Index

Show enters and exits. Hide enters and exits.

00:00:17rubuildius_amd64Adam Gardiner: 60a6b2003; 1788 files, 6220 examples, 22038 expectations, 0 failures, 0 errors; http://rafb.net/p/SeCufq19.html
00:00:17rubuildius_amd64Brian Ford: e5480fa09; 1788 files, 6220 examples, 22038 expectations, 0 failures, 0 errors; http://rafb.net/p/mmv15l33.html
00:02:07pluskid leaves the room.
00:02:30pluskid enters the room.
00:07:20rubuildius_ppcAdam Gardiner: 60a6b2003; 1788 files, 6223 examples, 22067 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/171297
00:07:21rubuildius_ppcBrian Ford: e5480fa09; 1788 files, 6223 examples, 22067 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/171293
00:07:22rubuildius_ppcAdam Gardiner: ef616fe2c; 1787 files, 6217 examples, 22057 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/171290
00:07:33qwert666 leaves the room.
00:10:39Arjen_ leaves the room.
00:12:43cored enters the room.
00:13:00pluski1 enters the room.
00:13:19crafterm enters the room.
00:13:45pluskid leaves the room.
00:13:58srbaker leaves the room.
00:14:37The_Linux_Lich enters the room.
00:16:29djwhitt enters the room.
00:19:12djwhittbtw, discussion of fonts earlier prompted me to try a few. conslusion: sgi screen font is really nice if you're on linux
00:21:58evannever seen that one, i don't think.
00:24:03FoobarWidget leaves the room.
00:24:45eventualbuddha leaves the room.
00:26:30d2dchat enters the room.
00:27:42macournoyer enters the room.
00:29:39pluskid enters the room.
00:29:56pluskid leaves the room.
00:30:58pluskid enters the room.
00:36:19brainopia leaves the room.
00:36:57dewd leaves the room.
00:39:40ruedjwhitt: That is not bad
00:42:57rueTerrible at <11 but 11, 12 are very nice and crisp
00:45:10rueI will give it a try for a bit
00:46:25wdperson enters the room.
00:48:59womble leaves the room.
00:50:16imajes_ leaves the room.
00:54:22mae enters the room.
00:54:46The_Linux_Lich leaves the room.
00:57:30srbaker enters the room.
00:58:01wdperson leaves the room.
01:10:41mae leaves the room.
01:12:03aotearoa enters the room.
01:18:20imajes_ enters the room.
01:19:24benburkert leaves the room.
01:20:29mark___ enters the room.
01:21:01benburkert enters the room.
01:22:35ezmobius leaves the room.
01:24:06benburkert leaves the room.
01:26:04imajes_ leaves the room.
01:30:16AndrewO enters the room.
01:30:46d2dchat leaves the room.
01:31:49srbaker leaves the room.
01:34:57binary42_ leaves the room.
01:37:35antares leaves the room.
01:40:42nicksieger leaves the room.
01:41:50cored leaves the room.
01:42:39srbaker enters the room.
02:09:18dkubb enters the room.
02:09:29mae enters the room.
02:12:24kw_ leaves the room.
02:16:16binary42 enters the room.
02:22:06mae_ enters the room.
02:22:25mae_ leaves the room.
02:22:58mae_ enters the room.
02:24:16VVSiz_ enters the room.
02:25:54KirinDave leaves the room.
02:27:19dkubb leaves the room.
02:28:52_martinS_ leaves the room.
02:29:48aasmith enters the room.
02:30:03dewd enters the room.
02:32:05d2dchat enters the room.
02:33:43dkubb enters the room.
02:34:17lopex leaves the room.
02:41:10kw_ enters the room.
02:41:25dkubb leaves the room.
02:42:23chop3 enters the room.
02:42:26VVSiz leaves the room.
02:45:11mae leaves the room.
02:45:34radarek leaves the room.
03:02:13agile enters the room.
03:05:07mae enters the room.
03:10:53webmatHas anybody tried the rake install task since Ryan's commit to the Rakefile and stuff? (commit 5dee7377f21dad0aecef82e8f6aa3be9dbd19d28)
03:11:42webmatI can't seem to install from edge tonight
03:12:13d2dchat leaves the room.
03:13:01agardinermay not help, but i never do an install... just run shotgun/rubinius
03:13:06benburkert enters the room.
03:13:36ruewebmat: Do you mean it worked after that commit but not currently?
03:14:17webmatthe reverse. the 24th's tarball was ok but it does not work anymore
03:15:54KirinDave enters the room.
03:17:23macournoyer leaves the room.
03:19:15webmatAnd by the way I haven't investigated this in detail, but now the build task seems to depend on the directory being a git repo. Which means it fails when run from a daily tarball.
03:20:15webmatIf I remember correctly, I had installed from the 24th's tarball without such a problem
03:20:54mae_ leaves the room.
03:21:05cyndis_ leaves the room.
03:23:38cyndis_ enters the room.
03:28:04ruewebmat: Ah, that is quite possible
03:28:24ruePaging Dr. zenspider
03:31:53AndrewO_ enters the room.
03:33:34webmatfollowing agardiner's comment, I'm considering using a small script named 'rbx' to just point to my edge directory's shotgun/rubinius, but then the uninstall task fails as well :-)
03:34:07webmatthe problem seems to be that some stuff in ENV is never set. therefore the task complains about having a nil instead of a string
03:34:08rueHeh, it would. We should try to resolve it anyway to avoid issues with lib locations and so on
03:36:53KirinDave leaves the room.
03:37:23KirinDave enters the room.
03:39:54evanwebmat: please open a ticket
03:39:59evanso zenspider can fix it
03:40:03webmatyup
03:40:04evaninclude as much info as possible
03:41:08rby enters the room.
03:42:15fbuilesv enters the room.
03:43:25benburkert_ enters the room.
03:43:25benburkert leaves the room.
03:48:00alexvollmer enters the room.
03:51:07KirinDave leaves the room.
03:52:13wycats enters the room.
03:54:49alexvollmer leaves the room.
04:01:17binary42 leaves the room.
04:03:20ttmrichter enters the room.
04:16:09webmatcreated ticket 439, don't hesitate to ask for more info.
04:16:20webmateyes ... burning ...
04:16:24webmatneeds sleep
04:17:55webmat leaves the room.
04:37:22AndrewO_ leaves the room.
04:48:44mark___ leaves the room.
04:49:32mark___ enters the room.
04:55:44yipstar leaves the room.
05:02:40mark___ leaves the room.
05:17:22ezmobius enters the room.
05:23:44fbuilesv leaves the room.
05:35:43mediogre enters the room.
05:44:58jicksta leaves the room.
05:47:07boyscout2 commits by Brian Ford
05:47:08boyscout * Rework and cleanup of various String methods.; 3145a74
05:47:09boyscout * Shuffle some String methods. Add specs for and rework String#substring.; 9ba3e51
05:55:04rubuildius_amd64Brian Ford: 3145a74a8; 1789 files, 6230 examples, 22062 expectations, 0 failures, 0 errors; http://rafb.net/p/qD7rx197.html
05:55:35tizianobis enters the room.
06:01:58rubuildius_ppcBrian Ford: 3145a74a8; 1789 files, 6233 examples, 22091 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/171408
06:03:53agile leaves the room.
06:04:16tizianobis_ enters the room.
06:20:19tizianobis leaves the room.
06:29:24tizianobis_ leaves the room.
06:37:27crafterm leaves the room.
06:39:17hassox enters the room.
06:40:26sdeming enters the room.
06:41:18dewd leaves the room.
06:48:55agardiner leaves the room.
06:53:55jartz enters the room.
07:10:01AndrewO leaves the room.
07:35:37rby leaves the room.
07:39:38yaroslav enters the room.
07:42:56ezmobius leaves the room.
07:50:42den1jay enters the room.
07:51:28TheVoice leaves the room.
07:56:50jartz leaves the room.
07:58:04benburkert_ leaves the room.
08:09:43den1jayhi
08:09:48den1jayanyone awake?
08:11:04GMFlash leaves the room.
08:11:13GMFlash enters the room.
08:13:14rueYep, aloha
08:14:03yaroslav leaves the room.
08:14:27den1jay:)
08:16:01kw_ leaves the room.
08:17:16den1jay leaves the room.
08:18:15jinjing enters the room.
08:22:51den1jay enters the room.
08:25:34srbaker leaves the room.
08:25:35srbaker_ enters the room.
08:25:53den1jayanyone currently active developing rubinius?
08:27:35thehcdreamer enters the room.
08:32:32rueMost of us, yep
08:32:51rueUnless you mean right this minute? In that case most probably will use some lame excuse like being asleep
08:33:11rueI am, incidentally, just writing some FFI specs
08:34:49den1jayHi rue
08:34:54den1jaythx for the quick message
08:35:39den1jaycan you plz provide me some tips or direct me to some urls where i can find info on how can I start developing rubinius in shortest time possible?
08:39:11den1jayrue: u there?
08:40:13rueI would recommend starting with the wiki-ish pages found on http://rubinius.lighthouseapp.com
08:41:05rueThe developer readme has some useful information, it is in the source or you can read it in the repository browser: http://git.rubini.us/?p=code;a=blob_plain;f=README-DEVELOPERS;hb=HEAD
08:42:13rueLargely, though, the specific course will depend on what in particular you want to do? You can write specs or write/fix core and library code or try to get a specific library or application to run or you can work on the VM.
08:42:34rueAmong other things. Is there anything that you are interested in doing, specifically?
08:44:18den1jayi am just thinking of porting builder library to rubinius
08:45:37zenspider leaves the room.
08:47:49rueCool
08:49:07rueThe typical process is 1. Try to run the library or its tests; 2. if Rubinius is broken, write a spec for the correct behaviour; 3. fix Rubinius; 4. repeat until library runs flawlessly
08:50:03rueOf course some problems could be things that you cannot fix: in that case it is of course perfectly fine to just omit step 3., someone else will work on fixing it
08:50:08contextden1jay. the wiki has a page for starting developing rubinius
08:50:15rueWriting specs for the broken behaviour is important though.
08:50:23contextden1jay. with links to books/blog entries to help you on your way of understanding how rubinius works
08:50:50den1jayok
08:51:01den1jaythx rue and context
08:51:12den1jayI am reading wiki and help doc now
08:56:20hassox leaves the room.
08:59:25qwert666 enters the room.
09:00:33mutle enters the room.
09:05:17jartz enters the room.
09:10:19Arjen_ enters the room.
09:13:23octopod enters the room.
09:20:53ruecontext: It does? I did not realize that
09:21:55dkubb enters the room.
09:29:31dkubb leaves the room.
09:33:55jartz leaves the room.
09:34:03jartz enters the room.
09:41:25antares enters the room.
09:42:17hassox enters the room.
09:47:13antares leaves the room.
09:48:49den1jay leaves the room.
09:51:34Skip enters the room.
09:58:56BlackEdder enters the room.
10:08:12jinjing leaves the room.
10:08:27jinjing enters the room.
10:11:59den1jay enters the room.
10:39:18aotearoa leaves the room.
10:44:32flori enters the room.
10:51:28loop_ leaves the room.
10:55:46pluskid leaves the room.
10:58:57tizianobis_ enters the room.
11:00:37Fullmoon leaves the room.
11:01:02webmat enters the room.
11:11:44tizianobis__ enters the room.
11:14:11tizianobis__ leaves the room.
11:23:19tizianobis leaves the room.
11:27:50Cu9YpD leaves the room.
11:39:20wdperson enters the room.
11:43:42ctennis enters the room.
11:52:07srbaker_ leaves the room.
11:52:08srbaker enters the room.
11:57:37radarek enters the room.
12:13:31chop3 leaves the room.
12:13:43fbuilesv enters the room.
12:18:32imajes_ enters the room.
12:19:32imajes_ leaves the room.
12:48:57mediogre leaves the room.
12:48:59cyndis_ leaves the room.
13:03:35BlackEdder leaves the room.
13:11:38AndrewO enters the room.
13:18:33dkubb enters the room.
13:28:17den1jay leaves the room.
13:31:48d2dchat enters the room.
13:32:17probablycorey enters the room.
13:34:00ruedjwhitt: Good find! I really like this font
13:34:22rueExcept for some reason MacVim's line spacing is incorrect
13:35:32fbuilesv leaves the room.
13:35:44wmoxam enters the room.
13:35:57rueJust single quotes too, which is weird
13:39:55fbuilesv enters the room.
13:46:07nemerle_afk enters the room.
13:54:31brainopia enters the room.
13:57:40agile enters the room.
13:59:22mae leaves the room.
14:03:31nemerle leaves the room.
14:05:21moofbong enters the room.
14:06:22den1jay enters the room.
14:07:57rueHm. I am getting conflicting info on .dylib vs. .bundle
14:11:42scooprI somehow remember the difference being that dylibs are referenced from mach-o headers and .bundles are more like plugins that are runtime loadable, I could be wrong though
14:13:13octopod_ enters the room.
14:14:09rype enters the room.
14:14:14octopod leaves the room.
14:14:53gnufied enters the room.
14:15:23rueYeah, it seems that .dylib are "statically linked shared libraries"
14:15:34gnufiedhi folks
14:15:40rueAnd .bundle are "dynamically linked shared libraries"
14:15:43rueHola
14:17:32skaar enters the room.
14:20:52djwhittrue: cool, glad you like it
14:21:37jney leaves the room.
14:24:01nemerle enters the room.
14:25:09pate_ enters the room.
14:26:46pateif you're headed to MWRC today ... pack a coat/jacket
14:26:55pateit's started snowing again. :)
14:30:04rueHeh
14:35:07nemerle_afk leaves the room.
14:38:21rueI wonder if this is why drbrain disabled externals
14:42:53BlackEdder enters the room.
14:43:05BlackEdder leaves the room.
14:45:07nicksieger enters the room.
14:46:28BlackEdder enters the room.
14:47:12evanpate_: will do
14:48:19rueYou are up early :P
14:49:19evanflying to SLC today
14:49:30evanhave to take fog to the kitty hotel first
14:49:43anonyMoin
14:49:45rueevan: Quick--ltdl can load either .dylib or .bundle files. We have it hardcoded to .bundle. This may be a problem because it is likely that most of the /usr/lib stuff are in fact .dylibs. This puts Mac at a disadvantage with the .so platforms
14:50:26rueevan: I tried a kitty hotel once. It was ugly
14:50:36rueHello, anony
14:51:10evanyou can't dynamically load .dylibs
14:51:13evanso no
14:51:16evanltdl can't load them
14:52:03rueI just did
14:52:04anonyevan, can't you?
14:52:11anonywhy can't you, rather?
14:52:22evanthe docs clearly say that dylib's not not meant to be used with dlopen()
14:53:08rueI think it goes through dyld
14:53:49anonyevan, what docs are you looking at?
14:54:12evancheck google for dylib versus bundle
14:54:33olabini leaves the room.
14:54:35anonyI looked at the developer docs and it says nothing in there, in fact it uses the dylib path for loading.
14:55:11evanhm
14:55:21evanperhaps i'm thinking legacy
14:55:28evanremember, i started with OS X in 10.0 days
14:55:28anonydlopen
14:55:29anonyLoads and links a dynamic library or bundle
14:55:34evanthere was a big difference back then
14:55:37evanperhaps there isn't anymore
14:55:39anony:)
14:55:50anonyTrust the source, not the nettertubes!
14:55:52evani vividly recall dlopen() refusing on .dylib's at one point
14:58:09anonyThat's probably true, I've only been doing OS X programming since 10.2ish
14:58:17evanhm. perhaps it was that you can't link against a .bundle
14:58:27rueI was reading our libltdl, looks like it is OK.
14:58:51rueIt just creates a bit of an issue because we need to support both extensions for FFI, I guess
14:59:20evanah ah!
14:59:21evani see
14:59:22evanok
14:59:30evanit has to do with how dyld handles it's symbols
14:59:34evana .bundle can be unloaded
14:59:36evana .dylib can not.
14:59:43rueYeah
15:00:08evanthe symbols tables are set up differently, because of the different purposes
15:00:32evana .bundle is simpler, looks like it just have a simple symbol table with minimual info
15:00:40rueI think I will enable both for now, need to split FFI loading from subtend_find_symbol() though.
15:03:38evanok
15:03:41evango for it
15:03:44evani'm off to the kitty hotel
15:04:05rueHave fun! Give Fog a scratch for me :D
15:04:13evanwill do!
15:06:18agile leaves the room.
15:11:28imajes_ enters the room.
15:15:03srbaker leaves the room.
15:15:10srbaker enters the room.
15:20:22cyndis enters the room.
15:26:20BlackEdder enters the room.
15:26:22imajes leaves the room.
15:28:43wycats leaves the room.
15:39:51den1jay leaves the room.
15:53:56imajes enters the room.
15:54:52srbaker leaves the room.
15:55:33srbaker enters the room.
15:55:37yipstar enters the room.
15:55:53srbaker leaves the room.
15:56:25srbaker enters the room.
15:57:15srbaker leaves the room.
15:58:32srbaker enters the room.
16:02:45srbaker leaves the room.
16:08:14imajes_ leaves the room.
16:08:43fbuilesv leaves the room.
16:22:59agile enters the room.
16:24:26Arjen_ leaves the room.
16:25:10tobyo enters the room.
16:31:59wycats enters the room.
16:51:07w1rele55 enters the room.
16:52:06wycats leaves the room.
16:56:53JimMc leaves the room.
16:57:57dewd enters the room.
17:03:16mutle leaves the room.
17:12:44dgtized enters the room.
17:15:02gnufied leaves the room.
17:15:48srbaker enters the room.
17:19:45dgtizedso are all of these string changes being benchmarked to ensure they actually speed things up?
17:20:20dgtizedthey look like they are good, changes, but I'm not noticing any speedup
17:20:43dgtizedin fact it seems like the bin/mspec ci is taking steadily longer
17:20:56rueI dunno, that brixen is a sneaky fellow
17:21:52djwhittdgtized: you're not noticing any speed up when doing what?
17:22:03dgtizedI mean if they are just intended as a cleanup that results in temporarily slower performance prior to some really nice optimization then I'm all for it
17:22:08dgtizedbin/mspec ci
17:22:10djwhittah, ok
17:22:20djwhittI think if you run the string benchmarks you'll find that they're a lot faster
17:22:24dgtizedit's taking about a minute to run now
17:22:35dgtizedright but bin/mspec ci is kind of a nice real world usage
17:23:20dgtizedso if the spot optimizations are causing a speedup in specific methods that's good, but that doesn't necessarily mean that it corresponds to overall performance
17:23:43djwhitttrue
17:24:34KirinDave enters the room.
17:26:29TheVoice enters the room.
17:28:15imajes leaves the room.
17:29:45DefilerBenchmarks in the ticket show a 10x speed improvement
17:29:56DefilerI doubt our bottleneck in ci has much to do with strings
17:30:21djwhittI think ci is actually fairly io bound
17:30:32djwhittwriting all those dots ;)
17:30:36Defilerhah
17:30:44antares_ enters the room.
17:37:27cremesdjwhitt: those dots incur about a 20% overhead on both my ppc mac and intel mac
17:37:54cremesyou guys should compare execution times between default "bin/mspec ci" and "bin/mspec ci -fm"
17:38:06cremes-fm suppresses the dots and gains back lots o' speed
17:38:37imajes enters the room.
17:38:53probablycorey leaves the room.
17:38:57benburkert enters the room.
17:39:25probablycorey enters the room.
17:44:22djwhittyeah, I do that on my rubuildius
17:44:25djwhittspeeds things up a lot
17:47:19ezmobius enters the room.
17:49:04dbussinkwondering about how much time is spend in File.stat, anyone ever looked at that?
17:50:16srbaker leaves the room.
17:50:28srbaker enters the room.
17:53:41scooprdamn
17:54:32Defilerdbussink: Yes. Most of the time =(
17:54:57evanfog is happily at her kitty condo, and now i'm waiting at the airport
17:55:51benburkert_ enters the room.
17:56:08dgtizedis File.stat just slow because it's FFI instead of primitive or something?
17:56:08scooprso, theres no one-command-way of building a statically linked shotgun currently, and -pg profiling doesn't play nice profiling dynamic libs (eg. librubinius-*.so), and LD_PROFILE doesn't seem to work.
17:56:38evandgtized: no, it's using a primitive now anyway
17:56:49evani'm pretty sure it is at least
17:56:56dgtizedevan: doesn't look like it
17:57:02dgtizedit's using a pointer to FFI::Struct
17:57:02evanthough, i could have gotten lost in the changes to it
17:57:38ezmobius leaves the room.
17:58:16evanthat should be fastish
18:00:15evanbenchmarks need to be written to verify it
18:00:18dgtizedwe are a good 3 times slower then MRI on that
18:00:39dgtizedrunning bm_stat anyway
18:00:49gnufied enters the room.
18:01:06evanwhere is bm_stat?
18:01:15dgtizedbenchmarks/rubinius/bm_stat
18:01:41evanwhy the hell is it compiling a file...
18:01:53dgtizedevan: good question
18:02:26GMFlash leaves the room.
18:02:34evanmaybe at one point, this benchmark compared our stat to a raw C stat
18:02:38evanusing that aux program
18:02:47evanuse -p when running bm_stat
18:02:58dgtizedlooks like it from the C program
18:03:12evanoh, it still does
18:03:14evanat the bottom
18:03:31evanthats nice actually
18:04:21brixendjwhitt: http://pastie.org/private/3yx05lijjdh4txyfluba
18:04:31brixendjwhitt: and see #359
18:04:50brixenI've compared -fd with -fm before and have seen little or no speed difference
18:04:54brixenI'll try it right now
18:04:57djwhittbrixen: did you mean that for dgtized?
18:05:08brixendjwhitt: heh, yeah, sorry
18:05:16brixentab completion
18:05:21brixendgtized: see above
18:05:49dbussinkthe file stat benchmark shows huge diffeerence
18:05:53GMFlash enters the room.
18:06:14dbussinkmaybe it's beneficial to convert it to a primitive for now, because it's so heavily used (every rbc / rb file that is hit)
18:06:38dgtizedbrixen: yea I see -- I saw your benchmarks on your ticket afterwards, that does look good -- I just was curious why bin/mspec ci seems to be getting slower at a rate non-proportional to the number of tests in it
18:06:53brixenI've been around and around the File.stat loop, let me look at it again
18:06:58dgtizedbrixen: and given that you had a lot of string updates I was curious
18:07:04djwhittdgtized: are you running on Gentoo 64?
18:07:17dgtizeddjwhitt: no ubuntu hardy
18:07:19evanbrixen: please don't rewrite it 8 times while i'm offline over the next week again
18:07:23evanok plz thxn.
18:07:30djwhittdgtized: 64 or 32 bit?
18:07:34brixendgtized: why wouldn't mspec get slower proportional to the number tests it runs?
18:07:49brixenevan: I won't, I'm just trying to avoid people making unfounded assumptions about that area
18:08:00dgtizedbrixen: it should get slower proprotional, it seemed like it was getting slower at a rate that wasn't proprotional to the increase in spec count
18:08:17dgtized32bit ubuntu hardy
18:08:18brixendgtized: well, seems or you have something to show that?
18:08:28brixendgtized: and, each spec is not constant time
18:08:42brixenthe ipaddr and socket specs have taken a long time
18:08:44djwhittdgtized: never mind then, seen weird stuff on 64 bit
18:09:22dgtizedbrixen: true, but I've been paying attention to spec checkin's too and it didn't seem like anything would be rediculously slow -- I just no that yesterday bin/mspec was going at around 40-50 seconds and now it's going at 55-65
18:09:46gnufied leaves the room.
18:10:07brixenbin/mspec ci => 41 sec; bin/mspec ci -fm => 39 sec
18:10:14evandid anything get found out about the be_empty problem?
18:10:22brixenI'd like to see someone else run that and show some numbers
18:10:25brixenevan: nope
18:10:49brixenevan: it only manifests that I can see on 64bit, and I have (conclusively to my mind) ruled out the be_empty matcher
18:10:51evanprinting to the terminal will always be slower than not, esp. while we have totally unbuffered writing
18:10:58evanthats one syscall per .
18:11:00brixenI can reproduce it with that code commented out
18:11:02benburkert leaves the room.
18:11:02djwhittevan: we figure out that Generator is screwed up
18:11:20dgtizedbrixen: that's your machine, and I'm sure it will be different from mine, what should be similar is that mine became slower over the past couple of days at a rate that didn't match the number of added specs
18:11:21djwhittevan: takes huge amounts of memory
18:11:28evanok, to me, it smells like something is up continuations
18:11:47evansince Generator uses them
18:11:53brixendgtized: I hear you, but I'm not seeing it. I can try on ubuntu 32bit and gentoo 64 if you give me two commit hashes
18:12:05evani think this a good time to let the rest of you know that i'm in the middle of a massive VM cleanup
18:12:07djwhittevan: I wrote this ticket about the 64 bit issues http://rubinius.lighthouseapp.com/projects/5089/tickets/434-slow-ci-runs-on-gentoo-54#ticket-434-3
18:12:32evansnapping off pieces and getting 100% test coverage on them as I go along
18:12:39patehey evan, are you EY guys heading to SLC together? do you know when you'll get here?
18:12:43evanand it's being transitioned to C++
18:12:49scoopro_O
18:12:52djwhittevan: C++ really?!?
18:12:58evanpate_: i'm not with other EY people, no.
18:13:08evanpate_: I'm arriving at 4pm
18:13:44evandjwhitt: yep, i think my reasoning is sound. It's that using an OO language allows us to match the VM to the semantics of the ruby much easier
18:14:00evanfor instance, the VM now has a Hash class, which has methods just like the ruby one
18:14:07dgtizedbrixen: let me a do a little testing, I am basing this on memory, and not written numbers so it's possible I misremembered
18:14:08djwhittyeah, I don't have a big problem with C++. just surprised
18:14:36evanthe idea is to use small parts of STL
18:14:44evanand simple C++
18:14:55evanto help make the VM more modular
18:15:09evanit's already helped me make the GC a ton cleaner
18:17:00evani know some people see C++ as terrible, and for quite a while so did I
18:17:00evanbut it's a tool like any other
18:17:05evanwith it's ups and downs.
18:17:17evanozy`: thats the spirit!
18:17:18brixenI think it's important for folks to have a realistic understanding of our current performance profile. I've profiled entire mspec ci runs at a high resolution (10,000 sps) and our profile is extremely flat
18:17:25evannext stop, FFI in fortran!
18:17:42brixenIOW, nothing contributes much more than a few percent to the whole time
18:18:01djwhittI don't think C++ is terrible. it's just abused
18:18:04djwhitta lot
18:18:16evanozy`: so, i'll put you down to work on that then? :)
18:18:40evandjwhitt: yes, it is. my hope is to avoid as much abuse that others do as possible
18:18:49evani've been reading a lot of C++ style guides and such
18:18:50scooprI'm also fine by c++, but being a developer in game industry kinda forces me being at least passable c++ coder :/
18:18:58evanand not much C++ code itself
18:19:10evanand using the simplest things I can
18:19:38scooprfor all kinds of cornercases etc. I like reading http://www.parashift.com/c++-faq-lite/
18:19:40evanfor instance, i don't forsee using any templates
18:19:47evanscoopr: thats what I read through
18:19:54scoopraight
18:20:11brixenbtw, if someone wants to investigate perf with File.stat, look into why File.stat is called like 12 times for each file required. that's what I was observing
18:20:18scoopryeah, templates wouldn't probably make much sense in a vm, except maybe some collection stuff
18:20:40evanscoopr: i've used templates in so far as vector requires you to use the vector template
18:20:57scooprand that's probably just enough
18:20:57evanmy thinking is that STL provides a body of functionality in the same way the ruby kernel does
18:21:14evanlean on it a bit to make things easier
18:21:22evandon't reinvent every wheel
18:21:23scooprstd::vector backed Array? =)
18:21:27evanha no.
18:21:37evanstd::vector is only used internal to the VM
18:21:59evanno std classes at all are used to implement actual ruby objects
18:22:04evanit's just to make implementing the VM easier
18:22:40evanthere are few more changes coming down the pike
18:22:45evanthat will show up in the new VM
18:23:26evanthe call convention changes have eliminated all instructions that pushed an unknown number of items on the stack
18:23:30scooprsomewhat related, I have a tiny request ;)
18:23:39evanso the compiler can now properly calculate max stack size
18:23:44djwhittevan: I assume you've seen this: http://www.sgi.com/tech/stl/
18:24:05evanand so i'm going to try moving towards more ST style MethodContexts
18:24:10evanwhere each MC has it's own stack
18:24:24evansized to what the compiler calculated the method required
18:24:29scooprin ruby Float is usually implemented as double. would it be too much trouble to make that configurable, say typedef double rb_real; or so?
18:24:29thehcdreamer leaves the room.
18:24:36evanthat will make Task much more lightweight
18:25:19evanand should (hopefully) eliminate the Continuation/Task problems that continue to plague us
18:25:35evani'm realizing that the one-stack-per-Task idea is really too complicated
18:27:26evanscoopr: sure, thats doable
18:27:31evanscoopr: may I ask why?
18:28:44scooprporting to platforms that don't have native double, or specialcases where precision doesn't matter (*cough* games)
18:28:55evan:)
18:28:56evansure.
18:29:06evanwhat other types might it be?
18:29:07evanfloat?
18:29:13scooprfloat, yeah
18:29:27evansure, i'll typedef it out.
18:29:37Defilerevan: That change should make it easier to write tests for the state of the stack, sounds like
18:29:46evanDefiler: very much so
18:29:49Defiler(one stack per MC)
18:30:10DefilerI have been worried about that for a while.. we don't have a lot of insight into what the stack is meant to look like after a break or return, say
18:30:16evanso, one problem with the current arch is that if you copy an MC
18:30:30evanthat MC still has indexs into the one stack shared with the original
18:30:47evanand if the original returns back up the stack... the copy loses it's data
18:30:49evanand all hell breaks loose
18:30:57evanwhich is just a stupid.
18:31:13NoKarma enters the room.
18:31:23NoKarmaHeya all
18:31:26evanby having each MC it's own stack, then when you copy an MC, it can just copy it's own stack
18:31:36NoKarmabrixen: you around?
18:31:36evanand life can happily continue
18:32:14brixenNoKarma: yo
18:32:16evanthe plan is to allocate the stacks in the lazy, stack-oriented way that MC's are allocated now
18:32:33evanthat should largely help with GC preesure
18:32:37NoKarmabrixen: You responsible for the Rubinius GSOC stuff?
18:32:50brixenNoKarma: responsible in what way?
18:33:04NoKarmawell, mentoring or something like that :)
18:33:06DefilerevaevCan ou explain why you opted to put the LRE in @lre instead of in, say, a slot on the SendSite?
18:33:10evanone problem now is also that to many parts of the VM have knowledge of the stack
18:33:13Defilerwhoa.. keyboard excitement
18:33:25brixenNoKarma: I'm going to sign up as a mentor because it sounds like they need more
18:33:35brixenNoKarma: and I'll help coordinate stuff with participants
18:33:37evanit would be a huge waste of memory in a SendSite
18:33:44evanplus, how would it be found if it were put in a SendSite?
18:34:00NoKarmabrixen: I'm intending to participate in this years GSOC
18:34:07brixenNoKarma: ok
18:34:12DefilerWell, ignoring that last part, how would it take up more memory in a SendSite?
18:34:23DefilerIt is already the most-frequently-allocated class in the system
18:34:24evanevery SendSite would require another slot for an LRE
18:34:28evanwhich would normally not be used
18:34:50evanbut the bigger question is not that
18:34:55NoKarmabrixen: And I was just glimpsing through the ruby-related ideas
18:34:56evanit's how would the other machinery find the LRE
18:35:01evanif it's located in a SendSite
18:35:17DefilerWell, we could pass the current sendsite as a pointer on 'state'
18:35:22evanhow would it know where to find the SendSite of the send that had the block?
18:35:27evanDefiler: that doesn't work
18:35:30evanbecause you have to come back to it
18:35:43DefilerCome back to what?
18:35:50evana block, which could be infinite depth from the send site of the first block
18:35:54evanhas to be able to find the LRE
18:36:23DefilerWell, in the trivial case, 'the_lre' could be a field of 'state'
18:36:39DefilerSince the VM already handles the return, it would just keep that field up to date like it does the rest of the 'world'
18:36:45srbaker leaves the room.
18:36:57evanwhere would it update it from?
18:37:03srbaker enters the room.
18:37:11evanthe_lre would constantly have to be swapped in and out
18:37:14evanevery method call
18:37:29evanand where would it be swapped in and out from?
18:37:36DefilerA lot of work is already done to achieve that
18:37:49evannot where to swap it in and out from
18:38:04evandon't forget, the LRE code required no VM support
18:38:05evanthat was big
18:38:17evanbecause it keeps the VM small
18:38:26DefilerYeah. It just keeps appearing as a hugely expensive thing in the profiler
18:38:31evanwhat does?
18:38:36evancreating one?
18:38:44Defilerallocating and collecting LongReturnException objects
18:39:08evanhm
18:39:27DefilerThere are more of them than Strings and Arrays put together, in a CI run, last time I checked
18:39:40evanwell, the compiler can help that
18:39:45DefilerThe approach is elegant, though
18:39:48DefilerIt just seems pricey
18:39:49evanif the compiler looks into the blocks
18:39:54evanit can figure out of an LRE is required
18:40:00evanif there is a return or break, it's needed
18:40:02evanotherwise, it's not
18:40:13DefilerAah. I will take a look at that
18:40:15evanso the LRE code could be omitted based on the code
18:40:21DefilerI suspect that will help a lot
18:40:59brixenyeah, that seems quite promising
18:41:05evanit might be as easy as using set/get inside the compiler
18:41:09brixenit is indeed the most allocated obj in a CI run
18:41:23evanto allow the block code, as it's processed, to set a property to indicate that it has a break or return
18:41:33evanand have the call then just perform a get() to see if the property was set
18:41:37evanif it was, then use the LRE code
18:41:40evanotherwise, dont
18:42:54gnufied enters the room.
18:45:55NoKarmaWhat are the "hot spots" of rubinus development right now? I'm having exams next week so I had absolutely no time in doing some rubinius related work (or even getting up to date about what you guys have been doing letely)
18:46:50brixenNoKarma: well, working on getting more of stdlib working/spec'd
18:46:53cremesNoKarma: IO specs, REXML specs, YAML specs... fleshing out the core libs
18:46:56brixenother apps, gems, etc
18:47:15brixenNoKarma: rubygems works in rbx now :)
18:47:55antaresand brixen did some great String methods performance optimizations
18:48:24NoKarmaWoooo! :D
18:48:26NoKarmaSounds cool
18:48:43antareslike it is ten times (or so) faster now
18:51:14antaresbrixen: I want to get myself familiar with FFI. What in the source code is a good way to start?
18:51:31brixenantares_: depends on your question. what are you looking to do?
18:52:10antaresbrixen: just want to understand how the interface work and document it somewhere
18:52:32brixenok, have a look at shotgun/lib/subtend in the files with ffi in the name
18:52:43brixenalso see MemoryPointer and FFI::Struct classes
18:52:58brixenthe stuff in kernel/platform generally
18:53:01antaresbrixen: thank you
18:53:08brixensure
18:56:48benburkert_ leaves the room.
18:57:37rueHey, what's with all the talkin'?
18:57:55brixentotally
18:58:00evansorry.
18:58:04brixenheh
18:58:12evanwell, i should go check my flight status
19:05:13TheVoicewhere you headed to?
19:06:21gnufiedTheVoice, oblivion!
19:08:33rueOr SLC
19:10:11NoKarmabrixen: I'm going to send in applications for 2 or 3 of the rubinius-related ideas. You guys can then select the one who is in the greatest need of someone working on it :)
19:10:50rueHm, do you think they would accept a proposal to provide back rubs to everyone?
19:10:54ruegoes to stretch
19:15:21benburkert enters the room.
19:18:28brixenNoKarma: sounds good
19:19:48agile leaves the room.
19:21:34NoKarmarue: Haha
19:23:16qwert666 enters the room.
19:23:39VVSizhi guys, a question about REXML specs
19:23:55VVSizseems like the fail on MRI 1.8.6 pl 111 (3 failures)
19:25:03obvio enters the room.
19:26:49DefilerVVSiz: That's not so good. Heh.
19:27:27VVSizin fact, it's a bit strange. with older MRI (default on Ubuntu 7.10), I see one failure. With pl 111 - three. With the very latest from svn 1.8 branch - zero.
19:28:05djwhittI think there were actually REXML bugs discovered in the course of writing those specs if I remember correctly
19:28:18djwhittfbuilesv was working on them
19:29:32VVSizOK, good. just checking. Ideally it would have been great to pass 100% on official "compatibility" target 1.8.6 pl111
19:30:50antaresVVSiz: there was a discussion on ruby-core about REXML bugs and behaviour AFAIK
19:31:20headius enters the room.
19:31:23antaresVVSiz: submit a ticket and assigned it to me. I was me who applied that spec patch
19:31:34brixenVVSiz: well, we need to up "official" to p114 I think
19:31:59headiusevening
19:32:05brixenmorning
19:32:06VVSizevening :)
19:32:23brixenok, technically afternoon :)
19:32:35brixenheadius: whereabouts are you today?
19:32:42headiustoday we are in Prague
19:32:50headiuslovely evening, a bit cold, but clear
19:32:51VVSizbrixen: right, pl114 is officially out! OK, I'll upgrade to 114 too.
19:33:02brixenVVSiz: yeah, I need to as well
19:33:19brixenheadius: sounds nice
19:34:34RyanTM enters the room.
19:35:54VVSizantares_: I'll ugrade to pl114 and we'll see whether there are any REXML failures then. thanks!
19:36:11antaresVVSiz: yeah I am already compiling p114
19:36:19VVSizsame here :)
19:36:31djwhittI'm a Gentoo user we've been on p114 for ages ;)
19:36:36VVSizand it seems that there are REXML related changes in it
19:36:58DefilerHorrible how many differences there are between 111 and 114
19:38:22headiusbrixen: so far so good
19:38:42headiussuper lagged though, we arrived in amsterdam at 6:30 and have had to just stay awake since
19:39:12VVSizantares_: hey, no Rexml related failures in pl 114. good.
19:40:57djwhittVVSiz: ah, looks like all the bugs were wrapped with ruby_bug in the specs
19:41:27VVSizhmmm, spoke too soon. I actually see 2 failures (previously was filtering them out with JRuby-specific excludes)....
19:42:34qwert666__ leaves the room.
19:43:07headius_ enters the room.
19:45:51antaresVVSiz: is there a changelog for p114 in source tarball?
19:46:34antaresVVSiz: I am still waiting for make doc, but go ahead and create a ticket if you see failures
19:50:34rubuildius_amd64 leaves the room.
19:51:42RyanTM leaves the room.
19:53:14headius leaves the room.
19:56:23benburkert leaves the room.
19:56:51VVSizbrixen: adding comments to exclusions of the form: ruby-core:55555 fails
19:57:05VVSizbin/mspec tag -t j --add 'fails(ruby-core:15981)' ....
19:57:18VVSizactually adds just ':' instead of fails(ruby-core:15981)
19:57:27brixenyeah, don't use :
19:57:33VVSiz:)
19:57:37brixenit's excluding
19:57:46brixenI believe there's a comment to that
19:57:57brixenexcluded
19:58:31VVSizah, ':' is such a useful thing to have. like in that ruby-core:xxx, or in URLs in comments
19:58:42VVSizbut I get it. I'll use ruby-core-xxxx for now
19:59:13antaresVVSiz: what's your resolution on this? Could you comment on http://tinyurl.com/2v23px ?
19:59:20brixenhmm, well if you really need it, I'll change the parser for the tags
19:59:50skaar leaves the room.
20:01:04VVSizantares_: done. thanks!
20:01:38brixenVVSiz: you need to come up with another character than : to separate tag or tag(comment) from the spec string
20:01:47brixenVVSiz: is it really that big a problem?
20:02:10VVSizbrixen: Understood. Yeah, it's not really that big.
20:02:41VVSizI thought if that's something trivial to correct, but indeed, there needs to be a char to separate tags from the spec
20:03:16brixenperhaps I could use ::
20:04:15djwhittanother cool thing about dwm
20:04:23djwhittit's smokin' fast
20:04:42djwhittthey're both similar in speed once they're running, but dwm is so fast to startup
20:04:57VVSizbrixen: if you're thinking of changing the format for excludes, please dont! :) I'll have to update all mine :)
20:05:00djwhittoops
20:05:03djwhittwrong window
20:05:07VVSiz:)
20:05:47brixendjwhitt: thanks for the info though :)
20:05:52djwhitthehe
20:05:53djwhittyep
20:05:56djwhittcomparing dwm and xmonad
20:06:03brixenVVSiz: well, it's a simple find/replace right?
20:06:10brixenVVSiz: : -> ::
20:06:59VVSizyeah, indeed
20:08:42brixenVVSiz: well, you say yea/nay then. I've just spec'd it and it seems to work
20:09:00brixenso, you can do: 'tag(this:5555)::Another#method'
20:09:40VVSizcool, I'm all for it! :)
20:09:51brixenheh, ok, I'll push in a sec
20:10:45VVSizyou're going to update rbx excludes too? if so, please post the one-liner for that :)
20:11:01brixenyeah, I'll update them
20:11:10brixenyou're asking me to pull out some sed-fu, huh?
20:11:12brixenhmm
20:11:43brixenfirst I have to fix the 12 mspec failures this causes in other parts of it :)
20:12:41VVSiz:)
20:13:31antaresVVSiz: ok I got it down to 1 failure
20:13:44antaresVVSiz: now need to figure out what changed in REXML in p114
20:14:45therealadam enters the room.
20:15:50VVSizantares_: if the latest 1.8 branch from svn passes all the tests, maybe we could just mark/exclude those that fail for 114 for now, sooner or later new patch level arrive and it will pass them :)
20:15:51brixenVVSiz: will you be around for a bit? I'm grabbing some lunch (and something for this major headache)
20:16:03VVSizyeah, I'm here
20:16:09brixenk, bbiab..
20:16:13VVSizthanks!!
20:17:09antaresVVSiz: hm, failure I have says the whole class is gone or got renamed, no less
20:17:20antaresVVSiz: I am looking for a changelog now
20:18:37VVSizman, I *really* with MRI folks would start using rubyspecs and run them before every change :)
20:19:13antaresor get hannibalized by Rubinius :)
20:19:40antaresI should have said "or Ruby gets hannibalized" I guess :)
20:22:27VVSiz:)
20:22:30w1rele55 leaves the room.
20:23:39headius leaves the room.
20:23:41antaresVVSiz: it was a missing require problem
20:23:47antaresshould have fixed now
20:23:49antareslet me try
20:28:35benburkert enters the room.
20:29:22brainopia_ enters the room.
20:29:52brainopia_ leaves the room.
20:30:28brainopia_ enters the room.
20:31:08antaresVVSiz: I should swap existing rexml in lib with a new copy from p114, right?
20:32:46VVSizantares_: dunno about rubinius impl. I mostly concerned with the specs and the fact that pass on our official MRI.
20:32:58VVSizso that we could use them against JRuby. :)
20:33:44antaresVVSiz: no, I mean we should have new REXML in Rubinius repository
20:33:59antaresVVSiz: nevermind, I'll read LH pages once more
20:37:14lopex enters the room.
20:39:29olabini enters the room.
20:40:20srbaker leaves the room.
20:46:01brainopia leaves the room.
20:46:15antaresVVSiz: REXML in p114 has a bug: method with 1 argument in constructor signature is called with 2 agruments in rexml/document.rb
20:46:26antaresVVSiz: should we tag this as Ruby bug?
20:47:50zenspiderantares_: pls report that to ruby-core
20:48:28antareszenspider: sure, will do. Let me try to check out stable branch of 1.8 first and check things there
20:50:51srbaker enters the room.
20:51:54ctennis enters the room.
20:57:38antaresIn Ruby 1.8 branch (rev 15833) REXML is completely reorganized compared to p114
20:59:05kw enters the room.
20:59:32antareszenspider: do you think providing failing spec we have as example would be good idea for ruby core? Or should I write a simplest possible example to prove that as a script?
21:00:37zenspiderantares_: prolly nicer to provide a minimal repro script and lower the bar for them
21:00:43rby_ enters the room.
21:01:34antareszenspider: roger
21:01:56zenspidertnx
21:04:38ndemonner leaves the room.
21:11:40antareszenspider: ok sent to ruby-core. Should we wrap this example in ruby_bug for now?
21:12:49obiejuan enters the room.
21:15:18macournoyer enters the room.
21:25:27boyscout1 commit by Brian Ford
21:25:28boyscout * Allow ':' in MSpec tag comments: tag(bug:555):Some#method.; 1da73a0
21:26:05zenspiderantares_: sure, why not?
21:26:19brixenVVSiz: it was much easier :)
21:26:40brixenVVSiz: so, I'm off the hook for the one-liner (which I couldn't easily produce anyway)
21:31:24dewd leaves the room.
21:41:02rubuildius_ppcBrian Ford: 1da73a015; 1789 files, 6233 examples, 22091 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/171701
21:42:02rubuildius_amd64 enters the room.
21:46:20antaresbrixen: wrapping example into ruby_bug does not save from Wrong number of arguments error, any workarounds?
21:46:51antaresbrixen: http://tinyurl.com/3c4c8o for more details on what I face with REXML specs
21:47:32tizianobis enters the room.
21:48:34brixenantares_: what's the exmple I'm looking at?
21:49:29antaresbrixen: REXML::Document#write returns document with transitive support (ruby/1.8/library/rexml/document/write_spec.rb line 33
21:49:38brixenk, sec
21:50:06rubuildius_amd64Brian Ford: 1da73a015; 1789 files, 6230 examples, 22062 expectations, 0 failures, 0 errors; http://rafb.net/p/ahRwtH98.html
21:50:19brixenantares_: is that in your local repo? if so, can you pastie me
21:50:30brixenI don't see anything but 'end' on line 33
21:51:21pastiebrixen: http://pastie.org/171706 by antares_.
21:51:59AndrewO leaves the room.
21:52:04brixenantares_: ok, so ruby_bug must be yielding then
21:52:10antaresbrixen: spec file and backtrace provided, sorry for plain text
21:52:21antaresbrixen: so tag it I guess?
21:53:46brixenantares_: what system are you running under?
21:53:50brixenrbx or mri?
21:54:06brixenrbx, right?
21:54:11antaresbrixen: it is rake spec:ci run
21:54:17brixenyeah, ok
21:54:22brixenruby_bug only guards for MRI
21:54:30brixenso, for rbx, either fix or tag I guess
21:54:50antaresbrixen: there's not way to fix it now, see the link above
21:55:09antaresbrixen: ok I'll tag it — need to figure out how to do it though :)
21:55:30brixenbin/mspec tag spec/file/name_spec.rb
21:55:44jptix enters the room.
21:56:22antaresbrixen: does it tag the whole file?
21:56:29brixenantares_: whatever fails
21:56:50brixenso, I'm still a bit lost. this fails rbx because the spec is for REXML behavior that doesn't exist yet in MRI?
21:57:19antaresbrixen: REXML is messed up in p114
21:57:39antaresbrixen: there are calls to method with "old" arity in the code
21:57:57antaresbrixen: and in ruby_1_8 branch as of today REXML is *completely* reorganized
21:58:07brixenahh, I gotcha
21:58:12brixenwe're using old REXML
21:58:18antaresbrixen: so there's no way to fix it before Ruby core team explains what the heck is going on with it
21:58:20wmoxam leaves the room.
21:58:22brixenok
21:58:34antaresbrixen: we are using p114 (I do in my repo)
21:58:52antaresit is screwed up in 114, before that everything is ok
21:59:01brixenso, you could: bin/mspec tag --add 'fails(see ruby-bug:XXXX)' spec/path/to_file_spec.rb
21:59:39antaresbrixen: I left a comment in spec, will update it when someone pops up on ruby-core
21:59:48antaresok rake spec:ci now passes
21:59:52brixenk, sounds good. thanks
22:00:50moofbong leaves the room.
22:01:20jptix leaves the room.
22:01:41jptix enters the room.
22:04:43hassox enters the room.
22:05:30kw leaves the room.
22:07:52antaresbrixen: now a stupid question. How do one use pastie bot (the way you did recently to get pastie from me)?
22:09:11boyscout1 commit by Michael S. Klishin
22:09:11boyscout * Update stdlib and specs for REXML from 1.8.6 patchlevel 114 (see details!); 6886ec5
22:09:32brixenantares_: not sure of your question. if you type pastie, it should pm you
22:10:22brixenpastie: for antares_
22:10:41antaresbrixen: that's what I was asking about, thanks
22:10:47pastieantares_: http://pastie.org/171721 by brixen.
22:13:07KirinDav enters the room.
22:15:05rubuildius_amd64Michael S. Klishin: 6886ec585; 1789 files, 6229 examples, 22061 expectations, 0 failures, 0 errors; http://rafb.net/p/HH9N2f74.html
22:15:31tizianobis leaves the room.
22:16:05agardiner enters the room.
22:17:23pate_ leaves the room.
22:17:32bitbang enters the room.
22:19:21VVSizbrixen: very nice. thanks! :)
22:26:15KirinDave leaves the room.
22:28:52rubuildius_ppcMichael S. Klishin: 6886ec585; 1789 files, 6232 examples, 22090 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/171733
22:34:23rueHaha, awesome
22:35:09rueThe download for p114 said "8 minutes remaining" I was thinking "when the hell did Ruby get that big?" and turns out it is coming in at 14kbps :D
22:35:48agardinersounds like a server may be... busy :-)
22:36:22jptix leaves the room.
22:38:15wycats enters the room.
22:41:26NoKarmawycats: Heya!
22:41:57wycatsNoKarma: ltns
22:42:40wycatszenspider: you were looking for me?
22:43:04zenspiderwycats: hey you.
22:43:12zenspideryou reported something odd about rubygems + hoe?
22:43:48wycatsyes
22:44:10wycatsif hoe is a dependency of any gem, it tries to connect to rubyforge when you activate the gem
22:44:30zenspiderwhat is "it" ?
22:44:41wycatsthe wizard behind the curtain
22:44:45wycatshere's an example
22:44:48wycatsI'm on an airplane
22:44:52wycatsI go into my merb directory
22:44:54wycatsI type merb
22:44:56zenspiderI'm the wizard behind the curtain
22:45:01wycatsI get an error about not connecting to rubyforge
22:45:04wycatsand I cannot start merb
22:45:09wycatsI then go into the gemspec for RubyInline
22:45:12wycatsand remove the hoe dependency
22:45:15wycatsviola... it works
22:45:47twbray enters the room.
22:46:04jinjing leaves the room.
22:46:09zenspidercan you show me that error? pref with actual traceback?
22:46:34wycatsummm...
22:46:45wycatsI have to figure out how to replicate it when I have internet ;)
22:46:55Arjen_ enters the room.
22:47:00drbrainhaha, viola
22:47:07NoKarmait's voila :)
22:47:16wycatsI meant the character in shakespeare
22:47:22wycatssweet Viola
22:48:10wycatszenspider: ideas on how to replicate it with connectivity?
22:48:16drbrainat least it wasn't "wala"
22:48:22drbrainI want to stab those people
22:48:25zenspiderso true
22:48:34zenspiderI hate wala... they need to be shot
22:48:50zenspiderwycats: something like little snitch maybe
22:48:55zenspideror. /etc/hosts
22:49:00wycatsgood idea
22:49:13wycatsand I need to add the hoe dep back in ;)
22:49:55zenspiderI am not hitting this
22:50:10wycatsok
22:50:11zenspiderso I need a merb project that winds up doing something with PT
22:50:16wycatsyeah
22:50:19wycatsjust install action-args
22:50:21zenspiderI added 127.0.0.1 for rubyforge in etc/hosts
22:50:24wycatsaction-args requires rubytoruby
22:51:09zenspider?? sudo gem install action-args ?
22:51:58hoopyso do y'all have rails running on mod_ruby yet or what!? </sarcasm>
22:52:08hoopyer mod_rubinius
22:52:12drbrainhoopy: EWRONGCHANNEL
22:52:16zenspiderI don't see ANYTHING directly networky in hoe
22:52:20wycatszenspider: sudo gem install merb-more
22:52:29zenspiderEDONTBECLEVER
22:52:31wycatshoe doesn't try to connect to rubyforge?
22:52:34wycatsvia the RubyForge gem?
22:52:43wycats?
22:52:51zenspiderthe tasks do
22:52:58zenspiderlike rake release
22:53:04wycatsthis was happening simply by activating the gem
22:53:12wycatsI wasn't even requiring hoe
22:53:14womble enters the room.
22:53:19wycatsI was requiring Ruby2Ruby
22:53:43brainopia_ leaves the room.
22:53:45drbrainas far as RubyGems is concerned, nothing hits the network there
22:53:47wycatslemme see if I can catch the request via little snitch
22:53:53wycats?
22:54:09wycatsit definitely happened
22:54:13wycatslet me see if I can reproduce
22:54:23zenspidergak... merb-more is icky!
22:54:33wycatshow so?
22:54:48wycatsI can try and make it less icky
22:54:59zenspiderfour hundred gems just installed!
22:55:30wycatsgem install merb-more is a convenience
22:55:37wycatsthe gem in question is merb-action-args
22:55:45wycatswe want people to be able to do gem install merb
22:56:08wycatsthe alternative is Rails... where everything is in one gem
22:56:16wycatstake your pick
22:56:22zenspiderso, now that this is installed, merb blah; cd blah; merb should repro
22:56:26zenspider?
22:56:41wycatsyou have to require merb-more in init.rb
22:56:48wycatsor at least merb-action-args
22:56:57wycatsI would also make a controller with a method that has action-args
22:57:01wycatsdef foo(bar); end
22:57:12wycatsLittle Snitch: "This software requires that you restart your computer. Click restart to complete this installation."
22:57:13wycatswhaaaaaa
22:57:44zenspiderI assume you mean config/merb_init.rb ?
22:57:50zenspiderI added require 'merb-more'