Index

Show enters and exits. Hide enters and exits.

00:00:23tarcieriheh, it's... hard to explain
00:00:28evanok.
00:00:34tarcierimainly I was just wondering why the output was screwey
00:00:37tarcierithat explains it
00:00:48tarcieria.sort! { rand(3).to_i - 1 } works
00:01:01evanha! true.
00:01:25tarcieriguess the to_i is unnecessary too
00:01:37tarcierievan: so I was going to ask you about gems for the "rbx" platform
00:01:41tarcieriis that ok?
00:01:46tarcierigems that are rbx-specific
00:01:48evanwe should ask drbrain
00:01:50evanbut I think so
00:01:55evanwe should probably use the name rubinius though
00:01:59tarcieriok
00:01:59evanto be explicit
00:02:19tarcieriI was talking to the mongrel people and they're willing to take a patch to let them generate a gem for Rubinius
00:02:20wycats enters the room.
00:02:32tarcierithat would include the Rubinius.asm parser
00:02:47evandopeness
00:02:57evanlet me ask drbrain
00:03:00evanhe's on a train
00:03:18wycatsevan: Defiler was telling me about using Rubinius.asm in random files, and how it takes locals to use in asm
00:03:22wycatsthat's dope
00:04:47evanwycats: yeah
00:05:01evanRubinius.asm basically lets you run specific code at compile time
00:05:57evanthat code typically drives the Compiler::Generator to create bytecode
00:06:40wycatsyeah
00:06:44wycatsit's pretty sweet
00:06:55evanyeah
00:08:28wycatsevan: any idea why Safari's new JS VM is not using LLVM?
00:08:41wycatsI would think given the amount of attention Apple is giving LLVM it would be central
00:09:29anonuserDifferent teams different preferences I'd assume.
00:11:02tarcieriApple's main interest in LLVM seems to be mixing and matching hardware and software implementations of various 3D graphics features depending on what's available on the 3D chip
00:11:28anonuserThat's what I took from it. They use it to provide shader support to lower end graphics cards.
00:12:29massI'd be a bit worried about doing a sort with rand as an index, because it means that a lot of assumptions don't hold true
00:12:58evantarcieri: drbrain is adding a feature to rubygems now to make this much easier
00:13:07evan(having rubinius specific gems)
00:13:29massin STL you can have segfaults because the data structures get so mucked up
00:13:47evanthats not going to happen to a ruby Array.
00:13:52evanmass: ^^
00:14:40anonuserevan, are you still playing around with LLVM for Rubinius?
00:14:49atduskgr_ enters the room.
00:14:50evananonuser: yep
00:15:01wycatsevan: I have a new favorite line of Railas code:
00:15:02wycatsmethod_scoping = method_scoping.method_scoping if method_scoping.respond_to?(:method_scoping)
00:15:14evanhah
00:15:16anonuserFor a C++ rewrite of the core VM or something?
00:15:39evananonuser: i'm doing it in the C++ VM i've been working on, yes.
00:15:58atduskgreg leaves the room.
00:16:03anonuserNeat-o, I'm just trying to get the story straight.
00:16:09evanyeah, no prob.
00:16:09moofbong leaves the room.
00:16:15evanit's pretty awesome
00:16:19evanfor a very small effort
00:16:21anonuserC++ or LLVM?
00:16:21masswycats: I cannot think of an emoticon to describe how that code makes me feel
00:16:27anonuser<3's both
00:16:28evanbytecode dispatch will be entirely eliminated
00:16:33anonuserbut I'm in a major minority with C++
00:16:47wycatsmass: yeah
00:16:48wycatsagreed
00:16:49evanthe C++ + bytecode + LLVM setup
00:17:02evani'm not minding C++ too much
00:17:31anonuserC++ has a bad rap in the OSS arena :(
00:17:40evani'd agree with that.
00:18:02massmost of that is because of past performance by a certain free software compiler though :)
00:18:05evanit has crazy parts
00:18:07evanbut so does C
00:18:13anonuserYeah.
00:18:16evanand it fixes some of C's crazy parts
00:18:22anonuserAnd adds it's own.
00:18:26evanyep!
00:18:27evan:D
00:18:31wycatsevan: it has a bad rap because people are lazy :P
00:18:36anonuserwycats, partially!
00:18:42massC++ standardizes a lot of things that C should have standarized by now
00:18:44wycatslol
00:18:44evanstuff like function template specialization has made the new VM nice and clean though.
00:18:52anonuserwycats, C++ is crazy complex but very powerful if you can wield it correctly.
00:18:53wycatsif you can read it ;) ;)
00:18:57wycatsanonuser: same with Ruby
00:19:06evanafter all, coming from Ruby, who can resist a little metaprogramming (even if it is at compile time)
00:19:07anonuserwycats, exactly but people love ruby for some reason!
00:19:11massbut C++ as an OO language tries to go too high for not having solved cross-cutting problems for the user
00:19:14anonuserI actually went from C++ to Ruby
00:19:44anonuserevan, partial template specialization is an easy thing to screw up.
00:19:53massbecause of that, library code becomes less readable as it becomes more useful
00:19:55mass:)
00:19:58anonuserI know I got in trouble with PTS going from compiler to compiler
00:20:06tarcierievan: cool (re: rubinius-specific gems)
00:20:17anonusermass, that is the truth, I fear reading complex container classes.
00:20:19evantarcieri: drbrain is on the train, but he should be able to push it soon
00:20:48evantarcieri: i'm pretty sure that all rubygems versions now will say they can't install them
00:20:52massanonuser: to be fair, a lot of that complexity is because the C++ standards committee decided all objects have to have a non-zero size (be referencable)
00:20:59evanand just the updated versions in rubinius will realize that they can
00:21:00evanso thats good.
00:21:06massoh the fun the STL does to optimize that decision away
00:21:32anonusermass, well that's one of my gripes with the C++ committee they don't really make a reference implementation of these things. They just kind of hope it works :(
00:21:39evanpersonally, I don't get why STL is so crazy complex.
00:21:43tarcierievan: so it's just going to be a platform setting, right?
00:21:47evantarcieri: yeah
00:21:52tarcieri"rubinius"?
00:21:52massevan: which part?
00:22:02evantarcieri: we'll tag a gem as platform rubinius
00:22:06tarcieriok, cool
00:22:14evanmass: all the containers, for instance
00:22:18anonuserevan, my take is performance for different cases.
00:22:20evani tried to read the code
00:22:23evanYIKES.
00:22:25anonuserDon't bother it.
00:22:37anonuserSGI wrote a lot of that crap in the early days.
00:22:41massevan, yeah thats what I meant with my 'all classes must have size' comment
00:22:42anonuserI love their variable names
00:22:47anonuser_m, _s, _sp
00:22:53evananonuser: yeah
00:23:06massoh yeah, the variable names are superb as well
00:23:08anonuserThat's why I loved using Visual Studio
00:23:19anonuserit had special cases for container classes.
00:23:23massthe compiler generates slower native code if you use long names, apparently ;-)
00:23:42anonuserso if it was a string or array it'd just show the content in the debugger as you'd use it, not the internals.
00:23:52rubuildius_amd64Brian Ford: 7a25999ab; 2188 files, 7231 examples, 25913 expectations, 1 failure, 0 errors; http://rafb.net/p/BKMMeh47.html
00:23:54anonusermass, :( name mangling happens anyways.
00:24:15massbut no, to keep the allocator from taking up space (except in the cases where you need space for the allocator), they do things like subclass internal types that have all the logic from your allocator
00:24:40massso there is a hashtable and a hashtable internal that will inherit from your less and allocator implementations
00:24:44massmindblowing.
00:24:56anonuserIt's all crazy stuff.
00:25:26evanmass: yeah, thats the kind of stuff i mean
00:25:31massanyway, I recommend "Modern C++ Design" if you want to get headaches :)
00:25:31evanseems overly complicated
00:25:40massI called it the anarchists cookbook of C++
00:25:41anonuserI know C is getting crazy cool things like jemalloc implementations and you can override malloc easily in unix but I've yet to see an arena based allocator for C++
00:25:44evanmass: i've actually decided to buy no C++ books.
00:25:49evanit's better I not know
00:25:55masslol
00:25:57evanand use more ruby-style idioms instead.
00:26:00anonuserevan, good show! I bought a reference book back in the day.
00:26:11anonuserJamas' C/C++ programmers bible..
00:26:14anonuserwhat a shitty book
00:26:16evanha
00:26:22evani have had C++ books over the years
00:26:24evandid it back in college
00:26:28evani have one C++ book
00:26:36anonuserMaybe I can find it.
00:26:36massi don't disagree evan, although effective C++ and more effective C++ might be a good idea to try and skim :)
00:26:37evanthe really old Oreilly C++ Core Language book
00:26:46anonuserTo give an idea the craziness of this book
00:26:52evanmass: ok, i'll skim at the bookstore :)
00:27:02anonusereffective C++ is an excellent book for people wanting to be decent C++ programmers
00:27:44massevan: if it is good enough, crazy people like me will contribute fixes to eliminate 4 bytes here or inline a method there.
00:28:09anonuser5 months later I've still not unpacked everything :(
00:28:25massthe next c++ will be rather nice imho, I just wish there was something I could play with now
00:28:27evanmass: heh
00:28:44anonusermass, yeah it's going to be neat. No more space between the angle and templating spec
00:28:48anonuserThat is going to be so awesome
00:28:51evanhah
00:28:54masslol there are other things I like about it
00:28:55evanonly took them 15 years!
00:28:58masslike the auto keyword :)
00:29:05anonuserFINALLY.
00:29:07massinferred typing ftw
00:29:12anonuserThat is the best thing evar!
00:29:30anonuserI always had to pull out some kind of iterator before a loop because it makes things too unreadable
00:29:30massrvalue references are evil and will take me a while to fully grok, but they fix so many issues its worth it.
00:29:48anonuserI'm hoping threading in the language will be complete.
00:29:57massunique_ptr will be great.
00:30:06massit'll be somewhat complete, if that makes any sense
00:30:09anonuserI hope it'll support multiple paradigms of concurrency and parallelisms
00:30:18evani don't get the reason for the null_ptr type
00:30:23tarcieri*boggle* @ parallelism in C++
00:30:37anonuserevan, there's a lot of history to it.
00:30:40massevan, I can explain null_ptr in one line :D
00:30:50anonuserI hope the garbage collector makes it in there.
00:30:58tarcieriI think the best way to do concurrent C++ is to write a VM that runs a language that solves the problems of parallelism at a higher level of abstraction :)
00:30:59anonusertarcieri, It can be done :)
00:30:59masschar *foo = false; // true here will throw a compilation exception
00:31:45anonuserActually C++ is a great candidate for threads for the sake of clarity.
00:31:46tarcieriOne VM per hardware thread per CPU core seems like a nice model for shared heap languages
00:31:50hemulen leaves the room.
00:31:55tarcierithreads are a lousy concurrency model
00:32:05anonusertarcieri, it depends on the case.
00:32:12tarcieriin the general case
00:32:17massactually, I used to work on threaded C++ apps. abandon all hope ye who enter here.
00:32:24anonusermass, same here.
00:32:30tarcieriused to do multithread C code
00:32:32tarcieriurgh
00:32:40massI'm sure its better now
00:32:46anonusermass, hah, you'd think.
00:33:02masssolaris gcc is still crappy I know
00:33:02anonuserI remember running into a memory starvation problem with threads.
00:33:05evanI still can't believe how bad threads and signals interact
00:33:10anonuserIt was trivially easy to trigger.
00:33:13evanand that no unix seems to have fixed that.
00:33:15tarcierievan: heh, that's exactly what I was going to say
00:33:19anonuserevan, you mean the all threads all the time get all signals?
00:33:25evananonuser: or do they?
00:33:26tarcierievan: signal management in multithreaded programs is a nightmare
00:33:29evanNO ONE KNOWS!
00:33:34massthey shipped 3.0.x with a bad atomicity.h for sparc
00:33:37anonuserhaha
00:33:38tarcierianonuser: you can pthread_sigmask()
00:33:46tarcieriand block certain signals on a thread-by-thread basis
00:33:58anonusertarcieri, yeah I know.
00:34:16massyeah, I think you pretty much have to sigmask everything to one point or else you cannot be sure what will happen
00:34:20tarcieriwhat's really fun is if you have a program that was never designed to be multithreaded that uses sigprocmask() and you try to create some threads inside it and want to block signals
00:34:25massthe signal handling thread :P
00:34:28tarcierilike, say, MRI
00:34:48tarcieriyou're basically screwed
00:34:49anonusermass, I had the signaling thread :(
00:35:03tarcieriyeah, that's what I tended to do... dedicate an entire thread just to signal handling
00:35:16tarcieriblock all signals on all other threads
00:35:29massat the time I was part of a small but growing group that wanted to ban all threading from all our c++ code and go to shared memory and mutex signalling :P
00:35:31yasuhito leaves the room.
00:36:03tarcierishared memory? urgh
00:36:09anonuserI wanted to give intels threading crap a try
00:36:32tarcieripreferred to implement message queues that would copy data into buffers and deliver those to other threads
00:36:44tarcieriand use only those
00:36:52anonuserI actually did something similar.
00:37:00lopex leaves the room.
00:37:09headiushowdy
00:37:17anonuserI wrote a networked framework for messages but I left before it was any decent.
00:37:52evanheadius: allo!
00:38:21tarcierianonuser: yeah I think given enough time people generally converge on some sort of message passing model
00:38:39anonuserYeah that's what I hope makes it into C++0x
00:38:51anonuserwhich is bound to become C++1x
00:39:07tarcierianonuser: sometimes referred to as Armstrong's Addendum to Greenspun's Tenth Rule
00:39:12tarcierior something to that effect
00:39:13tarcieriheh
00:39:19evanthey just just call it C++ 3000
00:39:23tarcieriheh
00:39:25evaner.
00:39:26evanshould just
00:39:28anonuserhahah
00:39:53anonuserDoes anyone know if the garbage collection will make it into the C++ standard.
00:40:20massnot for the next release, but the plan was to make changes to prepare for it
00:40:41headiuseveryone settling back into the daily grind?
00:40:50evanheadius: yep!
00:40:51masssome of the threading stuffs for next release: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2008/n2497.html
00:40:57evanheadius: you?
00:41:28headiusjust arrived home...gonna look at AST sharing across runtimes to reduce memory
00:41:33massgc stuff @ http://tinyurl.com/5ct3tm
00:41:38evanheadius: nice.
00:41:38massthat first URL was too long
00:42:02anonuserI was hoping it'd make it into the language standard :(
00:44:07evancezarsa: i found the bug.
00:44:18evanthere is a bug with a native method calling a primitive
00:44:22evanfixing it now.
00:44:34massshared_ptr is good and unique_ptr will be nice as well. The only time I've really run into a case where I would want to do full on GC management in C was when managing an internal cache
00:44:50anonuserNeat-o there's a STL thread safety proposal. Not sure why but okay
00:45:10massand even then I wanted to expire things least-recently-used, so I wound up having to just implement my own refcounting/ordering scheme
00:45:25anonusermass, if it's generation it can help with object instantiation overhead.
00:46:01massanonuser: the proposal may be that they have to be safe when threads are in use. Today, I don't know if STL guarantees two different maps on two different threads will work
00:46:15masseven if the threads never access one another's maps
00:46:20anonusermass, ah that makes sense.
00:46:55cezarsaevan: great! I think there many tagged specs on object_spec because of this bug, once you push the correction I'll check then
00:47:08evanok.
00:47:10evanyeah
00:47:13evanthis is a nasty bug
00:47:20evani'm surprised i didn't run into myself.
00:47:20evan:)
00:47:25massI think I got distracted again by a non-rubinius-related conversation :)
00:47:54anonusermass, it's tangentially related :)
00:48:04massI do really well with tangents
00:48:17RyanTM enters the room.
00:48:19RyanTM_ enters the room.
00:56:24benny leaves the room.
00:59:12Defilerwait, there's a daily grind? :)
00:59:55foysavas leaves the room.
01:00:07benstiglitz leaves the room.
01:00:25evangets out the whip for Defiler
01:01:01DefilerYou missed your chance in the hotel room.. Now I am free! FREE! hahaha
01:02:24benburkert enters the room.
01:02:27evantarcieri: you can put
01:02:37evanspec.platform = Gem::Platform.new ['universal', 'rubinius', '0.8']
01:02:42evanin the gem spec
01:03:01tarcieriok, sweet
01:03:01evandrbrain needs to sync the code into rubinius so rubinius knows to match properly
01:03:30tarcierithe Rubinius.asm parser is failing 3 of the tests... so it's gonna be a bit here before I actually get a patch together for them
01:03:51mahargevan: good stuff at railsconf. I went mostly for the alternative implementation stuff and wasn't disapointed :)
01:03:55tarcieriI'm not seeing any errors when running Mongrel on rbx though
01:04:14evanmaharg: :) thanks!
01:04:32evanmaharg: pardon my ignorance, but did we talk face to face?
01:04:45evani can never keep IRC nicks and real names/faces together
01:04:48mahargwe did, briefly
01:04:59massevan: Is that why you forgot I was in denver? :)
01:05:08evanmass: oh, probably.
01:05:09tarcieriusing real names as nicks is helpful in that regard :)
01:05:21evan^5s tarcieri
01:05:26mahargmy real name is the reverse of my nick heh
01:05:36evanor at least keeping your IRC name field correct
01:05:36evanhelps
01:05:43evanbecause i can /whois people
01:06:13mahargduring railsconf I was using my real name as my nick in the #railsconf channel
01:06:13DefilerWilson is just my 'street name'
01:06:28evanDefiler: Wilson is your slave name.
01:06:31massDefiler: I thought your real name was McLovin?
01:06:31Defilerhaha
01:07:09mahargI think we're gonna get one of our coders to start hacking on rubinius some time this week
01:07:11evanmass: the 27 year old hawaiian organ donor? he's a great guy.
01:07:22evanmaharg: awesome!
01:07:28evanmaharg: I forget, where do you work?
01:07:31mahargwe've got a good C/C++ guy who's been hacking in the guts of ruby itself for the last little while
01:07:37mahargnexopia
01:07:50tarcierihacking in the guts of MRI is pain pain pain
01:07:58evani hope you all realize by now that my memory for relating irc names to real people and their jobs is terrible
01:08:06tarcierihas monkeypatched C :(
01:08:12evanso I hope you don't hold it against me
01:08:23anonuserHas anyone ever worked at Turner broadcasting systems?
01:08:34maharghe's been doing crazy things with memory profiling hooks in mri heh
01:09:39maharghis nick'll probably be cthompson when he comes by
01:09:43tarcieriadded support for non-blocking SSLSockets through total absurdity
01:10:39tarcieriI have a native extension that couples to a bunch of externs in the openssl C extension through externs
01:10:40tarcieriugh
01:11:18mahargand I don't blame you. I'm sure you meet hundreds of people at these conferences. I can barely remember the names of people I've known for years
01:12:19evanmaharg: :)
01:13:09anonuserbbl
01:13:31foysavas enters the room.
01:14:11nullstyle leaves the room.
01:15:08anonuser leaves the room.
01:16:58ShayArnett enters the room.
01:17:05trythil enters the room.
01:17:38wycats leaves the room.
01:18:16wycats enters the room.
01:18:44foysavas leaves the room.
01:19:13boyscout1 commit by Evan Phoenix
01:19:14boyscout * Fix a subtend native method calling a primitive; 4a2cb03
01:19:18evancezarsa: there ya go!
01:19:26evangoing to go hang out with abby now
01:19:27evanbbiab.
01:19:47wycats leaves the room.
01:26:24foysavas enters the room.
01:28:11bitsweat enters the room.
01:32:10cored enters the room.
01:33:49rubuildius_ppcEvan Phoenix: 4a2cb0387; 2188 files, 7254 examples, 25973 expectations, 0 failures, 0 errors
01:49:00hemulen enters the room.
01:51:06yugui enters the room.
01:54:54jtoy enters the room.
01:58:45ShayArnett leaves the room.
02:00:56atduskgreg leaves the room.
02:02:07hemulen leaves the room.
02:02:52nullstyle enters the room.
02:05:28dctanner leaves the room.
02:06:46moofbong enters the room.
02:07:49dbussinkhey ppl
02:08:55dbussinkmost already back home?
02:09:49DefilerI've got a couple hours of travel left =(
02:09:58dbussinkDefiler: ah, well, me too
02:10:39DefilerAt least the WiFi actually works here
02:10:47dbussinkjust went hiking a bit today, we went to multnomah falls today
02:11:00dbussinkpretty nice spending of the free day
02:12:08DefilerNice
02:12:19jeremydurham enters the room.
02:12:56dbussinkDefiler: good to do something without a laptop :)
02:13:31Defiler:)
02:13:40DefilerSpeaking of which, time to head to my flight. See you later
02:13:57dbussinkDefiler: have a good flight
02:14:00dbussinkseen ya
02:14:02dbussinksee ya
02:18:38dbussinkbtw, on something totally different, anyone knows whether i can pickup an eee pc here around portland?
02:18:58dbussinka store that has them in stock would be really great
02:19:41trythil leaves the room.
02:20:06kevwil enters the room.
02:23:10RyanTM leaves the room.
02:23:46kevwil leaves the room.
02:24:01RyanTM_ leaves the room.
02:24:05hemulen enters the room.
02:24:53trythil enters the room.
02:25:06VVSiz_ enters the room.
02:26:19boyscout1 commit by Vincent Lu
02:26:20boyscout * Shows failed substring in InvalidCharacter error message.; 4489e32
02:26:56hemulen leaves the room.
02:28:02lstoll enters the room.
02:34:13Yurik_ leaves the room.
02:37:06Yurik enters the room.
02:37:39edwardam leaves the room.
02:38:15Yurik_ enters the room.
02:38:30Yurik_ leaves the room.
02:39:09Yurik leaves the room.
02:39:24rubuildius_ppcVincent Lu: 4489e320a; 2188 files, 7254 examples, 25973 expectations, 0 failures, 0 errors
02:39:27Yurik enters the room.
02:42:26VVSiz leaves the room.
02:46:24rubuildius_amd64Vincent Lu: 4489e320a; 2188 files, 7232 examples, 25917 expectations, 1 failure, 0 errors; http://rafb.net/p/tNnDJS78.html
02:46:25rubuildius_amd64Evan Phoenix: 4a2cb0387; 2188 files, 7232 examples, 25917 expectations, 1 failure, 0 errors; http://rafb.net/p/idvScm48.html
02:47:09shame enters the room.
02:56:07michalw leaves the room.
02:56:26wmoxam enters the room.
03:00:00atduskgreg enters the room.
03:17:17RyanTM enters the room.
03:17:58RyanTM_ enters the room.
03:22:59xhanjian_ enters the room.
03:24:12atduskgreg leaves the room.
03:30:42cored leaves the room.
03:32:56mernen enters the room.
03:33:56oliver_ enters the room.
03:35:50xhanjian leaves the room.
03:36:16wmoxam leaves the room.
03:36:45ShayArnett enters the room.
03:39:15moofbong leaves the room.
03:42:38sambo82 enters the room.
03:42:52moofbong enters the room.
03:45:02jeremydurham leaves the room.
03:55:04lstoll_ enters the room.
03:57:15wycats enters the room.
04:00:05trythil leaves the room.
04:04:37lstoll leaves the room.
04:10:13jbarnette enters the room.
04:13:55trythil enters the room.
04:16:45nicksieger leaves the room.
04:22:50oliver_ leaves the room.
04:31:15nicksieger enters the room.
04:34:09RyanTM leaves the room.
04:34:22RyanTM_ leaves the room.
04:35:50hemulen enters the room.
04:43:35ezmobius enters the room.
04:43:47nicksieger leaves the room.
04:44:47nicksieger enters the room.
04:47:33trythil leaves the room.
04:56:51trythil enters the room.
05:01:22carllerche leaves the room.
05:03:41boyscout2 commits by Cezar Sa Espinola
05:03:43boyscout * Removing fail tags from most subtend specs (only 1 failing spec left!); 21ac763
05:03:43boyscout * Fixing gc when using Data_Wrap_Struct with null mark and free functions.; 98a1b19
05:15:34ShayArnett leaves the room.
05:17:20rubuildius_ppcCezar Sa Espinola: 21ac76333; 2188 files, 7267 examples, 25998 expectations, 0 failures, 0 errors
05:18:53rubuildius_amd64Cezar Sa Espinola: 21ac76333; 2188 files, 7245 examples, 25942 expectations, 1 failure, 0 errors; http://rafb.net/p/xafc5699.html
05:24:44oliver_ enters the room.
05:26:33oliver_ leaves the room.
05:29:45trythil_ enters the room.
05:29:56trythil leaves the room.
05:40:48hemulen leaves the room.
05:42:12nicksieger leaves the room.
05:50:07sambo82_ enters the room.
05:50:22sambo82 leaves the room.
05:52:18obvio171 leaves the room.
05:52:46mernen leaves the room.
05:53:12moofbong leaves the room.
05:54:59headiusanyone here with an interest in subtend?
05:56:36tarcierihas interest in making Ragel parsers not use subtend :)
05:57:08headiusyeah
05:57:27headiuswe now have a library that can load extensions and a contributor is filling out the ruby.h bits for us
05:57:41headiusbut it's becoming increasingly apparent it's infeasible
05:58:12headiuseven the smallest extensions twiddle memory directly or expect object references to have a specific structure
05:58:39headiusI'm just not sure if it's worth the effort now
05:58:46ezmobiusyeah that seems like a rough task trying to support every extension
05:58:48masssounds painful unless you want to create a higher level interface
05:58:49tarcieriyou're talking about binary compatibility iwth MRI?
05:58:50headiusperformance would be terrible either way
05:58:54headiustarcieri yes
05:58:57tarcieriurgh
05:59:18headiusmass even then there's stuff like this from the json gem: if (st_lookup(RHASH(opts)->tbl, tmp, 0)) {
05:59:26headiusthere's no way we'll ever get that to work under rbx or jruby
05:59:48tarcieriisn't the JSON gem Ragel-generated?
05:59:53brixenheadius: what's FFI look like from IR land?
05:59:56headiusit is, but it's using that to pull some options
06:00:09brixenI wonder if we should just push FFI seriously and rewrite some of the C extensions
06:00:16massouch.
06:00:18headiusbrixen CLR has an "unsafe" mode where it can directly link to specific libs, so I suspect an FFI layer is just as feasible
06:00:27brixenheadius: oh cool
06:00:39massyu
06:00:47masserr, you also have externals in CLR
06:00:49headiusand yes, I'm convinced now we need to start pushing FFI and start a concerted campaign against ruby.h
06:00:53massmakes it stupid simple to call C api
06:01:09tarcieriI mean... I could look at converting the JSON gem to Ragel-generated Rubinius.asm
06:01:31brixenheadius: agreed
06:01:54headiustarcieri: I think the need to be able to call C libs is still key though, and that just avoids the issue
06:02:14tarcieriis just of the "less C is better" school of thought
06:02:14headiusmaybe I'll try to look at jruby-ffi this week
06:02:14tarcieriheh
06:02:24headiuswmeissner says it's basically done but needs integration work
06:02:52headiusit also introduces a new library that parallels JNA so we'd need to deal with that...but it could be possible to have FFI in JRuby within days
06:03:09ezmobiustarcieri: hows the perf of the Rubinius.asm mongrel parser?
06:03:12headiusdrbrain and I agreed we also need to provide an FFI for MRI/YARV
06:03:26tarcieriezmobius: two orders of magnitude below the C extension
06:03:27tarcieriheh
06:03:30headiusthat would be the last straw and we could really start pushing it
06:03:35tarcieriwhich should change dramatically with JIT
06:04:15headiustarcieri: is it ruby I could run in JRuby? I'd be curious to repeat the test
06:04:23tarcieriheadius: nope
06:04:36headiusthat's to bad
06:04:44tarcieriit's full of Rubinius.asm
06:04:49tarcierithat lets it do goto in Ruby
06:04:58ezmobiusragel has a mode to output goto machines using Rubinius.asm
06:05:55sambo82 enters the room.
06:06:13sambo82_ leaves the room.
06:06:14headiusmmm, I see, that's too bad
06:06:25headiuswe still have a partial rubinius machine, but no Rubinius.asm support
06:07:03tarcieriyou could modify Ragel to generate JVM bytecode :)
06:07:15ezmobiusragel already generates java though doesnt it?
06:07:37tarcieriI'm sure it can generate a table-based FSM using plain vanilla Java
06:08:18massyes it does, my json4java lib is at its core a ragel-based parser
06:08:37tarcieriI mean, there's already a Mongrel parser for Java
06:09:06headiusoh, we have mongrel and hpricot using the java stuff already
06:09:15headiusI was just curious to try running the ruby version
06:09:27ezmobiusahh it does output pure ruby but it uses continuations
06:09:31tarcieriyeah, you'd need to have it spit out Ruby that compiles to JVM bytecode
06:09:34ezmobiusand is massively slow on mri
06:09:39headiusezmobius: gack
06:09:48tarcieriezmobius: the Ruby it spits out is horrid
06:09:53ezmobiusyeah its fugly
06:10:26tarcieriit's so bad it exposed a Rubinius bug
06:10:26tarcieriheh
06:10:52massthats good then :)
06:10:58massits so bad its useful
06:11:05tarcieriI did a little postprocessing to clean it up
06:12:38elight enters the room.
06:12:57massI think I've had my fill of screencasts for the day
06:23:21jbarnette leaves the room.
06:34:08thehcdreamer enters the room.
06:35:40trythil_ leaves the room.
06:36:03trythil enters the room.
06:40:14maharg_ enters the room.
06:40:32elight leaves the room.
06:41:25trythil leaves the room.
06:41:48bitsweat leaves the room.
06:46:59cezarsa_ enters the room.
06:47:00cezarsa leaves the room.
06:49:37boyscout1 commit by Vincent Lu
06:49:38boyscout * Removing fail tags from most Iconv specs.; a66f3b6
06:50:42obvio171 enters the room.
06:54:45Skip enters the room.
06:58:42thehcdreamer leaves the room.
07:00:36maharg leaves the room.
07:02:15rubuildius_ppcVincent Lu: a66f3b678; 2188 files, 7304 examples, 26090 expectations, 0 failures, 0 errors
07:03:52rubuildius_amd64Vincent Lu: a66f3b678; 2188 files, 7245 examples, 25942 expectations, 1 failure, 0 errors; http://rafb.net/p/2h0t5B79.html
07:19:51shame leaves the room.
07:21:29shame enters the room.
07:24:02blakewatters enters the room.
07:25:51Maledictus enters the room.
07:44:34JonathanT enters the room.
07:44:48JonathanT leaves the room.
07:48:01blakewatters leaves the room.
07:48:29blakewatters enters the room.
07:51:20FoobarWidget enters the room.
07:51:25FoobarWidgethi guys
07:51:42FoobarWidgetwas there a talk on RailsConf on mod_rubinius?
07:53:09headiusno
07:53:15headiusjust a vanilla rubinius talk
07:54:17FoobarWidgetah. I thought I had missed something, but apparently not
07:55:06FoobarWidgethow did your talk go headius? :)
07:56:08blakewatters leaves the room.
07:56:11headiusI just did a little 10-min keynote which seemed to go pretty well and we had an open Q/A session on JRuby that went well
08:07:06tarcieritries to decide if Railsconf would've actually been worth going to
08:07:36hemulen enters the room.
08:08:07wycatstarcieri: it would have
08:09:09mahargI really liked the jruby session. It was the highlight of the last two days for me. Was pretty disapointed with everything else I saw on those two days
08:09:26tarcieriI'm not doing Rails anymore
08:09:35mahargI've never done rails heh
08:09:36tarcierior web development in general
08:09:36tarcieriso
08:09:40wycatsmaharg: :(
08:09:48tarcieriis on the client side of the web
08:09:54wycatsmaharg: did you see my talk?
08:09:56mahargI went for the alt-implementation stuff
08:10:02mahargwhich was yours?
08:10:17tarcieriI've never been to a RailsConf... I think RubyConfs are more up my alley
08:10:19wycatsDataMapper
08:10:21tarciericheaper too
08:11:07mahargno, didn't see datamapper. The coworker I went with did, though, and thought it was good though. He asked the question about sharding.
08:11:18tarcierialthough Florida seems a bit far out there
08:11:47wycatsmaharg: ah
08:11:47maharghe didn't agree with the answer though :P
08:11:50wycatstricky question
08:11:51wycatsyeah
08:11:53wycatsI don't either
08:11:58wycatsnot sure if he could tell ;)
08:12:14mahargheh
08:12:18wycatsbut Sam's argument is that sharding is a complex technique that doesn't let itself well to 80-20
08:12:28wycatsand that the people who will likely use the built-in support won't need it
08:12:41mahargwe built our own framework from the ground up and have it aware of sharding at pretty much every level
08:12:42tarcierilikes databases which use a tuplespace
08:12:55tarcieriautosharding
08:13:08tarcieripeer-to-peer distributed
08:13:37mahargthough tbh most of the awareness is lower level than our orm layer. It's mostly in our sql abstraction
08:13:46wycatsmaharg: it would not be hard, at all, to add sharding-awareness of DM
08:13:59wycatswe already support joins across multiple repositoriesw
08:14:03wycatsrepositories
08:16:07qwert666 enters the room.
08:21:26maharg leaves the room.
08:21:35thehcdreamer enters the room.
08:21:44maharg enters the room.
08:22:19mahargblah my tubes are so clogged tonight. I think I brought the 'bad wifi' bug back home with me
08:23:33wycatshaha
08:23:44wycatsmaharg: so was sharding like a make-or-break for him?
08:23:54mahargwhat's the last thing you say me say?
08:24:53wycats?
08:25:00wycatsthough tbh...
08:25:24maharg01:18 < maharg> unfortunately the boat has kind of already passed for us
08:25:25maharg01:18 < maharg> when we switched to ruby, there was really just activerecord and some very immature projects. And we found it unlikely we'd be able to sanely modify activerecord to do it
08:25:30maharg01:19 < maharg> but I think it was interesting to see how other non-activerecord frameworks are doing things. It sounded like datamapper's a lot like what we built up
08:25:39wycatsheh
08:25:40wycatsawesome
08:26:18nullstyle leaves the room.
08:26:52mahargbut yeah, sharding is definitely required for us
08:27:17maharg15 database servers split by user livejournal-style
08:28:00wycatsyeah
08:28:01ezmobius leaves the room.
08:28:04wycatswould be trivial to add to DM
08:28:09wycatsmaybe I'll bug ssmoot about it again
08:31:39mahargwe tend to think that lack of ability to properly scale your database (beyond the pretty brutal method of distributing reads to slaves) is a bigger part of ruby/rails scaling issues than it's given credit for
08:32:35wycatsmaharg: it's not like it's hard to add such a solution on top of something like DM
08:32:44wycatshow does Hibernate handle it?
08:33:19mahargI can't see it being anything but clunky on top. There are a lot of assumptions about how to connect to the database in most orm frameworks
08:33:28mahargand they don't tend to play nicely with sharding
08:33:45wycatsmaharg: DM is written with multiple repositories as a central notion
08:36:54mahargcool. I'll have to take a closer look at it some time
08:38:17wycats:)
08:38:18wycatsyeah
08:38:30wycatsI'll see what I can do about at least adding the appropriate hooks for sharding
08:38:33gnufied_ enters the room.
08:38:37wycatslike I said, the infrastructure is there
08:38:46wycatsmaharg: do you know how hibernate handles sharding?
08:39:03mahargnope
08:39:09wycatshttp://www.hibernate.org/414.html
08:39:13wycatsI'll show this to Sam
08:39:16wycatshe respects hibernate
08:45:03shame leaves the room.
08:53:10mutle enters the room.
08:54:17Arjen_ enters the room.
08:59:22edwardam enters the room.
08:59:24shame enters the room.
09:01:14maharg leaves the room.
09:01:28maharg enters the room.
09:05:35brainopia enters the room.
09:29:56imajes_ enters the room.
09:40:03brainopia leaves the room.
09:40:08qwert666 leaves the room.
09:42:01lstoll leaves the room.
09:44:05michalw enters the room.
09:59:15krisps enters the room.
10:08:48radarek enters the room.
10:11:04octopod enters the room.
10:22:10antares_ enters the room.
10:23:44be9 enters the room.
10:24:50FoobarWidget leaves the room.
10:47:04chris2 enters the room.
10:54:09lstoll enters the room.
11:03:26krisps leaves the room.
11:05:09krisps enters the room.
11:08:53jtoy leaves the room.
11:11:19krisps leaves the room.
11:12:28krisps enters the room.
11:18:15benny enters the room.
11:23:14imajes_ leaves the room.
11:48:15qwert666 enters the room.
11:59:23jero5 leaves the room.
12:15:03olabini enters the room.
12:19:32hemant__ enters the room.
12:25:15_mutle enters the room.
12:27:56gnufied leaves the room.
12:29:38antares_ leaves the room.
12:40:26mutle leaves the room.
12:42:29benny leaves the room.
12:43:28benny enters the room.
12:43:47charlenopires enters the room.
12:43:49NoKarma enters the room.
12:46:03Yurik leaves the room.
12:48:02RyanTM enters the room.
12:48:11RyanTM_ enters the room.
12:52:47tim_w enters the room.
12:55:14chris2 leaves the room.
13:05:58gnufied enters the room.
13:05:58hemant__ leaves the room.
13:18:39charlenopires leaves the room.
13:22:24charlenopires enters the room.
13:27:54gnufied_ leaves the room.
13:29:00michalw leaves the room.
13:29:58RyanTM leaves the room.
13:30:46RyanTM_ leaves the room.
13:31:17RyanTM enters the room.
13:31:47RyanTM_ enters the room.
13:43:28gnufied_ enters the room.
13:43:40gnufied leaves the room.
13:46:56xhanjian enters the room.
13:49:21blakewatters enters the room.
13:50:27xhanjian leaves the room.
13:53:51ShayArnett enters the room.
14:05:08wmoxam enters the room.
14:05:45boyscout1 commit by Arthur Schreiber
14:05:46boyscout * Updated MSpec source to fdcdcd35.; 5fbe1f8
14:13:13pauldix enters the room.
14:13:38AndrewO enters the room.
14:15:14qwert666_ enters the room.
14:18:37rubuildius_ppcArthur Schreiber: 5fbe1f8eb; 2188 files, 7304 examples, 26090 expectations, 0 failures, 0 errors
14:21:26benny leaves the room.
14:29:52nrlzd enters the room.
14:32:31Arjen_ leaves the room.
14:32:39qwert666 leaves the room.
14:38:41foysavas leaves the room.
14:39:37atmos enters the room.
14:41:38Yurik_ enters the room.
14:46:26oliver___ enters the room.
14:58:52macournoyer enters the room.
14:58:57lstoll leaves the room.
14:59:31moofbong enters the room.
15:01:10wmoxam leaves the room.
15:01:13wmoxam enters the room.
15:03:01obvio171 leaves the room.
15:05:42foysavas enters the room.
15:06:34oliver___ leaves the room.
15:07:00antares enters the room.
15:13:10yugui leaves the room.
15:21:00benny enters the room.
15:25:56kAworu leaves the room.
15:26:23kAworu enters the room.
15:27:15rubuildius_amd64Arthur Schreiber: 5fbe1f8eb; 2188 files, 7245 examples, 25942 expectations, 1 failure, 0 errors; http://rafb.net/p/1U10fe90.html
15:32:01kAworu leaves the room.
15:38:39kAworu enters the room.
15:38:41oliver___ enters the room.
15:42:00hemulen leaves the room.
15:47:08antares leaves the room.
15:48:31nullstyle enters the room.
15:50:43kAworu leaves the room.
15:55:56kAworu enters the room.
15:58:16octopod leaves the room.
16:00:23lopex enters the room.
16:02:59therealadam enters the room.
16:03:17benny leaves the room.
16:06:03gnufied_ leaves the room.
16:09:36octopod enters the room.
16:16:10benburkert enters the room.
16:20:33gnufied enters the room.
16:28:59enebo enters the room.
16:33:13nullstyle leaves the room.
16:43:31hemulen enters the room.
16:44:08hemulen leaves the room.
16:44:11hemulen enters the room.
16:45:41rue leaves the room.
16:47:16elight enters the room.
16:48:52nexcastellan enters the room.
16:48:52dgtized leaves the room.
16:49:24anonuser enters the room.
16:49:41jlindley enters the room.
16:53:21MontyGS enters the room.
16:55:21jbarnette enters the room.
16:56:00nullstyle enters the room.
16:58:26sambo82 leaves the room.
17:07:46bricolage enters the room.
17:19:46yugui enters the room.
17:26:00mutle leaves the room.
17:30:42macournoyer leaves the room.
17:30:57jbarnette leaves the room.
17:34:52benstiglitz enters the room.
17:35:50scooprhttp://rubystuff.org/ludicrous/ o_O
17:39:40evanyeah
17:39:44evani've heard him talk about i.t
17:42:03scooprdidn't quite get the gist of it from the description
17:42:28scooprthat might be because I'm unsure what libjit actually does
17:42:53scooprexcept, that it does jit =)
17:43:35evanyeah
17:43:46evanit's a simple C library for generating and execute machine code
17:44:58scoopris it usably good at it?
17:45:14evanit's ok.
17:45:19evanit's the most basic
17:45:22evanno optimizations
17:45:37evanthe API is nice
17:45:55evanit's a similar API setup to LLVM actually
17:46:49scooprany reason to use it over llvm, except perhaps simplicity?
17:47:08evani've used it before actually
17:47:20evanit had an incompatable license and wasn't being maintained for a long time
17:50:06thehcdreamer leaves the room.
17:51:29macournoyer enters the room.
17:51:55krisps enters the room.
17:55:05d2dchat leaves the room.
18:04:01atduskgreg enters the room.
18:11:53wycats leaves the room.
18:12:28jbarnette enters the room.
18:14:47enebo leaves the room.
18:19:16nexcastellanI just created a ticket and I'm not sure my patch made it. Can someone please quickly look at http://rubinius.lighthouseapp.com/projects/5089-rubinius/tickets/584 and let me know if there's an attached (trivial) patch? Please and thank you.
18:22:29anonuserThere is an attached patch
18:22:52nexcastellanThank you.
18:22:53jlindley leaves the room.
18:22:53anonuserhttp://rubinius.lighthouseapp.com/attachments/25697/0001-OS-X-fink-section-points-to-a-URL-which-n o-longer-ex.patch
18:23:00anonuserDoes that look correct?
18:23:13JonathanT enters the room.
18:23:25anonuserWell bbl Taco time!
18:23:28JonathanT leaves the room.
18:23:29nexcastellanYes. I just missed the "attachments" section on the lighthouse page. I was expecting it to be in a different place. :)
18:23:55octopod leaves the room.
18:29:10jlindley enters the room.
18:30:50wycats enters the room.
18:35:11dbussinktime to go to the airport, yay for a 10 hour flight :(
18:35:26evandbussink: have a good flight!
18:35:39evanit was tons of fun seeing and hanging out with ya
18:35:48dbussinkevan: did you take a look at the c++ stuff i made during these days?
18:36:02evandid you commit it?
18:36:10dbussinkyeah, it's all in
18:36:22evank
18:36:24evani didn't
18:36:25evani'll look now.
18:36:55evanoh..
18:36:58evanlots of commits!
18:36:58evanyay~
18:38:06dbussinki have 450 tests now :)
18:38:14evansweeeeet!
18:38:30evani love having tests for the VM
18:38:31dbussinkpretty much covered all the stuff i wrote
18:38:40dbussinkplus added some where stuff was missing
18:38:53octopod enters the room.
18:39:41anonuserTaco time was not that great :(
18:40:17dbussinkanonuser: it came out the same route as it got in? :P
18:40:30evanhah
18:40:47anonuserdbussink, no.
18:40:58anonuserIt tasted funny.
18:41:05JonathanT enters the room.
18:41:11JonathanT leaves the room.
18:41:14enebo enters the room.
18:41:18anonuserdbussink, though I wouldn't be surprised. I'm vegan and it tasted really odd.
18:42:28dbussinkevan: i'd love to get the ruby benchmarks running against these new primitives, i got them simplified quite a bit in various places
18:42:48evanok
18:42:56evanwe should be able to get to that soon
18:44:15anonuserA bit off topic, as usual, but does anyone have experience with pulseaudio and bluetooth headphones?
18:45:22evandbussink: do you fly direct back?
18:48:10dbussinkevan: yeah, straight flight to amsterdam, pretty nice
18:49:44evannice.
18:51:17benny enters the room.
18:53:08gnufiedanonuser, get rid of pulseaudio. ;)
18:53:22anonusergnufied, it works the best with bluetooth believe it or not.
18:54:00anonuserI get flashsupport with the thing. and I don't have to worry about rerouting gstreamer audiosinks
18:54:20gnufiedare you on fedora?
18:54:25anonuserNo
18:54:41anonuserUbuntu. The only distro that installed with ease :(
18:55:01anonuserFreeBSD doesn't boot on this machine, and I had to get rid of wireless to make linux work.
18:55:10anonuserdamn shoddy rt61pci driver :(
18:55:13dbussinkbut i've got to go, see you ppl later
18:59:24jlindley leaves the room.
19:00:58rue enters the room.
19:04:57JonathanT enters the room.
19:14:14gnufied leaves the room.
19:16:58vertiginous enters the room.
19:18:38xhanjian_ leaves the room.
19:19:24massI'm constantly surprised by methods which are part of rails rather than core ruby
19:20:00evanlike?
19:20:04massindex_by
19:20:31evanyeah, it's funny how thats worked
19:20:41evanbecause rails has moved so much faster than 1.8
19:20:48evanit's added all these methods people have always wanted.
19:21:02benny leaves the room.
19:21:35massthe implementation is pretty dirt simple as well
19:22:45cezarsaevan: one subtend question, isn't subtend_wrap_struct() supposed to set obj->obj_type = WrapsStructType?
19:22:57evani think so
19:23:02evankevinclark added that
19:23:07evanand it hasn't been touched much since
19:24:17cezarsahm ok, I'll try adding RDATA and RDATA_PTR, so I'll set the obj_type to know how to copy data back to the object
19:24:44evank
19:24:55evanthat sounds like the right way to do that
19:26:56drbrainhttp://webkit.org/specs/squirrelfish-bytecode.html
19:27:00drbrainneat
19:27:13evancool
19:27:19evanyeah, it's all register based
19:27:30evanwith auto-windowing
19:27:43evanwhich means that functions don't have to save and restore registers between function calls
19:27:46evanthats handled for them
19:31:20thehcdreamer enters the room.
19:31:49xhanjian enters the room.
19:33:03masssquirrelfish?
19:33:07masswtf is a squirrelfish? :)
19:33:32JonathanT leaves the room.
19:33:34masshmm
19:33:42massthat barely even looks like a fish, let alone a squirrel
19:34:00mass(looking at real squirrelfish picts)
19:34:09cezarsahttp://images.google.com/images?q=squirrelfish
19:34:20massyep :)
19:34:34masswow thats a big boost
19:34:46massI didn't know they were working on such a thing
19:37:18benstiglitz leaves the room.
19:38:22robin_dewd enters the room.
19:38:57benstiglitz enters the room.
19:39:44drbrainnew JS interpreter for webkit
19:40:00drbrainthe name follows Mozilla's SpiderMonkey
19:40:40jbarnetteif only it didn't suck so hard for cross-platform embedding -- we really wanted to add it to Johnson
19:41:40anonuserhttp://webkit.org/blog/
19:41:49anonuserThat gives a lowdown on squirrelfish
19:45:37rueHello
19:46:07evanrue: allo.
19:47:48michalw enters the room.
19:49:10massI'm bugging #webkit about it. Apparently they have a P1/Critical bug now regarding the lack of T-shirts
19:52:27zenspiderhaha
19:54:05evanha
19:57:01gnufied enters the room.
19:57:05cezarsawondering how great a rubinius T-shirt would be :)
19:58:01jicksta enters the room.
19:59:24nicksieger enters the room.
20:00:44evancezarsa_: i think we're going to make some
20:00:48evanwe did for railsconf last year
20:01:40wycatsis there a way to insert a module between a class and a module that was already included?
20:01:45wycatsin the method lookup chain
20:02:07evanyou can break the chain and insert it manually, sure.
20:02:40wycatshow?
20:03:00evancheck out IncludedModule in kernel/bootstrap
20:03:08evanyou'll make one
20:03:09anonuserYou should put some Rubinius shirts up for sale.
20:03:17benny enters the room.
20:03:17evanand hook it in where you want it manually
20:03:29evanby looking up direct_superclass to where you want it
20:03:36evanthen pointing the left to your new object
20:03:39wycatsnice
20:03:46evanand the new object's superclass to old super
20:03:53evanbasic linked list insertion
20:04:04wycatsdamn you and your pure-ruby implementation ;)
20:04:10anonusermmmm basic somewhat fast datastructures.
20:07:35krisps leaves the room.
20:09:14massI want a rubinius shirt too
20:09:54massshall I go to lighthouse? :)
20:10:00evanheh
20:10:12cezarsaI would certanly order a rubinius shirt here from Brazil :)
20:10:31massmaybe a [PATCH] with a .eps file
20:10:36evanhah
20:16:09trythil enters the room.
20:16:22ShayArnett leaves the room.
20:19:14trythil_ enters the room.
20:19:14trythil leaves the room.
20:37:29yugui leaves the room.
20:41:09krisps enters the room.
20:43:36ezmobius enters the room.
20:51:32nexcastellanShould bin/mspec ci -B full.mspec be expected to fail with an error about AF_INET6 (full error can be posted) on my OS X 10.5 laptop which does not use IPv6?
20:52:18DefilerIt happens, but it shouldn't
20:52:42DefilerOn Mac OS you get back the ::1 crap instead of 127.0.0.1 sometimes, but I don't understand the scenario yet well enough to fix the spec
20:52:52drbrainit may be due to the 10.5.3 changes, too
20:55:09nexcastellan::1 is a reasonable result under IPv6, no?
20:55:41drbrainoh, and I don't think you can turn off IPv6 on lo0 on OS X
20:55:47drbrainonly on your real interfaces
20:55:59benburkert leaves the room.
20:56:12pauldix_ enters the room.
20:56:22nexcastellanProbably true, the return value does seem to indicate that IPv6 is active. :)
20:56:23drbrainat least, there's no UI for it I've found in the Network Pref Pane
20:56:23pauldix leaves the room.
20:57:58drbrainyes, you should always receive an IPv6 address on lo0
20:58:11olabini leaves the room.
20:58:13drbrain(unless you find some utility to disable it)
20:59:57TheVoice enters the room.
21:00:15radarek leaves the room.
21:01:45DefilerWe should just change that spec to be 'should be either 127.0.0.1 or ::1'
21:01:46Defilerprobably
21:06:00be9 leaves the room.
21:07:52krisps_ enters the room.
21:08:25anonuserQuick question, is rubinius supporting 1.9 at all?
21:08:40evannah
21:08:45evanwe haven't spent any time on it
21:08:56nexcastellanActually, if I am reading the output properly, the output I am getting is '["AF_INET6", 80, "localhost", "fe80::1%lo0", 30, 2, 17]'. The problem is the fourth param, fe80::1%lo0, while the expected output only includes ::1.
21:08:59anonuserokay just checking, I was wondering about the new 1.8.7 release
21:09:12evanwe're not going to touch 1.8.7 for a while
21:09:20evanthey introduce all kinds of new APIs and stuff
21:09:26anonuserthe enumeration support would be one of those things that crosses 1.8 and 1.9 divides
21:09:31anonuserahh
21:09:31benny leaves the room.
21:09:52Defiler1.8.7 is the worst thing since 1.8.3
21:09:54Defiler=(
21:10:04anonuserwhy do you say that?
21:10:07ezmobius1.8.7 is very broken
21:10:07evanwasn't in 1.8.5 that everyone skipped?
21:10:09VVSizDefiler: how so?
21:10:11ezmobiuserb doesnt even work
21:10:14evans/in/it/
21:10:17Defilerhttp://svn.ruby-lang.org/repos/ruby/tags/v1_8_7/NEWS
21:10:20Defilerread that file. *shiver*
21:10:27evanthey didn't seem to run the tests before releasing
21:10:35evanso tons of incompatibilities and breakages
21:10:48evannot tons
21:10:54anonuser:(
21:11:01evanbut 100% more than their should have been
21:11:49anonuserthat is odd, minor releases are usually maintenance releases :-\
21:11:52nexcastellanForget what I said earlier, I think I'm looking at the wrong bit...
21:12:12headius1.8.7 is definitely not as minor as it should have been