Index

Show enters and exits. Hide enters and exits.

<
00:06:03evanyou need to click on the vm directory
00:07:24joachimmevan: I believe thats what I am doing,
00:07:40evanyou should see stats for all the files in vm/
00:10:44AndrewO enters the room.
00:10:45evanwhat should the title of by Kaigi talk be?
00:10:55evanprobably needs to be something that will translate well
00:13:46evandang, headius isn't around.
00:13:57brixenyeah, he's hiding from me
00:14:06brixenafter all this bitching about separate rubyspecs
00:14:26evanyou'd think he'd want to gloat that we're organizing the specs like he's talked about in the past
00:14:32brixenheh
00:14:33brixentotally
00:14:36evanhe's at J1 though
00:14:41brixenyeah
00:23:09drbraindo we have an equivalent of rb_syswait?
00:23:50yugui leaves the room.
00:24:29ezmobiushttp://www.news.com/8301-13577_3-9940042-36.html
00:24:32ezmobiusjava one has a virus
00:25:13evandrbrain: hm.
00:25:18evanwhats syswait do...
00:25:49drbrainit waits on a process until it exists
00:25:53drbrainsystem uses it
00:26:12drbrainsystem and IO#close
00:26:22evanezmobius: eeew..
00:26:29evanoh
00:26:33evanyeah Process.wait
00:26:56drbrainwell, its got some signal() stuff in here too
00:27:04evansignal?
00:27:04joachimmevan: ok, It works now. I had an issue with the coverage that prevented the build from completing.
00:27:05evanlet me look.
00:27:23evanjoachimm: ah! i'm going to clean up the build process this weekend.
00:27:46drbrainso I think it actually kills the process
00:28:06evandrbrain: ignore the signal stuff
00:28:06drbrainhrm, no, it just temporarily ignores them
00:28:08evanit doesn't effect us.
00:28:15joachimmevan: are you intentionally not testing exceptions?
00:28:18evanthats them working around being interrupted
00:28:22drbrainok
00:28:38evanjoachimm: how do ya mean?
00:29:25evansweet, ordered an interegrated ipod adapter for my car
00:29:34joachimmevan: I am looking at what is not covered. A quick glance seems to suggest that it is often lines that throws exceptions
00:29:39evanno more shitty fm modulator.
00:29:50evanjoachimm: ah!
00:30:03evanbecause of time, i'm not always covering stuff 100%
00:30:13evanas time goes on, the tests will need to be shored up
00:30:16drbrainI need to integrate an ipod adapter into my car
00:30:18evanusing the coverage and verification
00:30:19obiejuan enters the room.
00:30:27evans/and/as/
00:30:49evandrbrain: you'd be better off writing a boombox into your car i'll bet.
00:30:51evan:D
00:31:36drbrainif I use an actual dock, I'll just pull the ipod out to switch songs
00:31:52evanyeah
00:32:06drbrainI could also find a spare stereo and mod it
00:32:09evanlooks like someone made box to translate the cd changer controls into ipod controls
00:32:12drbrainso I could use the volume pots
00:32:56drbrainthat would be sweet, real push-button controls
00:33:30evanyeah
00:33:42evanhopefully it feeds track names through
00:33:47evanthough i don't care as much about that
00:34:02drbrainI don't know if I could keep the same *CHUNK* sound though, I don't remember how deep all the levers are
00:34:19evanheh
00:34:27evancrutchfield makes decks to go in old cars
00:34:35evanbefore they had a DIN standard
00:34:44evanthat come with customized faceplates
00:34:46joachimmevan: If I was to write a test, would I need to regenerate something or can I just start doing changes in test_*.hpp and rerunning the tests?
00:34:49evanthat had analog knobs and such
00:34:52evanhave
00:35:09evanjoachimm: should be able to just edit the test_* files and run 'rake'
00:35:42joachimmevan: ok, that generates a new runner.cpp. great
00:36:05evanyep!
00:38:16lstoll enters the room.
00:38:20ruejoachimm: How do you mean lines that raise exceptions?
00:39:13rueUp, scrollback, disregard
00:39:34joachimmrue: it was just a first observation. It was just coincidence that the files I checked had 'uncovered' throws. I see know that there are various things missing.
00:39:52joachimms/know/now
00:40:29ruejoachimm: Yeah, and depends on the nature of the coverage too of course. I recall someone who was fond of writing does_not_throw assertions :)
00:44:26rueHaha, 227-column terminal is wide enough for g++ error messages
00:47:15boyscout1 commit by Ryan Davis
00:47:16boyscout * Fixed up pack for base64 and uuencode to be MUCH MUCH cleaner and 2x faster; 2b24a1e
00:50:06ezmobius leaves the room.
00:51:08Maledictus leaves the room.
00:58:33rubuildius_amd64Ryan Davis: 2b24a1e84; 2107 files, 6858 examples, 25048 expectations, 1 failure, 0 errors;
00:59:30rubuildius_ppcRyan Davis: 2b24a1e84; 1280 files, 5652 examples, 20950 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/194642
01:02:24djwhittack, looks like rubuildius_amd64 isn't pasting correctly
01:02:31djwhittI'll get the error message for ya
01:04:16djwhittzenspider: there ya go: http://pastie.caboo.se/194645 (failure from last commit)
01:04:41benstiglitz leaves the room.
01:04:49evanyeah
01:04:51evanthat fucking spec
01:04:58evanit unstable.
01:05:00evanit's
01:05:08zenspiderdjwhitt: hrm... that's not me. can we exclude that one?
01:05:11brixenyeah, that spec is such a pita
01:05:20evanlets exclude it from ci.
01:05:25zenspideror should we get someone on it to stablize it?
01:05:30wycats__ enters the room.
01:05:30brixenzenspider: what's your github username?
01:05:40brixenlet's just exclude it for now
01:05:49zenspiderbrixen: I just made one... zenspider
01:05:52brixenI'm about to pull the trigger on this spec split
01:05:57brixenzenspider: sweet, thanks
01:15:31evandrbrain: SIGPIPE needs to be ignored
01:15:39drbrainok
01:15:40evanMRI flat out sets it to SIGIGN
01:15:58evanSIGPIPE is actually a crappy signal
01:16:05drbrainbut, that's a different problem than the IO#close one I'm working on, so I'll leave it for now
01:16:12evank
01:20:15zenspiderI'm fading... lack of sleep + workout == tarded
01:20:29evanany plans this weekend?
01:21:10wycats leaves the room.
01:21:26drbrainha! yeah!
01:21:43drbrainI switched it from trying to run ruby to trying to run true(1)
01:21:48drbrainnow it gets farther
01:22:30drbrain... but our reopen is only half implemented
01:23:35evanwhich half?
01:23:58drbrainit doesn't support path/mode
01:24:04drbrainonly reopening to ios
01:24:28evanthats easy to fix
01:26:48drbrainyeah
01:27:01drbrainswitching to cat and it goes back to failing :/
01:27:09drbrainso something weird is going on in popen
01:38:06dysinger leaves the room.
01:41:14cored leaves the room.
01:41:39ezmobius enters the room.
01:43:48AndrewO leaves the room.
01:45:10dysinger enters the room.
01:50:11eventualbuddha leaves the room.
01:53:50jtoy enters the room.
01:55:10jtoy leaves the room.
01:55:20jtoy enters the room.
01:56:09brixenman, we have come a long ways
01:56:16brixenwe can run specs with this output 1280 files, 5931 examples, 21145 expectations, 108 failures, 50 errors
01:56:34brixenand still run to completion without hosing the vm
01:56:43evan:)
01:56:46evanwe have, we have.
02:03:36rubuildius_amd64 leaves the room.
02:03:40rubuildius_amd64 enters the room.
02:04:00yugui enters the room.
02:14:03sambo82 enters the room.
02:20:56srbaker enters the room.
02:22:46technomancy leaves the room.
02:28:57blakewatters leaves the room.
02:30:27ezmobius leaves the room.
02:30:28gdagley leaves the room.
02:30:53gdagley enters the room.
02:31:07yugui_ enters the room.
02:31:10Yurik leaves the room.
02:32:05VVSiz_ enters the room.
02:37:30boyscout5 commits by Brian Ford
02:37:31boyscout * Change CI config to use spec/frozen rather than spec/ruby.; e51d8c0
02:37:32boyscout * Bandaid fixes for CI to spec/core/kernel specs. Further work needed.; cefb90f
02:37:33boyscout * Renamed spec/tags/ruby to spec/tags/frozen to track rubyspec submodule.; e7d09e5
02:37:34boyscout * Added spec:init logic to init and update spec/frozen submodule.; 7510986
02:37:35boyscout * Added rubyspecs as a submodule at spec/frozen.; f8ea3b6
02:43:27brixenholy crap, ci.rubini.us is fast
02:43:55brixenit had the run done 2 minutes after the push
02:44:11brixenevan: do you have a post-commit hook that notifies the ci bots?
02:45:31yugui leaves the room.
02:48:48rubuildius_amd64Brian Ford: e51d8c027; 2107 files, 6857 examples, 25046 expectations, 0 failures, 0 errors
02:48:59anteaya leaves the room.
02:49:50VVSiz leaves the room.
02:50:16rubuildius_ppcBrian Ford: e51d8c027; 1280 files, 5651 examples, 20948 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/194668
02:52:25yugui_ leaves the room.
02:52:41yugui enters the room.
02:55:31boyscout2 commits by Brian Ford
02:55:32boyscout * Fare thee well dear, dear specs.; e11d9b4
02:55:33boyscout * Changed toplevel spec_helper.rb to reference spec/frozen.; 33f805d
02:55:39GMFlash enters the room.
02:56:01rueI am tempted to just push in a change that says YAY
02:56:10brixenheh
02:58:15evanhehehe
02:59:19obiejuan leaves the room.
03:05:53cremesdjwhitt: you around? need help with finding new 'bot
03:06:55djwhittcremes: http://github.com/djwhitt/rubuildius/tree/master
03:06:59djwhittcremes: there you go
03:07:07cremesdjwhitt: okay...
03:07:45djwhittcremes: let me know if you have any questions or problems
03:08:17gdagley leaves the room.
03:08:21rubuildius_amd64Brian Ford: e11d9b4f6; 2107 files, 6857 examples, 25046 expectations, 0 failures, 0 errors
03:08:42cremesdjwhitt: tried: git clone http://github.com/djwhitt/rubuildius/tree/master rubuildius - no worky; what's the correct incantation?
03:08:45boyscout1 commit by Evan Phoenix
03:08:45boyscout * Ignore spec/ruby since it's a clone; 00c9c73
03:09:18djwhittcremes: yeah, that's just the url for the github web page
03:09:23djwhittcremes: this is the clone url: git://github.com/djwhitt/rubuildius.git
03:10:35brixenevan: heh, just about put push that :P
03:10:53cremesdjwhitt: okay, grabbed the latest... will customize for my box and relaunch
03:11:04cremesdjwhitt: any testing i should do?
03:11:14evanbrixen: :)
03:11:17cremesdjwhitt: i like to fire-and-forget. :-)
03:11:18evani'm off to get some greek food.
03:11:21lopex leaves the room.
03:11:26brixenyeah, I'm off for food in a sec
03:11:40brixenI'll finish these spec:update and spec:push task later
03:11:49brixendo the announce and all that
03:11:54brixenyada yada :)
03:12:12djwhittcremes: well, you could point it to #rubuildius_test for the first run, but other than that it should work if all the environment files in bot_nanny.sh are set right
03:12:44brixenamazing that nothing broke horribly today
03:12:47cremesdjwhitt: ok, will check it out. thanks for your work.
03:13:17djwhittcremes: you're welcome. thanks for updating :)
03:13:53cremesdjwhitt: no prob. been meaning to for a few weeks; vacation and life got in the way
03:14:25boyscout2 commits by Brian Ford
03:14:26boyscout * Added spec:clone task to clone rubyspec into spec/ruby for editing.; 47590d5
03:14:27boyscout * Updated .gitignore to ignore spec/ruby directory.; e9ff971
03:15:37rubuildius_amd64Brian Ford: 47590d505; build failed! http://rafb.net/p/7xJhQV35.html
03:16:16brixenhmm, that's a problem
03:16:33brixenI think people will have to do rake spec:clone manually if they have perms
03:16:49brixenI'll take that out of being automatic
03:18:09rubuildius_amd64Evan Phoenix: 00c9c73f5; 2107 files, 6857 examples, 25046 expectations, 0 failures, 0 errors
03:18:37djwhittbrixen: just consider that a test of my error paste fix ;)
03:18:48rubuildius_ppcBrian Ford: 47590d505; build failed! http://pastie.caboo.se/paste/194679
03:19:02brixendjwhitt: :)
03:19:22srbaker leaves the room.
03:19:24brixen19:12 brixen >> amazing that nothing broke horribly today
03:19:29brixenI jinxed myself :P
03:19:31jeremydurham enters the room.
03:19:58cremesdjwhitt: switching over as soon as latest compile completes; any last minute details i should be aware of?
03:20:45djwhittcremes: nah, don't think so
03:20:57cremesdjwhitt: k
03:21:26cremesdjwhitt: will watch for a few minutes then pass out after my usual friday night "binge" :)
03:21:41djwhittcremes: hehe
03:21:45boyscout1 commit by Brian Ford
03:21:46boyscout * Do not automatically attempt to clone committer url for rubyspecs.; d65cbc6
03:25:09cremesdjwhitt: did you put bot_nanny.sh into cron?
03:25:40djwhittcremes: nope, I typically run it via screen
03:25:57djwhittcremes: wouldn't really make sense in cron either since it would spawn a new bot everytime it ran
03:26:26cremesdjwhitt: damn... glad someone here is sober. ;)
03:26:32brixenheh
03:26:46brixenok, bbiab
03:26:53djwhittcremes: of course I still have this in cron to force the bot to check for update: */5 * * * * kill -USR2 `cat $HOME/rubuildius/matzbot.pid`
03:27:11djwhittcremes: you should have something like that in cron already for the old bot
03:27:24cremesdjwhitt: good catch; i'll update old cron to match that new signature
03:29:22cremesdjwhitt: when is EY taking over all 'bot action? i thought we were merely an interim solution...
03:30:10djwhittcremes: well, there is a ci.rubini.us going right now. I'm not sure if they're planning on transitioning completely to that at some point or not
03:30:27djwhittcremes: there aren't any bots in the channel from it yet
03:30:37cremesdjwhitt: k; will run as long as i am needed
03:30:52cremesdjwhitt: someone has to provide some big-endian lovin'
03:31:50cremessheesh... machine is idle but no response from the (original) bot yet
03:33:00djwhittcremes: hmm... did you move the old bot or something? perhaps it never got the signal from the cron job
03:33:07rubuildius_amd64Brian Ford: d65cbc69b; 2107 files, 6857 examples, 25046 expectations, 0 failures, 0 errors
03:33:46cremesdjwhitt: now i see more work from rubinius... but there was definitely a measurable lag
03:34:04cremesi'll switch to new 'bot as soon as current one is done
03:34:09maharg leaves the room.
03:34:09djwhittk
03:35:34rubuildius_ppcBrian Ford: d65cbc69b; 1280 files, 5651 examples, 20948 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/194684
03:35:35rubuildius_ppcEvan Phoenix: 00c9c73f5; 1280 files, 5651 examples, 20948 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/194683
03:35:36rubuildius_ppcBrian Ford: e11d9b4f6; 1280 files, 5651 examples, 20948 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/194673
03:35:50cremesokay, switching now...
03:35:53rubuildius_ppc leaves the room.
03:37:13cremesdjwhitt: switched; watch for fireworks :)
03:37:22cremesquick; someone commit something!
03:37:45djwhittwaits for something to explode
03:38:08djwhittI think it'll run at least once even without a fresh commit
03:38:43djwhittthough, this one actually saves session so it won't re-run after restarts
03:38:53cremesdjwhitt: hmm... it's happily sitting here watching irc without any response
03:39:59djwhittcremes: oh, one thing I did forget. make sure you have a clone of rubinius in repo/rubinius/HEAD
03:40:18cremesdjwhitt: it doesn't clone it automatically?
03:40:25djwhittcremes: not the first time
03:40:38cremesk; will restart after i do a manual clone
03:41:10djwhittcremes: k. also, keep in mind that it might take a few minutes until cron sends the bot the signal to pull changes
03:47:28srbaker enters the room.
03:51:38ShayArnett enters the room.
03:53:07seydar enters the room.
04:13:26MenTaLguY enters the room.
04:15:01MenTaLguYhello
04:16:39brixenhi MenTaLguY
04:16:57brixenthe actor page on LH is sweet btw, I just noticed it today
04:17:25MenTaLguYcool
04:17:34MenTaLguYI haven't really sat down and gone over it yet
04:17:56brixenMenTaLguY: did you do the graphic for it?
04:18:12MenTaLguYnope
04:18:43brixenoh interesting, evan and I speculated it was yours :)
04:18:54brixenthe nice gradients and whatnot heh
04:20:20MenTaLguYit does look nice
04:20:58MenTaLguYI'm not sure how accurate it really is
04:21:04MenTaLguYbut then again it's a design sketch
04:21:17MenTaLguYand I think the stuff I'm implementing now will end up looking a bit more like the diagram in the end
04:21:29MenTaLguYthan it does currently
04:21:46brixenwell, it's certainly cool to have this stuff being developed
04:23:42srbaker leaves the room.
04:24:29MenTaLguYyeah
04:28:56trythil enters the room.
04:29:28ShayArnett leaves the room.
04:32:36trythil leaves the room.
04:35:24benburkert_ enters the room.
04:35:24benburkert leaves the room.
04:37:33seydar leaves the room.
04:42:24trythil enters the room.
04:49:21trythil leaves the room.
04:57:02boyscout2 commits by Wilson Bilkovich
04:57:03boyscout * Clean up OpenSSL digest implementation; e4fc5e2
04:57:04boyscout * Print a readable error message for nameless modules with cvars; f097fcf
05:08:10rubuildius_amd64Wilson Bilkovich: e4fc5e2ed; 2107 files, 6857 examples, 25046 expectations, 0 failures, 0 errors
05:09:30stepheneb leaves the room.
05:16:25MenTaLguY leaves the room.
05:39:46mernen leaves the room.
05:45:22shame enters the room.
05:46:55benburkert_ leaves the room.
06:01:45ruebrixen: Think it was mainly rosejn setting up that page. tarcieri added a couple things too
06:01:59brixenrue: ahh ok
06:03:16brixenhrm, looks like ci.rubini.us bot is hung or otherwise awol
06:03:58DefilerCan someone remind me what that tool is that will list the functions in a binary library?
06:06:32brixenon darwin, there's otool
06:06:42manveru leaves the room.
06:06:48brixenalso nm
06:07:38Defileraah, right. nm. that was it
06:10:24manveru enters the room.
06:11:26Defilerbrixen: Care to look at a header file with me?
06:11:43DefilerI am in /usr/include/openssl/hmac.h (Leopard)
06:12:15rueAlso known as Clearest Code On The Planet.h
06:12:48Defilerline 101 in my version is: unsigned char *HMAC(const EVP_MD *evp_md, const void *key, int key_len,
06:12:51Defilerblah blah
06:13:01rueWonder if there would be money in becoming a hitman for people who have to use some ghastly library
06:13:06Defilerbut that looks like a regular old function, and the decl here isn't conditional
06:13:24Defilernm doesn't show _HMAC as a function in libssl.dylib
06:13:33Defilerbut the other HMAC functions in this header are there, and they work
06:13:34brixenhmm, one sec..
06:14:54Defilergotta love the mixed tabs and spaces as well
06:15:14brixenheh, yeah
06:16:02DefilerMan, the license in this file is intense
06:16:40rueCould be a macro somewhere else, or the symbol may be manually removed/hidden
06:16:49rueAlso, you may have a version in /opt/local
06:16:52evanit's probably just #define'd to be something
06:17:11DefilerYou can implement something that was declared as a function in a header via a define?
06:17:16Defilerbrutal
06:17:27evanyou #define it before the declaration
06:17:36evanbecause cpp is a stupid text substitation program
06:17:41rueI see no #defines, though
06:17:42evanit doesn't even know it's a declaration.
06:17:44evanyeah
06:17:48evanit looks legit.
06:17:53evanlet me look with nm
06:18:06Defiler set_ffi_lib "libssl"
06:18:07Defiler attach_function "HMAC", "ossl_hmac", [:pointer, :pointer, :int, :string, :int, :string, :uint], :string
06:18:09rueDefiler: I am currently actually #defining connect(), send() etc. to mock them out
06:18:10Defileris what I have to wire it up
06:18:33evanah ha!
06:18:37evanthe wonders of openssl
06:18:39evanit's not in libssl
06:18:41evanit's in libcrypto
06:18:44Defilerwhat the fuck
06:18:46evanwhy oh why it's split
06:18:47evanno one knows.
06:18:51Defilerno what the fuck
06:19:04evanI hear ya
06:19:07rue'Cause it was more fun that way, maybe
06:19:07DefilerIt's wood chipper time for these kids
06:19:10evani'd have 2 modules
06:19:19DefilerYeah, it's cool. I have them split up already
06:19:21evanSSL and Crypto
06:19:23DefilerIt's just dumb
06:19:25rueHa, maybe there would be customers in that area.. :)
06:19:26evanagreed
06:19:41rueI have to go to ninja school first though
06:19:54evani *think* it's because crypto is the one thats crypto algos gathered up over time
06:19:59evanand libssl is ONLY the ssl parts
06:20:03evanor that was it originally
06:20:08evanit's not that anymore.
06:20:10rueIt is easier to use outside openssl, sure
06:20:12DefilerFFI::NotFoundError: Function 'HMAC' not found! (Looking in 'libcrypto')
06:20:27evanhrm
06:20:30rueWell, dunno easier but lighter-weight
06:20:32DefilerIt's probably due to those old irritating export restrictions
06:20:40Defilerthus making openssl easier to embed
06:20:49evanyeah
06:21:02evanthats why libtomcrypt rocks
06:21:05evandeveloped in canadian
06:21:07evanpublic domain
06:21:29Defilerso openssl/foo.h is routinely also #included by libcrypto as well as libssl. interesting
06:21:37rueIt is both in the /usr/lib and /opt/local/lib here
06:22:04evanDefiler: i got HMAC attached here
06:22:17brixenDefiler: otool -T -v /usr/lib/libcrypto.dylib | grep HMAC
06:22:18evan>> module Crypto
06:22:18evan>> set_ffi_lib "libcrypto"
06:22:18evan>> attach_function "HMAC", [], :pointer
06:22:18evan>> end
06:22:25evanworked
06:22:33Defileraha.. so maybe there really is a problem with > 6 args
06:22:40evanno
06:22:56TheVoice enters the room.
06:22:58evanwhen it's attached, the args do nothing
06:23:00Defileryep. boom.
06:23:02evanthats not until you do a call
06:23:09Defilerif I cut the , :uint off the end
06:23:10evanunless there is still some code to check
06:23:10Defilerit attaches
06:23:15evancheck ffi.rb
06:23:19evansomething must still be checking.
06:24:02evanhuh.
06:24:05evani get it here too...
06:24:25evancrap
06:24:32evanthere are still 6 arg length checks in th VM
06:24:33evanone sec.
06:24:43Defileroh, hah
06:24:48Defileryeah. it's in ffi.c
06:25:05Defilerand it returns Qnil so it looks like a lookup failure
06:25:30rueOn purpose, if my recollection serves :)
06:25:39evannot anymore
06:25:41brixenevan: are you able to restart the ci.rubini.us bot?
06:25:47evanthat was a gnu lightning problem
06:25:54evanbrixen: ack, i'll do that.
06:26:01brixenk
06:28:22evanDefiler: yeah, that fixes it
06:28:23evanone sec.
06:28:25evani'll commit.
06:29:18boyscout1 commit by Evan Phoenix
06:29:19boyscout * Remove final vestage of GNU lightning; 77aabb3
06:30:06evanwtf
06:30:28evanyou have the submodule connecting via ssh?
06:30:43brixenI changed that
06:31:03brixenI'm adding spec:anon and spec:committer like the git:anon, etc ones
06:31:13evanperfect.
06:31:25brixenand spec:init will depend on spec:clone which will use the anon url
06:35:37Defilerevan: super fixed. thanks.
06:38:14rubuildius_amd64Evan Phoenix: 77aabb391; 2107 files, 6857 examples, 25046 expectations, 0 failures, 0 errors
06:40:12lstoll leaves the room.
06:42:29mkrauskopf enters the room.
06:43:38DefilerHrm. There is no way to make a multi-line string that doesn't add additional whitespace from the second line, right?
06:44:13Defilere.g. "this is a long string \<cr><whitespace to align with above line>that should print on one line"
06:44:35DefilerThe MRI error being raised is this mouthful: TypeError: wrong argument (String)! (Expected kind of OpenSSL::Digest::Digest)
06:45:26evanDefiler: no, thats why heredocs always look ugly if you care about left side white space
06:47:51brixenDefiler: in such cases, I just do adjacent strings on separate lines with the \ continuation
06:48:36DefilerWhat do you mean by adjacent?
06:49:05Defilerhttp://rafb.net/p/ACBikj70.html
06:49:07Defilerthis is the scenario
06:49:10trythil enters the room.
06:49:15DefilerI need output identical to what that code generates
06:49:33Defiler(I don't actually care to match MRI's error, but I'm curious now)
06:50:42yugui leaves the room.
06:51:34benburkert enters the room.
06:53:03brixenDefiler: that works
06:53:11brixenthis is what I meant by adjacent: http://pastie.org/194739
06:54:11Defilerevan: So, now that I have these structs declared.. can I use them as attach_function return values?
06:54:19Defilerto auto-'box' them?
06:54:39Defilerbrixen: oh man I didn't know that worked
06:55:34benburkert leaves the room.
06:58:01benburkert enters the room.
06:58:01benburkert leaves the room.
06:58:58evanDefiler: no
06:59:03evanDefiler: you have to return pointers
06:59:12evanand give the pointer to a Struct.new
07:01:26Defilerk
07:03:26benburkert enters the room.
07:03:43benburkert leaves the room.
07:03:51boyscout1 commit by Brian Ford
07:03:52boyscout * Fix spec:clone and add spec:anon, spec:committer, spec:update, spec:push.; dd6d0e6
07:06:40Defilerhaha. rad. hexdigest is a two-line method now
07:10:05evanDefiler: hows merb coming?
07:10:14manveru leaves the room.
07:10:27Defilerevan: Merb doesn't work on webrick right now (even on MRI)
07:10:38Defilerezmobius was going to fix it last night, but I haven't heard from him
07:10:43evanand rubinius mongrel isn't working?
07:10:49evanwe should talk to tarcieri
07:10:53evani think he had it working
07:11:27boyscout1 commit by Wilson Bilkovich
07:11:28boyscout * Implemented OpenSSL::HMAC.hexdigest; 68b7fc3
07:11:34DefilerThat push right there is the last of OpenSSL that Merb appears to depend on
07:11:42brixenawesome
07:11:52evandope
07:12:03Defilerall that screwing around and HMAC is 25 lines included whitespace and ends
07:12:03evanstops micro managing.
07:12:15manveru enters the room.
07:12:39brixenevan: hehe, at this rate, I will have to eat my words about redoing stuff in FFI :)
07:12:55evanDefiler: hexdigest is 2 lines become you just moved everything to MemoryPointer
07:12:59evanDefiler: i'd prefer you didn't do that
07:13:02DefilerYeah, it's ridiculously easy once you get the hang of figuring out which things should be strings and which pointers, etc
07:13:07brixenwe'll probably end up rewriting all of ruby stdlib soon
07:13:09evanwhy does anyone else need a read_string_as_hex method on MemoryPointer?
07:13:11Defilerevan: Well yeah, but I needed it here as well
07:13:15rubuildius_amd64Brian Ford: dd6d0e60f; 2107 files, 6853 examples, 25042 expectations, 0 failures, 0 errors
07:13:38evanoh, you need it in digest and hmac?
07:13:40Defileryeah
07:13:43evanhrm.
07:13:48DefilerIt's really the right place for it, in my opinion
07:13:53Defilerbut I am open to other ideas
07:13:54evani don't think so
07:14:00evanit's so specific
07:14:14evanit's like having a method that returns the number 42
07:14:49DefilerOK. I'll move it. Where do you think it should go?
07:15:06evanOpenSSL.read_as_hex(ptr)
07:15:09evani'd think.
07:15:29DefilerOK
07:15:39evanwhy do you think thats the proper place for it?
07:15:52evani don't mean to tell you whats what
07:15:59evanjust giving my opinion
07:16:01DefilerBecause it's a MemoryPointer, and it has the responsibility to return various results based on how you ask to read it
07:16:23evanThen why not having Strin#as_shakespeare
07:16:41DefilerSure. You get to implement it, though
07:16:43evan"do this".as_shakespeare # => "perform doth"
07:16:54evani'm just say
07:17:03DefilerI mean, other than namespace issues, I don't really see a problem with that
07:17:08evanreading a string as hex out of a MemoryPointer is a very specific operation
07:17:17evanso much that i doubt anyone will need it
07:17:25evanother than the 2 cases you have
07:17:30DefilerOK. I guess it seemed pretty generic to me
07:17:39evanin all the FFI i've done
07:17:44evani've never need to read a string as hex
07:17:58evani've needed ever other method
07:18:13DefilerWhat code are we using to print those 0xfoo addresses in Object#inspect? Heh
07:18:14evanit seems very specific to the problem domain of OpenSSL
07:18:26evanobject_id.to_s(16)
07:18:30evanit's not a memory address.
07:18:42Defileryeah, they are too short. Heh
07:18:56DefilerI do like how this version reads, though
07:18:56DefilerOh well
07:19:21evanyou're not arguing very well
07:19:39evan:)
07:19:40DefilerIt's after 2
07:19:51DefilerI'm amazed I'm sitting up
07:20:07evanleave it then.
07:20:17DefilerNah, I've got it changed
07:20:29DefilerOpenSSL.read_buffer_as_hex
07:20:29evani can't have everything the way I want
07:20:30wycats__ leaves the room.
07:20:40DefilerI'll pick my battles
07:20:40wycats_ leaves the room.
07:21:23mkrauskopf leaves the room.
07:21:49evanrails has really gotten people in the mindset that the builtin classes are where they should embed their problem specific logic
07:21:54evanit's not something i've very fond of
07:22:36DefilerWell, getting back the kind of subclass you want can be tricky in typical Ruby code
07:22:58DefilerHere, for example, if I could easily get back a subclass of MemoryPointer that had my wackiness in it, I would do that
07:23:11rubuildius_amd64Wilson Bilkovich: 68b7fc3ba; 2107 files, 6857 examples, 25046 expectations, 0 failures, 0 errors
07:23:16DefilerHexStringPointer.new etc
07:23:16evanthats what extend is for
07:23:31evanptr = thing_that_returns_a_pointer()
07:23:42evanptr.extend WilsonMadness
07:23:49evanptr.hex_o_rama(12)
07:24:16Defilerptr.hex_this_shit_so_we_can_get_going
07:24:49evanDefiler: got anything going on this weekend?
07:25:02DefilerMight play some disc golf
07:25:08DefilerWith the last of the tastiest weather
07:25:46evannice
07:29:09DefilerOh. Hrm.
07:29:36DefilerYou know.. I should change this up
07:29:44Defilerbecause it makes sense to implement hexdigest in terms of digest
07:29:55DefilerI just did it backwards because Merb uses hexdigest and not digest
07:30:09Defilerbut in reality, hexdigest is just digest + convert_to_hex durh
07:33:02evanyep
07:33:08evani was going to ask you that.
07:33:18Defileroh god that is way less everything.
07:36:13boyscout1 commit by Wilson Bilkovich
07:36:14boyscout * Implement OpenSSL digest methods, and rewrite hexdigest using them; 3502540
07:36:29manveru leaves the room.
07:37:28Defilerbrixen: OK.. so.. new spec layout
07:37:35brixenyes
07:37:43Defilerbrixen: we now pass the OpenSSL specs that TM added earlier
07:37:55DefilerWhat do I do to tag them as passing for ci these days?
07:38:02brixenahh, hmm
07:38:17brixendid you try bin/mspec tag --del fails path/to/spec_file ?
07:38:27DefilerDefinitely not. :)
07:38:30brixenI haven't tried that since I moved spec/tags/ruby to spec/tags/frozen
07:39:11manveru enters the room.
07:39:34Defilerhttp://pastie.org/194748
07:39:50brixenhmm
07:39:58brixenle'me try here
07:40:01Defilerk
07:40:26brixenoh, hrm
07:40:39brixenI get giving it the spec/ruby path will confuse it
07:40:49brixendoesn't know where to look for the tags
07:40:54brixenuse spec/frozen instead
07:41:37brixenI'm going to have to write so many docs in the next couple days (ugh)
07:41:49Yurik enters the room.
07:42:02brixens/get/bet/
07:43:29brixenDefiler: http://pastie.org/194751
07:43:36DefilerSo if I want to run a subset of ci, I would say mspec ci spec/frozen/1.8/blah?
07:43:43brixenyes
07:44:53boyscout1 commit by Wilson Bilkovich
07:44:54boyscout * Remove failing tags from openssl/hmac specs; adba40f
07:46:17DefilerMan, VPRI is really kicking ass
07:46:39DefilerThis is about 130-ish lines of non-whitespace to support OpenSSL digests and hmac
07:46:45Defilerand they did javascript in 170
07:49:01brixenheh
07:50:08Yurik leaves the room.
07:54:21rubuildius_amd64Wilson Bilkovich: adba40f62; 2107 files, 6859 examples, 25050 expectations, 0 failures, 0 errors
07:54:22rubuildius_amd64Wilson Bilkovich: 350254024; 2107 files, 6857 examples, 25046 expectations, 0 failures, 0 errors
07:57:49brixenok, I've updated this thread with some info http://groups.google.com/group/rubinius-dev/browse_thread/thread/da19a433520afad5
07:58:07brixenthe channel topic has a link to the LH page as well
08:02:45dbussinkhmm, do changes to ENV propagate down to the vm too?
08:03:05dbussinkbecause ffi doesn't seem to pick up changes i make to LD_LIBRARY_PATH
08:03:16dbussinkit does work if i set the env variable in the terminal beforehand
08:03:22Defilermerb app running on rubinius muhaha
08:03:37DefilerI found and fixed the merb bug ezmobius promised to handle and now it is golden
08:04:23wycats enters the room.
08:05:27brixenDefiler: sweet!
08:06:24dbussinkevan: any ideas what could be the cause of the env changes not being picked up?
08:06:30brixendbussink: changes to ENV should be visible to C functions, since ENV is implemented with calls to the system
08:06:42dbussinkbrixen: yeah, that's what i thought
08:06:43brixener, visible to the VM
08:06:51dbussinkbut it doesn't seem to work :(
08:07:37brixenhmm, but is this something that would affect ld?
08:07:51dbussinkwell, ffi uses dlopen
08:08:29brixenyeah
08:09:44brixendbussink: you could put a call to getenv in the ffi code and see
08:09:57dbussinki was just doing that )
08:09:59dbussink:)
08:10:01brixenheh
08:10:15brixenI still wonder if dlopen might cache that or something
08:10:20brixenbut I really have no idea
08:10:24dbussinkit's set there
08:10:49dbussinkso that probably leaves something like a caching mechanism yeah
08:12:16dbussinkhttp://hathawaymix.org/Weblog/2004-12-30
08:12:17dbussinknasty
08:12:55manveru leaves the room.
08:13:06dbussinkthat makes it impossible to work around it
08:13:33dbussinkand forcing rubinius users to setup LD_LIBRARY_PATH is not something i would like to do
08:15:36brixenhrm, that sucks
08:16:11dbussinkyeah, really bigtime
08:16:26ruebrixen: I can rewrite whatever docs needed; new stuff I obviously may not be able to help with
08:16:41headius enters the room.
08:17:16manveru enters the room.
08:17:35ruedbussink: Library paths are a huge pain in the ass, and generally not worth it. If for whatever reason the lib cannot go in the system paths, require full path instead
08:18:07dbussinkrue: well, full path creates nastyness if it needs other libraries outside the search path
08:18:24dbussinkrue: this is a huge problem for example for all macports libraries
08:18:29rueYes, but that is generally a sign of something being wrong to begin with
08:18:40rueWhuh? Those should all BE in your ldpath
08:18:57Defilernope. macports doesn't update the LD path
08:18:57headiusevening
08:19:02dbussink/opt/local/lib is not standard
08:19:11dbussinkit compiles everything against it though
08:19:23dbussinkso that basically creates havok in our situation
08:20:04rueBe mindful of the difference between dlopen and lt_dlopen there
08:20:23dbussinklt_dlopen?
08:23:08brixenrue: thanks man. helping guide folks will be big help. I'll be out climbing for at least 1/2 the day tomorrow
08:23:14brixenheadius: what's your github username?
08:23:39headiusI haven't set one up...was halfway through the process when I had to run off to an event
08:25:06brixenheadius: check the link in the channel topic
08:26:24brixenrue: I think the docs need some pretty significant work to remove the rbx-centric stuff. I'll work on it tomorrow
08:26:52headiuslooks good at a glance
08:28:27brixenheadius: did you see http://rubyspec.org/ ?
08:28:35headiusI had not
08:28:48headiusvery nice
08:29:09brixenheadius: a fair amount of doc work needed, but I'll do the ruby-talk announce sat or sun
08:29:17brixenglad you like it :D
08:29:38headiuswe have a bunch of companies interested in sponsoring jruby hackfests at railsconf...I'm thinking we might try to steer it toward porting our various suites to specs
08:29:54brixenthat would be awesome
08:30:12brixenmust say, as much as I detest every single bug-tracker I've used, redmine is pretty nice
08:30:15headiusthat would probably be most useful, and nobody's going to re-write an optimizing compiler or something anyway
08:30:41brixenyeah, I was just looking over the mri openssl tests the other day
08:30:52brixenhaving all that stuff in specs would be awesome
08:30:58headiusredmine's pretty nice, we have a few people deploying on jruby now
08:31:06headiusyeah
08:31:07brixenoh sweet
08:31:17headiusI just want to start dumping our suites or pieces of them
08:31:25brixenyeah
08:31:34headiusthey're all governed by separate _index files now, so we can cull them one at a time
08:31:47brixencool
08:33:23brixenwell, must sleep. catch y'all late tomorrow
08:34:01kw leaves the room.
08:36:09trythil leaves the room.
08:36:18dbussinknight
08:40:40TheVoice leaves the room.
08:46:30ruebrixen: Announcements always on Monday or Tuesday EST working hours :) Otherwise they get eaten by the trickle traffic over the weekend
08:46:46manveru leaves the room.
08:48:35ruedbussink: GNU Libtool's ltdl is being used to load libraries. Most platforms have a separate dlopen implementation, but OS X for example also has its own libtool
08:48:35kamal_farizbrixen: just checking in. my github account is 'kamal'
08:49:27robertpostill enters the room.
08:49:28robertpostill leaves the room.
08:49:42dbussinkrue: yeah, was looking into it
08:49:59dbussinkbut it doesn't make any difference wrt the ld_library_path behavior
08:50:57manveru enters the room.
08:52:28rueDoes too :)
08:52:42dbussinkrue: does it?
08:52:47dbussinkrue: how?
08:52:48ruehttp://www.gnu.org/software/libtool/manual.html#Using-libltdl There should be functions there if it is really necessary to add stuff in
08:52:57GMFlash leaves the room.
08:54:19dbussinkrue: but we don't seem to be using libltdl everywhere
08:54:34dbussinknot on freebsd for example
08:55:11rueFreeBSD is the only exception currently, I think. In that case, one would need to fall back
08:55:28dbussinkbut is libltdl simply not supported there?
08:55:37dbussinkor is it juts not used
08:56:06dbussinkbecause i could use lt_dladdsearchdir to add paths
08:56:17dbussinkthat seems to be runtime possible
08:56:39dbussinkld_library_path changes are not runtime possible, so that would still not make it work for freebsd
08:58:15rueI would imagine LTDL_LIBRARY_PATH worked on FreeBSD, not sure though. There was some specific problem why it is split
08:59:59Maledictus enters the room.
09:00:22rue...Which no-one remembered to write down, I guess
09:00:26dbussinkwell, from what i see in the libltdl documentation, the only way to change it runtime is to use lt_dladdsearchdir and it's frieds
09:00:28dbussinkfriends
09:04:41ruelt_dlopen documentation, #2 says the env would work. Naturally they do not elaborate
09:05:00atmos leaves the room.
09:05:33dbussinkyeah, it says the env should work, but not that it's runtime changeable
09:05:49dbussink#1 has a specific remark that the program can change the search path itself
09:06:36headius leaves the room.
09:07:26dbussinkbut nothing like actually testing, it does seem to work :)
09:07:57dbussinkrue: but isn't the easiest solution for freebsd to libltdl working there?
09:11:55dbussinkrue: onto the next bug then
09:12:01dbussinkbut i got to run some errants first
09:19:02imajes enters the room.
09:22:13ruedbussink: I think the problem is that libltdl on FreeBSD is not able to attach to the same process
09:25:45manveru leaves the room.
09:28:14manveru enters the room.
09:38:28trythil enters the room.
09:43:45rueBe back in a few
09:45:51rueAlso, I think Merb running should get a bit more exposure--hurray for Defiler
09:55:10benny enters the room.
09:59:22blakewatters enters the room.
10:08:50jtoy leaves the room.
10:15:39jtoy enters the room.
10:26:25VVSizmornig, folks
10:26:41Yurik enters the room.
10:29:32jtoy leaves the room.
10:31:44yugui enters the room.
10:39:50trythil leaves the room.
10:39:58trythil enters the room.
10:40:25VVSizrue: do you know what else should I do to be able to commit to rubyspecs repo?
10:40:49VVSizcurrently, git push returns: fatal: The remote end hung up unexpectedly
10:45:42Skip enters the room.
10:48:17imajes leaves the room.
10:48:32VVSizoh, nevermind, figured that out. for some reason github (or browser) was appending mailto: to the URL :)
10:55:27trythil_ enters the room.
10:55:28trythil leaves the room.
11:08:15NoKarma enters the room.
11:09:37Blinchik_ leaves the room.
11:10:41Blinchik_ enters the room.
11:16:56trythil_ leaves the room.
11:34:11dbussinkVVSiz: nice work on the bigdecimal stuff :)
11:35:56lopex enters the room.
11:45:38VVSizdbussink: heheheh :) there's still *lots* to test there :)
11:45:48dbussinkVVSiz: hehe, probably yeah
11:45:55dbussinkthe first run of date also resulted in 200+ specs
11:46:13VVSizthe most evil method is #mode, it basically changes most other methods behaviours!!! :)
11:48:02dbussinkthat's really ultimate evil
11:49:13dbussinkrue: are you still there?
11:49:19VVSizso, effectively, we need to multiply the current tests by number of modes :)
11:49:37dbussinkrue: i just removed the freebsd ifdef conditional and it works fine on freebsd still
11:49:43dbussinkjust tested iconv
11:50:23dbussinkVVSiz: guess you assumed the default case now?
11:50:35VVSizyes, all current specs are for the default case
11:51:11VVSizplus, some test don't really honor the mode (like sqrt and others), some have hardcoded precisions and some have hardcoded algorithms on how to calculate the precision
11:51:19VVSiznothing of which is documented :)
11:51:40dbussinkman, you already have a ton of bugs to file against mri? :P
11:52:54VVSizI probably should, yeah. also, all methods that accept additional precision parameter, have in their docs specified that the parameter is optional. Which is not true, it's required :)
11:53:11VVSizalso, #zero?, according to the docs, must return 0 for 0, but returns 1 :)
11:53:28dbussinkrue: hmm, not true though, indeed failures for the current process
11:57:40VVSizanyone knows what the project is used for those nice project's web pages for mspec and rubyspecs? http://rubyspec.org/
11:57:52dbussinkVVSiz: you mean redmine?
11:58:08VVSizoh, it seems so.
11:58:28VVSiznot that bad at all.. :)
11:59:01dbussinklooks pretty nice
11:59:04dbussinknicer than trac :P
11:59:46VVSizI've been using Trac for years, so Redmine looks comfortable and kinda known
12:22:17yugui_ enters the room.
12:23:01yugui leaves the room.
12:24:02dbussinkrue: i seem to be able to fix it if i compile it with RTLD_NOW instead of the default RTLD_LAZY
12:52:24Fullmoon leaves the room.
12:55:03rubuildius_ppc enters the room.
13:00:48chris2 enters the room.
13:01:26blakewatters leaves the room.
13:04:25jeremydurham leaves the room.
13:10:36qwert666 enters the room.
13:14:32rubuildius_ppcWilson Bilkovich: adba40f62; 2107 files, 6861 examples, 25076 expectations, 0 failures, 0 errors
13:18:59jeremydurham enters the room.
14:26:06ttmrichter leaves the room.
14:38:23seydar enters the room.
14:54:28seydar leaves the room.
14:57:44seydar enters the room.
15:06:39kw enters the room.
15:07:33seydar leaves the room.
15:31:46ShayArnett enters the room.
15:44:06rubuildius_ppc leaves the room.
15:44:39rubuildius_ppc enters the room.
15:45:31trythil enters the room.
15:49:31GMFlash enters the room.
16:06:50GMFlash leaves the room.
16:11:04jeremydurham leaves the room.
16:12:12chillywilly enters the room.
16:26:35seydar enters the room.
16:32:18trythil leaves the room.
16:44:47yugui_ leaves the room.
16:56:54stepheneb enters the room.
17:03:15anteaya enters the room.
17:09:25imajes enters the room.
17:09:36kw leaves the room.
17:10:21anteaya leaves the room.
17:11:25obiejuan enters the room.
17:12:24benburkert enters the room.
17:19:27febeling enters the room.
17:23:52seydar leaves the room.
17:31:52enebo enters the room.
17:45:23obiejuan leaves the room.
17:45:27Yurik leaves the room.
17:45:55Yurik enters the room.
17:50:17manveru leaves the room.
17:53:28manveru enters the room.
17:55:02BluR enters the room.
18:04:36NoKarma leaves the room.
18:05:07NoKarma enters the room.
18:07:13rueHola
18:07:30VVSizbrixen: I've filed first three tickets against rubyspec and mspec projects, #1, #2 and #3 :)
18:07:37VVSizrue: hola indeed!
18:09:44rueHehe
18:09:55ruedbussink: So long as it works, cool!
18:10:08dbussinkrue: still fighting with it
18:10:14dbussinkrue: didn't fix all cases
18:10:50rueAh, perhaps not then.
18:11:59dbussinkrue: there is however an init for libltdl missing in library.c
18:12:03xmlhacker enters the room.
18:12:07dbussinkat least, it looks like that
18:12:57yaroslav enters the room.
18:13:26qwert666_ enters the room.
18:16:34brixenVVSiz: sweet :)
18:17:23ruedbussink: Should be in setup.c
18:17:30brixenVVSiz: I'll look into the standalone stuff today. I knew there would be issues with that
18:17:31dbussinkrue: looks like it works now :)
18:18:36dbussinkrue: the missing init is the key
18:18:44dbussinksurprised it works on other platforms though
18:19:21rueshotgun/lib/subtend/setup.c:37: ffi_dlinit();
18:19:44rueSo the others should be starting up there
18:20:06dbussinkrue: http://pastie.caboo.se/194849
18:20:15dbussinkrue: that's my current diff
18:20:16dbussinkrue: it works with this
18:21:40benburkert leaves the room.
18:22:27benburkert enters the room.
18:22:39dbussinkrue: weird stuff
18:23:08dbussinkrue: because it i don't put in the init there, it doesn't properly set it all up
18:24:40ruedbussink: Use ffic.
18:24:43rueErm, ffi.c
18:24:59dbussinkinstead of library.c you mean?
18:25:03rueYes
18:25:15dbussinkyeah, could refactor something there
18:25:17rueIf it works, excellent. Is this 7.0?
18:25:29dbussinki have a 6.3 vm here
18:26:54rueCould be one of the earlier versions with the problem. Have to special-case that if it comes up. But use ffi.h and ffi.c for anything FFI-related. The init can go in one of the bootstrap files
18:30:55qwert666 leaves the room.
18:32:02wycats_ enters the room.
18:33:00enebo leaves the room.
18:36:36rueLooks like Init_ffi is hooked in now too, so can just use that
18:45:38dbussinkhmm, do we link those subtend libraries with export-dynamic?
18:45:56VVSizah, it's a bit unfortunate that the whole rubyspecs history is not preserved...
18:46:34djwhittooo... rubyspecs.org
18:46:36djwhittlooks nice
18:46:41VVSizyeah
18:47:00djwhittbrixen: nice job on splitting the specs out!
18:47:13djwhittis super impressed nothing exploded
18:47:46djwhittcremes, cremes_: so, looks like you got the bot going alright?
18:48:37wycats leaves the room.
18:50:20dbussinkrue: hmm, there is a new release of libtool
18:52:24kw enters the room.
18:54:50benburkert leaves the room.
18:56:55chris2 leaves the room.
18:58:31kw leaves the room.
18:59:34rueVVSiz: About 2600 commits in there :) I will push a revision dump
19:02:35brixendjwhitt: thanks!
19:02:43brixenfinally packed, see y'all in a few hours
19:08:26rueYou are coming over?
19:08:32rueI need to clean up..
19:29:35rueAway a bit.
19:39:53BluR leaves the room.
19:51:13benny leaves the room.
20:18:00yaroslav leaves the room.
20:33:24wycats_ leaves the room.
20:33:42BluR enters the room.
20:36:55wycats enters the room.
20:39:22rueBack-ish
20:39:58imajes leaves the room.
20:40:31trythil enters the room.
20:42:58ezmobius enters the room.
20:44:36wycats_ enters the room.
20:47:55rueMhehee.. let us see how long I can subscribe to the Github news feed
20:50:11ezmobiuswaves
20:50:32evanallo!
20:50:39evani'm off to lunch
20:50:49evannow that my comment feed problem has been solved :)
20:51:11dbussinkevan: any suggestions whether getting libltdl working on freebsd?
20:51:17ezmobiushah merb-core + merb-more + merb-plugins + dm-core + dm-more all together is less lines of code the ActiveSupprt alone :P
20:51:35evandbussink: i don't think you need it
20:51:37djwhittezmobius: how many lines is it?
20:51:51evandbussink: on freebsd, it doesn't use it
20:52:01dbussinkevan: well, dlopen doesn't support changing ld_library_path on the fly
20:52:02ezmobiusactivesupport is 28k lines
20:52:09dbussinkevan: libltdl does
20:52:14evanon freebsd it doesn't?
20:52:22dbussinkit doesn't anywhere
20:52:26evansure it does
20:52:30evanoh oh
20:52:31evanchanging.
20:52:35dbussinkyou can change ld_library_path after a process started
20:52:36evanhm.
20:52:38dbussinkcan't
20:52:53dbussinkbut you can with libltdl with another variable
20:52:55dbussinki got that working
20:52:58evanwell, perhaps we'll have to handle that at startup
20:53:07evanchange the path and re-exec
20:53:18dbussinkwell, dunno what the easier solution is
20:53:25evanhm, ok.
20:53:34evanlibltdl had problems on FreeBSD
20:53:43evanbecause it wouldn't find symbols in the current process
20:53:49evanbecause it doesn't pass in the correct flags
20:53:56evanand i really didn't want to debug libltdl
20:53:58evanit's sooo ugly
20:54:01evananyway, i have to run.
20:54:01dbussinkyeah
20:54:06dbussinki've digging into it
20:54:09evandbussink: lets talk later.
20:54:12dbussinkbut no succes so far
20:54:12dbussinkok
20:56:16ruedbussink: I spent quite a bit of time on it (in September, I think) but with the various issues it is a bit of a pain
20:56:24ruedbussink: What are you running into?
20:56:38dbussinkrue: well, it's the current process attaching again
20:56:43rueYeah
20:57:00dbussinkkinda hit the same wall as you and evan did :P
20:57:27rueMind you, I do not think it would be terrible to just say that the library path must be set before running or something
20:57:49dbussinkwell, i think it should be easy user configurable
20:58:06dbussinknot force people into setting up LD_LIBRARY_PATH stuff etc.
20:58:11rueAfter all, if you at runtime enter the paths, you might just as well make absolute paths
20:58:41dbussinkwhether it's through libltdl or by a re-exec doesn't really matter that much to me
20:58:46rueNo-one should ever need to touch LD_anything. If MacPorts is doing that, they are seriously fucked up
20:59:14dbussinkmacports does all that stuff compile time, they put it in the port definitions
21:00:03rueI do not understand why they would not add /opt/local/lib to libpath
21:00:29dbussinkdunno, maybe they don't want to change behavior too much or something
21:00:48dbussinkor they fear certain collisions with certain libraries
21:00:55wycats leaves the room.
21:01:19rueI suspect it is because they are stupid :)
21:01:38rueIn all seriousness though, sure, there could be a legitimate issue
21:02:35MenTaLguY enters the room.
21:03:03ruedbussink: But I look at it from the point of view of what is a reasonable effort. If MacPorts says, indeed, that any libs downloaded through there only work with other MacPorts stuff, then that is exactly what the user would want to do: get Rubinius through MacPorts (or just symlink the libraries to /usr/lib or set LD_)
21:03:16vvs enters the room.
21:03:25anteaya enters the room.
21:03:32rueAssuming, naturally, that there is no easy way of fixing the problem otherwise and I suspect there is not
21:03:40MenTaLguYhello
21:03:45rueHey, MenTaLguY
21:04:10dbussinkrue: but in that case they would put the /opt/local/lib in the config option we try to create
21:04:31dbussinkrue: but you're more in favor of the re-exec option?
21:26:40kw enters the room.
21:26:53ezmobiusDefiler: pingage
21:28:10BluR leaves the room.
21:42:02EugZol leaves the room.
21:45:57kw leaves the room.
21:49:51mernen enters the room.
21:50:57febeling leaves the room.
21:52:55trythil leaves the room.
21:54:09trythil enters the room.
22:00:57ShayArnett leaves the room.
22:04:35FoobarWidget enters the room.
22:05:57yipstar enters the room.
22:09:50imajes enters the room.
22:11:50xhanjian_ enters the room.
22:12:15trythil leaves the room.
22:13:27ruedbussink: I am most in favour of MacPorts getting their brains from the storage
22:13:31kw enters the room.
22:13:37wycatsno way: http://spreadsheets.google.com/pub?key=pCXpD9gyYzbW5IuxGT-aSaw
22:13:45dbussinkrue: hehe, but there is more than just macports :)
22:13:54dbussinkrue: os x mysql installer to name something
22:14:11imajes leaves the room.
22:15:01rueBut we can change the path on OS X so that works fine. FreeBSD is somewhat saner in that respect. Just make them use the env, though I suppose you could exec
22:15:43