerAre we thread-safe yet? :)
18:31:34atduskgreg enters the room.
18:37:06Fullmoon enters the room.
18:37:22rubuildius_ppcDaniel Luz: 27db60cae; 2377 files, 7769 examples, 26855 expectations, 0 failures, 0 errors
18:37:45wycatsthat conversation was clearly the result of a bit too much blow
18:37:47wycatsafaict
18:37:53bitsweatwondering where I should add failing specs for hash[index] op= expression
18:37:53evanmorning
18:38:04bitsweatthere's variables_spec.rb
18:38:14bitsweatbut unclear to me whether that's specking variables or this assignment behavior
18:38:19evanwell
18:38:26evanthats actually a language element
18:38:30evanfor assignment
18:38:35bitsweatright
18:39:14bitsweatit's specked for []= which happen to return value arg
18:39:26evanwhats the failure?
18:39:29evanjust curious.
18:39:31bitsweatok to spec for an object that doesn't in the same place?
18:39:43bitsweat[]= which return something other than the value
18:39:52evanah
18:39:55evanok
18:40:09evanthats a spec on generic []= actually
18:40:12evanit should go in language
18:40:17bitsweatmri always returns value, rbx returns result of #[]=
18:40:21bitsweatok
18:40:26evanright
18:40:34evanbecause it actually doesn't matter which #[]= is called
18:40:39evanMRI always returns the RHS
18:40:56bitsweatyup
18:42:33dfg59 enters the room.
18:42:44gnufied enters the room.
18:48:44rubuildius_amd64Daniel Luz: 27db60cae; 2377 files, 7770 examples, 26834 expectations, 1 failure, 1 error; http://rafb.net/p/kF8Dsi90.html
18:48:52boyscout1 commit by Wilson Bilkovich
18:48:53boyscout * Another README revision; 3776070
18:52:02DefilerWe are now amazingly close on madness like: some_proc.instance_eval { caller }
18:52:07DefilerSpec time. Heh
18:53:21bitsweatwhat's frozen dir for in rubyspec?
18:54:48DefilerThat's the 'known green' snapshot of the ruby specs that we use for our CI process
18:55:01DefilerWe pull in new specs weekly or so
18:55:45NoKarma enters the room.
18:59:06Defilerdrbrain: What is "2 or 3 digit year used: 104" all about?
19:00:50TheVoice enters the room.
19:02:20rubuildius_ppcWilson Bilkovich: 3776070e7; 2377 files, 7769 examples, 26855 expectations, 0 failures, 0 errors
19:03:34Defilerevan: Got a minute to look at some 'caller' fun with me today?
19:03:47evansure
19:04:28Defilerhttp://pastie.org/212403
19:04:48DefilerI am trying to either understand or at least track down why we have such an incorrect line number showing in spec_parser_fixture.rb
19:05:14DefilerI can tell you about what I know is still wrong with our caller/backtrace infrastucture over the phone more easily, I suspect
19:05:55nexcastellanRight now, tainted? will always return false in Rubinius, correct?
19:08:40nexcastellanThat is, we don't do tainting.
19:09:52pfarley enters the room.
19:09:53boyscout1 commit by Daniel Luz
19:09:54boyscout * Check type of $_ before invoking String methods.; 3c49ed2
19:10:56pfarley_ enters the room.
19:10:56pfarley leaves the room.
19:11:31<