Index

Show enters and exits. Hide enters and exits.

00:00:46Arjen_ leaves the room.
00:02:07KirinDave leaves the room.
00:08:12crossblaim_ enters the room.
00:08:12crossblaim leaves the room.
00:08:41rueAlright, now I have: x86 Arch Linux 3/08, amd64 Arch Linux 3/08, x86 Xubuntu 7.10, amd64 Xubuntu 7.10, x86 OpenBSD 4.2, x86 NetBSD 4.0, x86 DragonflyBSD pre2, amd64 FreeBSD 7.0R, amd64 Solaris 10, x86 OS X Leopard, x86 plan9 a, x86 Haiku current pre-A1, x86 Syllable 0.6.4
00:08:51crossblaim_ leaves the room.
00:08:58ezmobiusdamn
00:09:15rueShould help a bit with the platform testing. When I get my mips and arm from storage, should be golden :D
00:09:23rueVagabond has the UltraSPARC covered
00:09:38brainopia leaves the room.
00:09:48aotearoa enters the room.
00:10:00antaresrue: pretty impressive, even Plan9 is covered
00:10:27rueYep, I expect plan9 will be a first-tier Ruby platform any day now
00:10:31antaresrue: Rubinius definitely should save old enterprises with cobol on plan9
00:10:52KirinDave enters the room.
00:11:07antaresrue: how about windoze?
00:11:13rueEh
00:11:47antaresrue: I am seriiously considering putting some efforts to Windows-specific issues when I get back from my trip in 2 weeks or so
00:12:19antaresrue: I've bee windows user working with Ruby myself and I hope something will change with Rubinius :)
00:12:47ezmobiusheh all of our coraid SAN devices run plan9
00:14:58rueantares: Me too, but Microsoft has not gone under yet..
00:15:24antaresrue: pain can be decreased anyway I think
00:15:34rueThe APE should actually allow porting to plan9 relatively easily
00:17:30rueI want to get one of these: http://www.pegasosppc.com/
00:18:55rueMainly because of MorphOS
00:36:16tarcierioh yeah, I remember looking at those boards back in the day
00:36:34rueSolaris loads up slow with 128MiB :D
00:36:46tarcierihaha
00:37:51antares leaves the room.
00:39:45rby leaves the room.
00:44:56rubuildius_amd64 leaves the room.
00:45:59rubuildius_amd64 enters the room.
00:47:24bricolage leaves the room.
00:48:10boyscout3 commits by Adam Gardiner
00:48:11boyscout * Enable storage of CM#args in unused (persisted) slot 9; f86a6b7
00:48:12boyscout * New stables that don't use CM slot 18; 612bc2f
00:48:13boyscout * Re-work CompiledMethod#args generation, and temporarily disable; 6feaf99
00:48:43agardinerbrixen: looks like you were right about the stables... had no problem removing the CM slot this time round
00:58:31rubuildius_amd64Adam Gardiner: f86a6b77c; 1845 files, 6269 examples, 22362 expectations, 0 failures, 0 errors; http://rafb.net/p/6SNzxa93.html
00:58:50AndrewO enters the room.
00:59:41halorgiumpastie: yo
01:01:21rubuildius_ppcAdam Gardiner: f86a6b77c; 1845 files, 6272 examples, 22391 expectations, 0 failures, 0 errors;
01:05:27macournoyer enters the room.
01:09:11lstoll enters the room.
01:22:23webmat enters the room.
01:25:51avibryantbrixen: fyi, the next annoying thing I hit in mspec was in shuffle,
01:26:01avibryantary[r], ary[i] = ary[i], ary[r]
01:42:14rueagardiner: Oh, I locally changed :bonus to :metadata. Should I push it?
01:42:37agardineryeah, i was gonna do the same - you beat me to it! :-)
01:42:40tarcierisweet
01:42:44drbrainpush it real good
01:43:22agardineri was also thinking about pulling the local_names out of bonus/metadata and into their own slot, since we have something like three unused slots on CM still
01:43:46therealadam leaves the room.
01:43:49agardinerwhat do you think? this would mean metadata would be unused (for now)...
01:44:10wmoxam enters the room.
01:44:23rueI was actually thinking making it even more explicit, :metadata_container or something. Perhaps that is 'too' clear
01:44:59agardinerworks for me - you won't use the long name in user code
01:45:31rueagardiner: Unless there is a scenario where we do /not/ want the local names generated, I see no harm in it
01:45:59agardinerno, we always generate the local names, and need them for debugging...
01:46:12agardinerwhen they were put in bonus, we didn't have any free slots
01:46:20agardiner...but now we do! :-)
01:46:37rueHeh
01:46:53agardinerwhat's the status/thoughts about putting the sexp into CM?
01:47:24agardinerthis one probably ought to be in bonus and something you can enable/disable...
01:47:36ezmobiusthat shoudl probably be optional no? otherwise woudln't that have pretty large memory footprints?
01:47:41agardineryeah
01:48:06agardineri think you'd probably want to use a compiler flag to have it saved - default is no sexp
01:49:19agardinerbut the issue i see is that the sexp does not map neatly to a CM - i.e. CMs are nested, so the the sexp for an outer CM probably ought not duplicate the sexp for an inner CM...
01:49:40agardinerbut what do you put in its place? a special node maybe?
01:50:02rueWell, are you storing the sexp or the AST?
01:50:20agardineri figured it would be the sexp...
01:50:46agardinerast is probably harder to manipulate
01:51:39agardinerand also gets normalised, potentially losing info in the process
01:52:19rueDanng, Xubuntu performs far worse on 128MiB than /Solaris/
01:53:39rueTrue, the sexp would probably be better
01:54:00agardineri think it would be smaller than the ast as well...
01:54:08agardinermostly symbols
01:54:09rueDefinitely
01:54:44rueOn one hand you could always re-parse the sexp if you have the lines.
01:55:25agardinerwell, if you've got the sexp you can always recompile to get the AST... and there is no guarantee you'll have the source .rb around
01:55:26rueBut a sexp with a special substitution node should be good; ideally would have a reference to the subsexp
01:55:37agardineryeah, that's what i was thinking
01:56:10agardinerthe CM the sexp is from has a reference to the sub-CM, so it should be easy to recreate the entire sexp from that
02:00:01rueYep
02:00:40rueHehe, you could try the conditional compilation to exclude it ;)
02:03:22agardineryeah! nice feature btw!
02:19:22radarek enters the room.
02:20:07rubbish enters the room.
02:20:58d2dchat leaves the room.
02:24:02VVSiz_ enters the room.
02:31:55agile enters the room.
02:33:17wmoxam leaves the room.
02:36:03fbuilesv enters the room.
02:36:45radarek leaves the room.
02:40:49wmoxam enters the room.
02:42:20VVSiz leaves the room.
02:44:22agardinerwow! just run some static code analysis...
02:44:42agardinermost commonly occurring opcode is pop, at 9.88%
02:44:53agardinernext up is send_method at 9.30%
02:45:21agardinerwe could definitely optimise away a lot of these pops...
02:46:38agardinersend in all its guises is just over 17% of all instructions!
02:47:44eventualbuddha leaves the room.
02:47:56agardinerinlining will probably prove a very fruitful endeavour, performance wise
02:52:50smparkes enters the room.
02:56:43rueSounds like a worthy place to look
02:56:47aotearoa leaves the room.
02:58:17agardinerhmm... excluding the specs gives a slightly different profile
02:58:24agardinernot surprising really...
02:59:42agardinersends come down to just over 13% of all instructions
03:00:24agardinerbut pop is still the most common at just over 10%
03:06:33rueDo we currently erase the "frame" when returning?
03:06:43rueOr is that handled manually?
03:06:55agardinernot sure i follow...
03:07:56ezmobius leaves the room.
03:08:04wycats leaves the room.
03:11:59agardinerok, when a return is performed, the context is retired, the sender is activated, and the sp is reset to the sender sp value
03:12:07agardinerthen the return value is pushed on the stack
03:12:51agardinerand then in many cases, it is popped right off again
03:15:59rueRight, I was wondering if it would be faster to just clear the entire frame; that would allow a sort of a fast-and-loose mechanism where we do not need to care so much about whether there is junk on the stack below the useful values
03:17:15rueI need to look at that part more, it is possible that it would not work out that way. It might be too difficult to produce the code where that did not matter
03:17:45agardineryeah, it is doing that by resetting the sp to the sender sp - essentially throwing away the callee frame
03:18:08agardinerbut send is not the most common case where a pop immediately follows
03:18:22agardinerthat is actually on set_local
03:19:11agardinerand that can be completely optimised away if we modify set_local or added a new opcode that didn't leave the value on the stack as well as storing it
03:19:42agardineri spoke to evan about this before, just haven't got round to implementing it
03:20:17d2dchat enters the room.
03:20:36rueWe can determine whether that value will be used in most cases, right?
03:20:54agardinerfor set_local, we can tell in every case whether it will be used
03:21:06agardinerif the next op is a pop, it ain't needed
03:21:37agardinera keyhole optimiser could sort this out really simply
03:21:37macournoyer leaves the room.
03:22:20kofno leaves the room.
03:22:21agardinerit isn't easy to tell at bytecode generation time, but it is easy to tell in generator
03:23:22djwhittare the docs at http://rubini.us/doc/vm/ kept up to date?
03:23:38agardinernot automatically...
03:23:44agardinerevan has to regenerate them
03:23:47kofno enters the room.
03:23:52djwhittah, ok
03:24:01agardinerthey are a little out of date, insofar as some opcodes have been removed
03:24:08djwhittk
03:24:20djwhittI'll look at them in my checkout then
03:24:36agardinerand the source code listings may have changed a little, but by and large, the descriptions ought still be accurate
03:24:53agardineryeah, that is a safer bet
03:25:21headius leaves the room.
03:26:25eventualbuddha enters the room.
03:28:35wycats enters the room.
03:31:15rueActually, I can update them.. I just tend to forget
03:31:51agardinerhmm... i think they are in two places now on rubini.us
03:33:05agardineri tend to use to the old rubini.us/doc/vm link, rather than the rubini.us/rbx_documentation/vm-documentation link
03:33:13agardiner(less typing) :-)
03:33:30rueNo, actually that is the same. rbx_documentation just links to the old stuff
03:33:44agardineroh, ok - neat!
03:36:08agardinerrue: you going to push that bonus rename soon, or should i just rename it as part of moving local_names?
03:37:04rueOh, yeah.. I meant to run a sanity and forgot. Sec, CI & stable CI
03:37:06agardiner(not that i'm hassling you or anything) :-)
03:37:24agardinerk, np
03:39:52wyhaines leaves the room.
03:41:16AndrewO leaves the room.
03:44:41ezmobius enters the room.
03:45:49rubbish leaves the room.
03:49:21kamal_farizare the stars back from the whirlwind ruby world tour?
03:49:25kamal_farizheh
03:49:42kamal_farizoh, rubyfools and scotland on rails are on
03:51:59crafterm enters the room.
04:01:36wmoxam leaves the room.
04:03:47loincloth enters the room.
04:05:52loincloth leaves the room.
04:05:58loincloth enters the room.
04:14:57wmoxam enters the room.
04:24:17nicksieger leaves the room.
04:26:43loincloth leaves the room.
04:30:53agile leaves the room.
04:32:48AndrewO enters the room.
04:33:08ruekamal_fariz: Heya
04:33:27rueYeah, evan is still in Denmark or wherever. He has actually been offline mostly
04:34:35antares enters the room.
04:37:35loincloth enters the room.
04:38:26obvio leaves the room.
04:38:39obvio enters the room.
04:40:11kamal_farizhey rue
04:40:22kamal_farizhow's mod_rubinius coming along
04:40:24brixenrue: were we in #rubyspec?
04:42:21ruekamal_fariz: 'S coming, coercing the best behaviour out of Apache has been most of my efforts at this point
04:43:01kamal_fariza pleasant experience
04:43:34ruebrixen: I dunno, I idled on the channel for a bit to reserve it but I never saw anyone so I have not bothered
04:46:31ruekamal_fariz: Heh, yeah not so much
04:47:01rueFortunately the timeframe I am working with allowed a little bit of experimentation in that respect
04:47:49rueI am now moving to finalise the Apache side and then look at the options I have worked on for the Rubinius side
04:48:09brixenrue: ok, I'm idling there now
04:48:17brixenxmlhacker: you around?
04:48:42xmlhackerbrixen: yup... what'x up?
04:48:53xmlhackers/what'x/what's
04:52:56rueHm, should I generate new stables for the name change?
04:53:43agardinerit shouldn't really be necessary, since it is compiled down to push_my_field <field_num>, and you haven't changed the num
04:53:54rueYeah.. saves a little repo space
04:54:01agardinerhehe
04:55:21boyscout1 commit by Eero Saynatkari
04:55:22boyscout * Renamed the bonus field of CM to metadata_container for clarity.; 14256ab
04:55:50rueMan, I have been paranoid about pushing code lately
04:55:59kamal_farizi'm gonna be in the seattle area in a couple of weeks, wondering if drbrain and zenspider would be around
05:03:14brixenrue: hmm, seems to be broken on my system
05:05:05rueWhere is it brokenating?
05:05:25agardinerseems ok to me... are you trying to make rue even more paranoid? :-)
05:06:20be9 enters the room.
05:07:29brixenrue: damn, just realized it's 4.2
05:07:39rubuildius_amd64Eero Saynatkari: 14256aba2; 1845 files, 6269 examples, 22362 expectations, 0 failures, 0 errors; http://rafb.net/p/0GembO24.html
05:08:05rueHm, 0.1 here, what timezone are you in?
05:08:31brixenheh, 08.04.02
05:09:51rueOh! Yeah, I already bought the coconuts and an umbrella. I got two because I was not sure of the fabric. Did you find the armadillo yet?
05:10:08brixenhaha
05:10:32brixenarmadillo? I thought you said anteater
05:10:57rubuildius_ppcEero Saynatkari: 14256aba2; 1845 files, 6272 examples, 22391 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/174437
05:11:03rueNo, it will not go that high
05:13:36wyhaines leaves the room.
05:13:36dbussink leaves the room.
05:13:36rphillips leaves the room.
05:14:07rphillips enters the room.
05:14:07dbussink enters the room.
05:16:39ruebrixen: Aaaanyway, what about 4.2? I am not hip to the groove
05:17:22rueAlthough a day versioning system would be great. Could maybe sell it to Emacs enthusiasts 'cause it goes to version 23
05:17:32brixenahh, just that it's hard to say april fool's on 4.2 :)
05:17:50rueHa! Meany
05:17:54brixenrue: but the moment of comedy is long past
05:18:46rueYes
05:19:16rueUnless April 1 itself is a hoax that will be revealed on some upcoming second of April
05:19:52dbussink leaves the room.
05:19:52rphillips leaves the room.
05:20:18brixenheh
05:20:39rphillips enters the room.
05:20:39dbussink enters the room.
05:21:40brixenI wish irc had sound, you could get these shredding effect every time the net splits
05:22:10rueThat would get really annoying in about 3 minutes, I think
05:23:48agardinerespecially when you get the big splits... might need a more serious earth-crumbling-metal-wrenching sound effect for that
05:23:50brixenhunny, what's that sound? oh nothin'. the net's splitting again
05:29:36loincloth leaves the room.
05:29:57brixenozy`: yeah, tried colloquy for about 3 minutes
05:31:18eventualbuddha leaves the room.
05:32:40agile enters the room.
05:33:03RyanTM leaves the room.
05:35:00AndrewO leaves the room.
05:35:15wmoxam leaves the room.
05:38:22loincloth enters the room.
05:38:50yaroslav enters the room.
05:40:51agile leaves the room.
05:48:57ezmobius leaves the room.
05:52:20loincloth leaves the room.
05:55:24wycats leaves the room.
05:57:50kofno leaves the room.
06:07:41obiejuan leaves the room.
06:11:57nkpart_ enters the room.
06:14:08wycats enters the room.
06:14:30wycats leaves the room.
06:18:29nkpart leaves the room.
06:21:00wycats enters the room.
06:32:50yaroslav leaves the room.
06:33:29yaroslav enters the room.
06:51:24mediogre enters the room.
06:52:01yaroslav leaves the room.
06:52:39yaroslav enters the room.
06:59:38boyscout2 commits by Adam Gardiner
06:59:39boyscout * New stables; moved CM#local_names to own slot; f573d5c
06:59:40boyscout * Move CompiledMethod#local_names to unused slot; 15df180
07:01:00antaresagardiner: mm, what slots do you mean here?
07:01:11agardinerslots == fields
07:01:33agardineri moved local_names from what was previously the bonus tuple into a free slot on CompiledMethod
07:01:59antaresagardiner: I see, thanks. This is a good move.
07:02:20agardinerwell, its a little cleaner
07:03:55agardinerpersonally, i think the bonus/metadata_container slot should be hash (rather than a tuple) that contains optional metadata
07:04:49agardinerlocal_names are used by eval and are not optional
07:07:08agardineranyways, CompiledMethod is a lot cleaner now, with just one unused field and no deprecated fields
07:07:47antaresagardiner: what is unused? I did not closely follow logs last 2 days, quite a lot of work :(
07:08:09agardinerer... perhaps "available" would have been a better term
07:08:20agardinerthere is no field defined that is unused
07:08:29agardinerwe just have a slot free that is not being used
07:08:38agardinerif that makes sense?
07:09:12agardineri.e. slot 14 is not being used for anything
07:09:25antaresagardiner: that is ok I think
07:09:48agardineryeah, its a bit of a pain to remove an unused slot, and we'll probably find something to put in there later anyway
07:11:04rubuildius_ppcAdam Gardiner: f573d5c03; 1845 files, 6272 examples, 22391 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/174462
07:12:35rubuildius_amd64Adam Gardiner: f573d5c03; 1845 files, 6269 examples, 22362 expectations, 0 failures, 0 errors; http://rafb.net/p/K0t7Up92.html
07:13:40Skip enters the room.
07:20:05agardinerlater!
07:20:16agardiner leaves the room.
07:28:04rby enters the room.
07:35:52rby leaves the room.
07:36:01lstoll leaves the room.
07:41:11rueI was going to say that it does not really matter what goes in the metadata_container field except when if/marshalling it
07:49:50crafterm leaves the room.
08:08:29cypher23 enters the room.
08:10:50wycatsis there a good POSIX syscall reference around?
08:12:26ruebrixen: What is the web front looking like? I am adding an IRC nick list which needs to be editable obviously so LH for now
08:13:28brixenrue: which?
08:14:13ruewycats: I have a Ruby script that generates that for you, just a sec
08:15:16ruewycats: Ah, here we go
08:15:23rue`wget http://www.google.com/search?q=posix+system+call+reference`
08:15:43rue;)
08:16:05wycatsrue: sadly it's not bringing up much useful
08:16:25ruewycats: The ChorusOS one probably has all of them
08:16:39rueWell, depending on your exact need
08:17:12ruebrixen: Any plans that are currently ongoing to add software, is someone working on it etc.?
08:17:49thehcdreamer enters the room.
08:18:50rueI am editing that page someone mentioned a typo-ish on
08:20:47d2dchat leaves the room.
08:30:24brixenrue: you mean rubini.us?
08:34:02benburkert leaves the room.
08:42:46ko1_away leaves the room.
08:45:18Rich_Morin_ leaves the room.
08:47:58KirinDave leaves the room.
08:51:07rueRight
08:55:01TheVoice leaves the room.
08:59:31mutle enters the room.
09:02:01headius enters the room.
09:03:11rueDamn it, I KNEW I should not have started this stupid editing
09:03:17headiushowdy
09:03:36rueHallo
09:09:53crossblaim enters the room.
09:10:23crossblaimhi
09:12:28octopod enters the room.
09:13:41jartz enters the room.
09:17:57Arjen_ enters the room.
09:20:11KirinDave enters the room.
09:21:51KirinDave leaves the room.
09:31:25benburkert enters the room.
09:31:28benburkert leaves the room.
09:33:30rueStupid stupid stupid, I always end up rewriting the entire thing
09:45:26wycats leaves the room.
09:47:21yaroslav leaves the room.
09:47:52yaroslav enters the room.
09:53:05yaroslav leaves the room.
09:53:41yaroslav enters the room.
09:57:58yaroslav leaves the room.
09:58:36yaroslav enters the room.
10:00:18yaroslav leaves the room.
10:00:58yaroslav enters the room.
10:04:03BlackEdder enters the room.
10:06:48qwert666 enters the room.
10:08:38wycats enters the room.
10:09:31lstoll enters the room.
10:18:18zenspider leaves the room.
10:30:13VVSizA = 12; class A::B; end - excellent one-liner
10:30:25VVSizcrashes Ruby 1.9, hangs or crashes rubinius, crashes JRuby
10:32:35rueIs it the parser?
10:33:40rueYeah, that should be a pretty trivial check actually, if it makes it past the parser which it should
10:36:43VVSizpastie: for rue
10:36:47wycats leaves the room.
10:37:00pastierue: http://pastie.org/174505 by VVSiz.
10:37:11VVSizthat's what I see in interactive shell
10:37:43VVSizif I put the whole line at once, it just hangs. if I first add A = 12, and the rest on the next line, I see that crash.
10:38:09rueI just get an assertion on the one-liner
10:40:09VVSizmaybe I should do the full cleanup/rebuild then, to make sure there are no side-effects
10:42:43VVSizafter clean rebuild, on Linux. in interactive shell, entering the full line just leads to: Segmentation fault
10:43:05brainopia enters the room.
10:45:21headiusit's not a syntactic issue
10:45:47headiuswhen walking the const chain it's probably expecting each element will be a module or class without confirming it
10:45:55radarek enters the room.
10:49:16tokengeek enters the room.
10:59:31obvio171 enters the room.
11:03:40cypher23 leaves the room.
11:08:12yaroslav leaves the room.
11:08:51yaroslav enters the room.
11:13:13yaroslav leaves the room.
11:13:30yaroslav enters the room.
11:13:46yaroslav leaves the room.
11:17:15obvio leaves the room.
11:18:14dysinger leaves the room.
11:25:58rueVVSiz: Hah, my assertion was from bad stables..
11:26:49headiusso you're getting the same result then, yes?
11:29:31cypher23 enters the room.
11:34:54obvio171 leaves the room.
11:35:29rueArgumentError, rescued from an invalid field access
11:41:51rueI thiink it is an easy fix, lessee
11:43:02scoopr*37 hours passes by*
11:44:17rueAh, fixed.. are there related cases, maybe? It is quarter till seven, my brain is not at its sharpest
11:53:18kofno enters the room.
11:53:54jtoy enters the room.
11:55:40ruivaldo enters the room.
11:56:03rueHm, why is this so slow..
12:00:24ctennis leaves the room.
12:10:23rueOh, haha, DEV=1
12:10:53ariekeren enters the room.
12:21:05evanrue: you around?
12:24:48dbussinkso, i tunneled my way through from work now :)
12:24:48rueYeh, afternoon
12:24:56ruedbussink: Sweet
12:25:01evanrue: whats up with the compile_if
12:25:04evanit's GROSS.
12:25:09evanthe catch/throw stuff makes my tummy ill.
12:25:27evanand it must slow down complication, having to throw constantly
12:25:39rueWhat, it is the perfect use case for catch/throw! :D
12:25:44evanno it's not!
12:25:49evani can't even follow it
12:25:59rueIt is a bit iffy, though. I do not think compilation was affected, let me check
12:26:09evanwhats the whole point?
12:26:30evanthe ability to insert zero cost debug code?
12:26:36rueevan: Oh, while you are around--can you push the new stuff in a branch?
12:26:40dbussinkevan: how's the new vm architecture going?
12:26:46evanit's going
12:26:54dbussinkeverybody is kinda anxious so see it :)
12:27:12evanwith traveling and talking and such, i've haven't had much time to work on since I left
12:27:28ctennis enters the room.
12:27:32rueEnjoy the vacation, good to get a break
12:27:52dbussinkno hurry, there is just a lot of curiosity :)
12:28:02evanrue: anyways, i think we have to figure out a better way to handle what you have that with catch/throw
12:28:12evani can't follow it
12:28:28rueevan: It simply drops code that is not wanted at all at runtime
12:28:42evani can't follow how the use of catch/throw is used
12:28:48evanor why it's needed at all.
12:29:27w1rele55 enters the room.
12:29:28rueevan: Can certainly change that part--the way it works is that Rubinius is detected in :const
12:29:40evanthats wrong
12:29:45evanit should be detected in call
12:30:01evansame as the primitive plugin
12:30:35rueSure, that would work
12:30:48evanif you do it in call, you have access to the block
12:30:58evanthen you don't need any of the catch/throw madness
12:32:02headius leaves the room.
12:32:23evani don't see a reason it has to operate on the sexp either
12:32:29evanit should work like the other plugins and operate on the AST
12:32:50rueOh, wait, I did put it in :call. But this is the AST processing, not bytecode generation so it will not actually have full info on the block until later
12:33:01evanwho cares?
12:33:14evanit should be in AST generation.
12:34:18evanif it has to be in sexp, i think we need to add a simple sexp macro system, to perform sexp mutations before doing the AST conversion
12:34:23evanthat would make the code much more sane
12:34:28evanand not so hacked in
12:34:34rueThe throw part is rather simple, though. It goes up to one of two catches: one is in Newline (default Node#consume), other is in :iter
12:34:51ariekeren leaves the room.
12:34:53evanthat feels like an INCREDIBLE hack
12:34:57evanwhy those 2 nodes?
12:35:22evani feel like the plugin is not a plugin at all
12:35:33evanbut rather that the logic for using it is now laced into the compiler itself
12:35:43rueThat is where the block will be housed
12:35:53rueTrue enough, I suppose
12:36:02evanthats why you just use a call plugin
12:36:07evanbecause it gets all the information
12:36:19evanwithout having to hack it hook points at a dozen random places
12:36:34headius enters the room.
12:37:26evanthis plugin should be almost EXACTLY like assembly plugin
12:38:08webmat enters the room.
12:38:26evanbecause you then have the ability to conditionally generate bytecode for the block
12:38:34evanoh, i think i see
12:38:41evanyou don't want the code in the block to actually be in a block
12:38:49evanyou want it to be embedded in the context the block is in
12:39:11evanand thus thats why you wanted sexp access
12:40:00headiushow long until you're home, evan?
12:40:14evanwe fly home on sunday
12:41:12evanwhy?
12:41:23headiusyou should go up to bergen
12:42:11headiussupposed to be pretty nice...my inlaws took a trip up there a few months ago
12:42:23evani think we're going to spend the days in oslo
12:42:28evanholy cow it's expensive
12:42:30evanYIKES
12:42:43headiusyeah, it's absurd
12:43:03headiusdenmark is bad, but norway is downright painful
12:43:18dbussinknorway is really nasty
12:43:20rueevan: Right. You can strip it out of the block later too but it is a bit more complex
12:43:23dbussinkprice wise
12:43:28rueIt is all that oil money :)
12:43:33evanrue: i think the whole thing should jsut go away
12:43:34scooprevan in norway?
12:43:45evanrue: and if you really want it, do it simply
12:43:47dbussinkcheap dollar doesn't help either for you guys
12:43:53evanrue: ie, don't worry about the block scoping or anything
12:43:57evanjust use a normal call plugin
12:44:13evanand output the whole thing if the gvar is set
12:44:25evanand write a Rubinius.compile_if in the runtime
12:44:28evanthat just calls the block
12:44:35evanmuch much much much much simpler
12:44:44evanscoopr: yep
12:44:54scooprneat
12:45:12scooprmake a detour to finland, I'll offer you a coffee/beer ;)
12:45:17evanheh
12:45:20headiusyou're in finland?
12:45:26headiuswhereabouts?
12:45:28evanif we were in the region longer, I might have
12:45:49scooprvantaa, next to helsinki
12:46:10headiusI'd love to get to finland some time...not a lot of conferences over there though
12:46:16rueevan Well, a bit simpler sure. Tradeoff though. Sexp macros was on the radar but the problem then is that it requires two passes over the sexp which seems wasteful since we are already going to be doing the exact same thing later
12:46:36rueevan: Mind you, I am not disputing that a cleaner implementation would be in order
12:46:40evanrue: well, the current approach is not extensible
12:46:50evananother plugin can't reuse it
12:46:57evanit's too tied to the way that compile_if needs
12:47:02scooprwe've been having some rubybrigade.fi meetings, but those have been small, would love to get you presenting though! :P
12:47:25dbussinki'm for a rubinius sprint at least somewhere in europe :)
12:47:27evani'm fine having sexp macros, and we can do them in one step
12:47:30evanone pass
12:48:39evanCompiler.add_macro :call, :matching => [[:const, Rubinius], :compile_if]] do |input| ... end
12:49:09evanthen consume can easily check if a macro wants to try and expand each sexp as it sees it
12:49:14evanno catch/throw
12:49:42rueBut there is still the AST building or do I misunderstand?
12:49:43evancould even just eliminate the :matching bit and do some checking in input
12:50:09evanthis would occur before the AST was generated
12:50:18rueRight, that is the second pass I meant
12:50:20evanthe AST would be generated from the output of the macro
12:50:26evanum no.
12:50:44evanthen we have 2 passes now
12:51:05hassox enters the room.
12:51:07evanpass one converts the sexp into an AST, pass 2 converts the AST into bytecode
12:51:20rueYou have to go through the sexp in some way to implement the macro
12:51:30evanthrough?
12:51:56rueHow would .add_macro match that sexp snippet?
12:52:10evanadd_macro wouldn't
12:52:17evanit would just register the pattern and the block
12:52:46evanin a primary #consume method, it would try to match the registered macros to the current sexp
12:53:25lazyat0m enters the room.
12:53:29rueWhich is what I am doing now--again, not in a particularly generic manner but nevertheless
12:53:41evanlazyat0m: hello!
12:53:55lazyat0mevan: hello! everyone: hi!
12:54:20evanis sitting next to lazyat0m
12:54:24rueThat is exactly what it does except of course for generic we need to determine how far/deep the lookahead goes
12:54:26lazyat0mnow ignore me, i'm going to lurk until it looks like i can say something that makes me look smart :)
12:54:32ruelazyat0m: *wave*
12:54:48_martinS_ enters the room.
12:54:53evanrue: hm. well, we need to come up with a better way of wiring it in than catch/throw
12:55:09rueevan: All I use the catch/throw for is unwinding the parse tree without explicitly doing it
12:55:15evani'd fell better if even it was using normal exceptions
12:55:20evanrather than catch/throw
12:55:53rueAw, exceptions are not for control flow, catch/throw is! :) But sure, there is certainly opportunity for improvement
12:56:10evansure exceptions are for flow control
12:56:16evanand this is an exception
12:56:22evanraise AbortContinuingToFollow
12:56:24evanor whatever
12:56:26evan:)
12:56:31lazyat0mlol
12:56:42lazyat0mwe had a debate about this a while back
12:56:57lazyat0mi think the position we came up with is that exceptions are for the benefit of people using your library
12:57:04lazyat0mrather than something to use yourself
12:57:22lazyat0m(although often the "person using your library" is yourself. if that makes sense)
12:57:50rueHeh, to me exceptions are error conditions and catch/throw are non-error conditions. Theoretically you could use the same mechanism for both
12:58:00evanrue: anyways, we'll figure out how to improve it
12:58:03evani'm headed to another room now.
12:58:17evantalk to ya later.
12:58:38lazyat0m leaves the room.
12:58:47rueYep yep, have fun over there
13:01:49agardiner enters the room.
13:07:03rubbish enters the room.
13:07:58boyscout4 commits by Eero Saynatkari
13:07:59boyscout * Added doc/rubini.us/ to store static website texts.; 8c8b88d
13:08:00boyscout * Spec to check `A = 12; class A; end` raises TypeError. Works as is.; e88fdb6
13:08:01boyscout * Fix cpu_open_class() to check its argument is a Module.; 746ddba
13:08:02boyscout * Spec for const lookup: `A = 12; class A::B; end` should raise TypeError.; 3c0db09
13:08:19rueWith that, /bed
13:08:35rueVVSiz: Fixed + spec if you did not add one yet ^
13:09:21VVSizrue: nice, thanks! we'll fix it soon too
13:09:54JimMc enters the room.
13:10:14rueVVSiz: Did it get fixed in 1.9 too or did I just misread the commit?
13:10:41VVSizcurrently MRI 1.9 crashes on it
13:11:07rueruby: * insns.def (defineclass): check if cbase is a class or a module.
13:11:17rueI think they may have :D
13:11:21rueAnyway, nite/day!
13:11:26VVSizallright! yeah, they might. will check :)
13:11:28VVSizsee ya
13:13:00Fullmoon enters the room.
13:17:25rubuildius_amd64Eero Saynatkari: 8c8b88df5; 1845 files, 6271 examples, 22365 expectations, 0 failures, 0 errors; http://rafb.net/p/oswmWF36.html
13:20:55rubuildius_ppcEero Saynatkari: 8c8b88df5; 1845 files, 6274 examples, 22394 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/174564
13:22:27brainopia leaves the room.
13:22:35agardiner leaves the room.
13:24:34lazyat0m enters the room.
13:28:23headius leaves the room.
13:28:32lstoll leaves the room.
13:28:42RyanTM enters the room.
13:30:46_martinS_ leaves the room.
13:44:38lstoll enters the room.
13:55:20jtoy leaves the room.
14:05:53rubbish leaves the room.
14:11:20fbuilesv leaves the room.
14:13:58fbuilesv enters the room.
14:14:23fbuilesvevan: you there?
14:19:04dbussinkfbuilesv: no, he's gone now
14:19:39fbuilesvdamn, been trying to get him for like 2 weeks for bit but nothing, can anyone else update the keys to the repo?
14:20:14dbussinknot that i know of
14:20:31francoi1 enters the room.
14:20:38dbussinkbut he's going home this weekend, so somewhere next he's probably here more often again
14:20:53fbuilesvnice
14:20:57dbussinkfbuilesv: but if you have some stuff you want in, you can always post a patch on lighthouse
14:21:19fbuilesvdbussink: some more specs, cleaning them right now before upload
14:21:47rosejn enters the room.
14:22:45francoi1dbussink: Hi! I'm working on build:checkdeps again this morning
14:22:57dbussinkfrancoi1: ah ok
14:23:04macournoyer enters the room.
14:23:22francoi1macournoyer: Good morning!
14:23:59macournoyerhey francoi1 ! good morning
14:24:35macournoyerworking w/ us today?
14:24:48francoi1macournoyer: no, can't. Too much stuff to do elsewhere.
14:24:57francoi1macournoyer: such as getting my first Rubinius patch accepted :)
14:25:03macournoyeryay!
14:25:06macournoyerwhat is it about?
14:25:28djwhitt enters the room.
14:27:19lazyat0m leaves the room.
14:33:33wmoxam enters the room.
14:33:50AndrewO enters the room.
14:34:57JimMc leaves the room.
14:36:58kofno leaves the room.
14:37:07kofno enters the room.
14:42:24francoi1macournoyer: this patch is for dependency checking on build/install
14:42:30francoi1dbussink: http://rubinius.lighthouseapp.com/projects/5089-rubinius/tickets/462-patch-rake-build-checkdeps#ti cket-462-5
14:45:22dbussinkfrancoi1: do you have some different systems to test with?
14:45:41francoi1dbussink: No, but I might ask macournoyer if he wants to test it on OSX
14:45:47obvio171 enters the room.
14:46:10macournoyersure
14:46:26francoi1macournoyer: Thanks! Simply run rake build:checkdeps
14:47:34francoi1macournoyer: You can pull from git://github.com/francois/rubinius.git
14:48:22macournoyerI applied the patch
14:48:28macournoyerit prints I'm missing libtool
14:48:57macournoyerhttp://pastie.caboo.se/174596
14:48:59jtoy enters the room.
14:49:08macournoyerbut I can build and install
14:49:14jtoy leaves the room.
14:49:42scooprfor the record, $? in win32 doesn't have .success?
14:50:20kofno leaves the room.
14:50:28kofno enters the room.
14:51:20dbussinkfrancoi1: you probably want to silence stderr too
14:51:35dbussinkfrancoi1: and apparently not all accept the --version option
14:51:57francoi1dbussink: 2>/dev/null ?
14:52:08dbussinkfrancoi1: something like that yeah
14:52:10fbuilesvare current specs failing for anyone else?
14:52:43scoopr>/dev/null ain't that portable either >:)
14:53:13francoi1I have used open4 many times
14:54:24francoi1I should probably use open3 though, which is part of code
14:54:26francoi1I mean core
14:56:28d2dchat enters the room.
14:56:59scoopractually, on windows that whole thing has to be a bit different :/
14:58:03francoi1scoopr: Is Rubinius runnable on Win32 ?
14:58:07scooprnope
14:58:18moofbong enters the room.
14:58:42francoi1Okay, so at the moment, this bit of code can be left as-is and changed when we start targetting Win32 ?
14:58:52scoopryeah I guess
14:58:53francoi1I have a Windows machine I could test this on, so I might as well do the work now
14:59:13francoi1But I'd just like this to be stable before I touch Windows though
14:59:33francoi1Okay, open3 doesn't return a usable Process::Status
14:59:41francoi1Any bright ideas anyone ?
15:00:02francoi1I believe open4 does return a correct Process::Status
15:00:08francoi1But that means another dependency
15:00:38scooprthe thing is, backticks seem to be a tad lower level on windows than nix, on nix it starts a shell and returns with a returnvalue anyway, but on windows it throws errno errors when it can't find something
15:00:52francoi1I see
15:01:30scoopralso, gem is a .bat and for some reason it throws a different error on `gem`, but `gem.bat` would work :/
15:02:19francoi1Nowhere does it say that the list of dependencies must be the same on all platforms :)
15:02:24ruivaldo leaves the room.
15:02:24francoi1scoopr: Check this: http://github.com/francois/piston/tree/master/lib/piston/svn/client.rb#L54
15:02:30francoi1This is how I run commands in Piston
15:03:25rosejnAnyone have experience playing with the rubinius Actors?
15:06:11scooprfrancoi1, perhaps something like ret=`cmd` rescue nil ; if ret && ret.success? ; yay! ; end .. doesn't take care of the .bat issue though
15:06:45francoi1scoopr: Yes, but the dependencies on Windows might be "gem.bat msvc" and on Linux, "gem make"
15:10:22qwert666 leaves the room.
15:10:37qwert666 enters the room.
15:11:23scooprmingw all the way >:)
15:11:47chris2 enters the room.
15:12:25scooprbut yeah point taken
15:13:55qwert666_ enters the room.
15:13:58qwert666_ leaves the room.
15:15:02fbuilesvI've added some more specs for REXML in case someone wants to test/apply: http://rubinius.lighthouseapp.com/projects/5089-rubinius/tickets/467-spec-more-specs-for-rexml#tic ket-467-2
15:19:28francoi1dbussink: I have it working with open4
15:19:53francoi1dbussink: If you wish, add my public repos as a remote and pull the checkdeps-open4 branch
15:19:54JimMc enters the room.
15:19:55francoi1dbussink: git://github.com/francois/rubinius.git
15:20:30dbussinkfrancoi1: hmm, dunno about adding another dependency
15:21:02dbussinkfrancoi1: don't think people will like that
15:21:03francoi1dbussink: I know. I tried open3 (which is part of ruby), but I get no usable Process::Status back
15:22:24agile enters the room.
15:22:27dbussinkfrancoi1: yeah ok, but then the question is whether the dependency is worth it
15:23:05francoi1dbussink: When I ran "rake build" the first time, the message I ended up with was "grammar.c: file not found"
15:23:17francoi1dbussink: It took me a while before I found out that I didn't have bison
15:23:33dbussinkfrancoi1: well, maybe the solution is then to document it better instead of solving it like this
15:23:36francoi1dbussink: So either we check dependencies early, or we attempt to compile, and fail when the dependency can't be met
15:23:44dbussinkfrancoi1: i don't think the open4 dependency will be accepted
15:23:57dbussinkfrancoi1: especially because we don't depend on anything but ruby and rake now
15:24:03francoi1dbussink: No, I agree. But I'd like to attempt to anyway :)
15:24:23dbussinkfrancoi1: well, maybe hint to stuff and not abort things
15:24:24francoi1dbussink: Ah, that's nice then. But you already have many dependencies: bison, cc, make, and others...
15:24:38dbussinkfrancoi1: yeah, but we want to prevent each one we can
15:24:59francoi1dbussink: Okay, so back to square one. You know, I skimmed INSTALL before I ran rake build
15:25:25francoi1dbussink: There's a General Requirements in INSTALL
15:25:45francoi1dbussink: But I'm on Linux, and I thought "Oh well, I must have all of these, since I compile/install stuff all the time"
15:26:39dbussinkmaybe we can add some stuff in http://rubinius.lighthouseapp.com/projects/5089/installation
15:27:16dbussinkbecause in my experience the bison issue is the issue that occurs most often
15:30:03francoi1dbussink: Ok. Thanks for all the time you spent on this with me. I really appreciate. Too bad it didn't work out.
15:30:14obvio171 enters the room.
15:31:10dbussinkfrancoi1: well, don't look at my opinion as authoritative, but the thing here is that the solution shouldn't be more complex then the problem
15:31:35francoi1dbussink: No, no. I'm not faulting you, or anyone for that matter.
15:31:53francoi1dbussink: Compiling software is already complex as it is. Adding more dependencies isn't really a solution either, as you said.
15:31:56dbussinkfrancoi1: i don't want to scare away people who want to contribute, or the contrary :)
15:32:05qwert666 leaves the room.
15:32:06dbussinkon the contrary
15:32:15probablycorey enters the room.
15:32:56francoi1It's a question of finding the right compromise. backticks are OK, but stderr isn't silenced. rake's sh doesn't let us trap stdin, stdout, stderr, but leaves a usable process::status.
15:33:11francoi1open4 is an extra dependency. open3 doesn't have process::status.
15:37:33dbussinkfrancoi1: but is there stuff in documentation that you find lacking?
15:37:52skaar enters the room.
15:37:54dbussinkfrancoi1: because we could definitely add the missing bison stuff to the common build problems list
15:38:08francoi1dbussink: No, the documentation was very clear. My system didn't have a required program.
15:38:22francoi1dbussink: I'm guessing most people don't read much beyond the first 2 paragraphs of INSTALL
15:38:42francoi1dbussink: which is what I essentially did. I saw the dependencies list, and incorrectly guessed that I would have everything
15:38:59dbussinkyeah ok, but i've seen the question pop up once in a while and therefore i think it doesn't hurt to add it
15:39:51dbussinkfrancoi1: do you still have the output somewhere?
15:40:57francoi1dbussink: Yes, it's on pastie somewhere
15:41:04francoi1dbussink: I'll get it for you
15:45:02headius enters the room.
15:47:12tokengeek leaves the room.
15:47:25obvio leaves the room.
15:49:44francoi1dbussink: http://pastie.caboo.se/174633
15:49:48francoi1dbussink: Look at line #77
15:50:09francoi1dbussink: But the messages that are last on screen at #104 and forward
15:50:25francoi1dbussink: So I never even saw the "bison: command not found" message until much later
15:51:01dbussinkfrancoi1: thnx, i'll update the lighthouse page with this
15:51:26francoi1dbussink: NP. At least I'll have contributed to the documentation :)
15:53:00lopex enters the room.
15:55:08dbussinkfrancoi1: http://rubinius.lighthouseapp.com/projects/5089/installation
15:55:25dbussinkfrancoi1: is that clear enough?
15:58:47crossaway leaves the room.
15:59:13crossblaim enters the room.
16:00:39francoi1dbussink: Yes, and I updated the text slightly here: http://pastie.caboo.se/174638
16:00:56francoi1dbussink: You haven't "forgotten" to install bison, you're just missing it.
16:01:44dbussinkfrancoi1: ok, updated
16:01:50dbussinkfrancoi1: but i have to go now
16:01:57dbussinkfrancoi1: thnx for the help
16:02:25dbussinkfrancoi1: don't hold back to work on some parts in rubinius ;)
16:02:55francoi1dbussink: No no, I won't. I'm the guy behind Piston, and I was curious to see if Piston could be used.
16:03:06francoi1dbussink: So, I'll probably work on getting there at some snail's pace.
16:03:09francoi1dbussink: :)
16:03:11tokengeek enters the room.
16:03:30dbussinkfrancoi1: you mean whether it runs in rubinius now?
16:03:39therealadam enters the room.
16:04:12francoi1dbussink: Yes. I don't think it can currently.
16:04:31francoi1dbussink: I tried installing, but I had problems just getting RubyGems installed. I probably need a clean machine to test this on anyway.
16:04:42francoi1dbussink: "it" being Piston.
16:09:30dbussinkfrancoi1: i'm not an expert on the installing stuff, dunno how well that all works
16:09:37dbussinkfrancoi1: but i really have to go
16:09:39dbussinksee ya
16:13:49jartz leaves the room.
16:14:32anonuser leaves the room.
16:21:12rosejnIs there currently an IRB equivalent in rubinius?
16:21:36NoKarma enters the room.
16:21:44NoKarmafbuilesv: Hi
16:21:48NoKarmaHey all
16:22:38rosejnUsing a new IRC client here... can anyone see my messages?
16:22:47NoKarmarosejn: yep
16:22:52rosejnhaha, ok thanks
16:23:11rosejnany idea what the status of a rubinius IRB is?
16:23:19fbuilesvNoKarma: hey, how's it going
16:23:53NoKarmafbuilesv: everything's fine. I'm having exams these days, tho
16:24:06yaroslav enters the room.
16:24:08fbuilesvNoKarma: I'm running out for some multivariable calc exam right now too :P
16:24:08NoKarmarosejn: IRB is working fine in rubinius
16:24:17NoKarmafbuilesv: heh :)
16:24:23NoKarmafbuilesv: got a reply from pat?
16:24:31fbuilesvNoKarma: Nope, not yet
16:24:41NoKarmafbuilesv: ok
16:24:49fbuilesvNoKarma: they extended the deadline for another week
16:25:19NoKarmafbuilesv: ooh, I see
16:25:45agile leaves the room.
16:27:18rosejnNoKarma: ok, so I can use it in a script, but it looks like rubinius doesn't include an executable IRB?
16:27:19fbuilesvNoKarma: anyway, I'm out, gl on the exams :P
16:27:37NoKarmafbuilesv: Thanks, you too
16:27:58fbuilesvrosejn: check the original irb executable
16:28:01NoKarmarosejn: just start up shotgun -- ./shotgun/rubinius
16:28:07fbuilesvrosejn: it's like 5 lines long
16:29:36fbuilesv leaves the room.
16:29:41rosejnahhh, cool. Didn't realize shotgun/rubinius is also an IRB. Thanks.
16:30:06headiusit's the default behavior if you don't specify something to run
16:30:19headiuswhich makes me wonder...can you not pipe to shotgun/rubinius?
16:34:33rosejnyeah, you can pipe to it
16:34:55rosejnThis actually makes a lot more sense... I never understood the reason to have both "ruby" and "irb" commands.
16:35:55jlindley enters the room.
16:36:57rosejnSo I've been playing with Actors, and I've got some nice example code and some tests. It looks like the tests are extremely minimal. Do I just write-up some specs and post them to lighthouse?
16:37:54headiushmm, weird:
16:38:02headius~/NetBeansProjects/rubinius ➔ echo "puts 'foo'" | shotgun/rubinius
16:38:02headiusfoo
16:38:02headiusnil
16:38:35headiusis there some documentation on actors api? I was thinking about adding them to jruby tonight
16:44:15xmlhacker leaves the room.
16:44:16xmlhacker_ enters the room.
16:44:19loincloth enters the room.
16:44:22rosejnnope, not that I've seen
16:44:45rosejnthe Actor class and it's required Mailbox, Filter and Channel are pretty minimal though
16:45:01headiusI'll have a look there
16:45:48antares leaves the room.
16:49:37jxt enters the room.
16:53:03rosejn leaves the room.
16:55:27jxthi
16:56:03jxtI'm having difficulties building rubinius... followed this guide: http://rubinius.lighthouseapp.com/projects/5089/installation and get the following stack trace with rake build: http://pastie.caboo.se/174659
16:56:48jxtAnybody got some suggestions? :/
17:01:11obvio leaves the room.
17:01:22brixenjxt: looks to me like you need readline dev package installed
17:01:22Arjen_ leaves the room.
17:01:43ch0wda enters the room.
17:04:22brixenoff to class, bbl..
17:04:34jxtbrixen: works like a charm... thanks
17:08:06d2dchat leaves the room.
17:22:10mutle leaves the room.
17:24:24jxt leaves the room.
17:24:42joshbuddy enters the room.
17:24:57enebo enters the room.
17:26:39nicksieger leaves the room.
17:27:32yaroslav leaves the room.
17:30:01jayWHY enters the room.
17:33:20thehcdreamer leaves the room.
17:36:18yaroslav enters the room.
17:38:31yaroslav leaves the room.
17:39:10yaroslav enters the room.
17:40:57nicksieger enters the room.
17:42:21justinweiss enters the room.
17:42:43NoKarmawooo
17:42:48NoKarmafound a bug in kind_of
17:44:25agile enters the room.
17:45:13imajes enters the room.
17:47:28qwert666 enters the room.
17:48:55yaroslav leaves the room.
17:57:45NoKarmaHey, yo, guys, is there some special mojo I have to do before checking in a failing spec?
17:58:37headiusput your lips together and blow
17:59:45NoKarmawell, either that or see the lighthouse wiki before bothering people with stoopid questions :)
18:00:01NoKarmaheadius: gonna try your suggestion first, tho
18:00:14headiusexcellent
18:02:43nicksieger leaves the room.
18:03:19eventualbuddha enters the room.
18:04:13nicksieger enters the room.
18:05:49wycats enters the room.
18:06:45hassox enters the room.
18:09:23boyscout1 commit by Arthur Schreiber
18:09:23boyscout * Add specs for #kind_of? / #is_a? behaviour that are failing in Rubinius.; 4a9cb7c
18:09:31NoKarmawoo, it worked!
18:13:58enebo leaves the room.
18:14:04headius leaves the room.
18:20:41headius enters the room.
18:20:55rubuildius_amd64Arthur Schreiber: 4a9cb7cc0; 1845 files, 6271 examples, 22365 expectations, 0 failures, 0 errors; http://rafb.net/p/9ul3HK32.html
18:21:33rubuildius_ppcArthur Schreiber: 4a9cb7cc0; 1845 files, 6274 examples, 22394 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/174720
18:21:38headius leaves the room.
18:24:16NoKarmawoo, it worked!
18:27:24loincloth_ enters the room.
18:29:50moofbong leaves the room.
18:30:42moofbong enters the room.
18:31:11rubbish enters the room.
18:35:00imajes leaves the room.
18:35:48TheVoice enters the room.
18:39:19VVSizheh, Object#is_a spec is wrong
18:39:40VVSizit has description of Object#kind_of
18:41:45NoKarmais_a is an alias of kind_of
18:42:43ko1_away enters the room.
18:43:36dgtized leaves the room.
18:43:55VVSizright, but the spec description still should be Object#is_a, not Object#kind_of
18:53:26loincloth leaves the room.
18:54:28loincloth enters the room.
18:56:05d2dchat enters the room.
18:56:33anonuser enters the room.
19:00:15benburkert enters the room.
19:01:54dgtized enters the room.
19:03:54loincloth leaves the room.
19:05:03boyscout1 commit by Vladimir Sizikov
19:05:04boyscout * Fixed copy-paste error in Object#is_a? specs.; 0e7d1c6
19:10:36rubbish leaves the room.
19:15:10bitbang enters the room.
19:15:54rubuildius_amd64Vladimir Sizikov: 0e7d1c6e0; 1845 files, 6271 examples, 22365 expectations, 0 failures, 0 errors; http://rafb.net/p/DGkHUw62.html
19:21:00rubuildius_ppcVladimir Sizikov: 0e7d1c6e0; 1845 files, 6274 examples, 22394 expectations, 0 failures, 0 errors; http://pastie.caboo.se/paste/174748
19:21:05Fullmoon leaves the room.
19:23:19eventualbuddha leaves the room.
19:23:38eventualbuddha enters the room.
19:32:18thehcdreamer enters the room.
19:38:15bitbang_ enters the room.
19:38:30bitbang_ leaves the room.
19:39:53imajes enters the room.
19:40:01thehcdreamer leaves the room.
19:52:52bitbang leaves the room.
19:53:59imajes_ leaves the room.
20:06:46jlindley leaves the room.
20:09:45jartz enters the room.
20:11:28dodecaphonic leaves the room.
20:15:34rueHuh, how in the hell did that bug get in #kind_of?
20:19:37VVSizah, what?
20:20:27VVSizrue: just checking that you're not saying that I should revert the spec change for #is_a? / #kind_of?
20:22:03ch0wda leaves the room.
20:22:10be9 leaves the room.
20:22:57dbussinkhmm, anything changed on the gitweb interface?
20:22:59dbussinkit's giving me xml parse errors
20:23:54kofno leaves the room.
20:25:26rby enters the room.
20:30:54nicksieger leaves the room.
20:30:59benburkert leaves the room.
20:31:30nicksieger enters the room.
20:32:37brixendbussink: http://git.rubini.us looks ok to me
20:32:44VVSizworks fine here too
20:33:22rueVVSiz: No, I am wondering how the bug has managed to be there this long
20:33:32VVSizoh :)
20:38:47Defilerrue: Which bug?
20:39:34rueDefiler: #kind_of? does not check the metaclass for ancestors, it looks like. Fixing
20:40:34Defileroof
20:50:38dbussinkbrixen: strange, what browser?
20:50:49dbussinkbrixen: i get a xml parse error with both safari and firefox
20:51:05dbussinkbrixen: on the same line too
20:52:03brixendbussink: what's the full url?
20:52:22dbussinkbrixen: just http://git.rubini.us/
20:52:22brixenI'm in ff, but I can try safari too
20:53:02brixenno error in either
20:53:13d2dchat_ enters the room.
20:54:04dbussinkbrixen: what does line 273 say in the source?
20:54:32brixen273 in source of what?
20:54:40brixenoh, in the page, heh
20:55:11loincloth_ leaves the room.
20:55:13brixenno line 273 in mine
20:55:56dbussinkhmm, weird stuff, the page is served as application/xhtml+xml here and it borkes on that
20:56:12brixendbussink: http://pastie.org/174823
20:56:15brixenthat's my page source
20:56:24dbussinkmaybe it's the squid proxy that in between, i'm using my 3g network connection right now
20:56:54rueFine here too
20:56:59brixenhmm
20:57:08brixenlunchin' bbiab...
20:57:10dbussinkfound it
20:57:37dbussinkapparently there is something called NG cache in between which inlines some CSS here
20:57:40dbussinkand it borks on that
20:57:44dbussinknasty stuff
20:59:14benburkert enters the room.
21:01:37d2dchat leaves the room.
21:17:41kirindave enters the room.
21:19:20ctennis enters the room.
21:25:46ezmobius enters the room.
21:29:26jicksta enters the room.
21:31:12wycats_ enters the room.
21:32:38wycats leaves the room.
21:45:33ch0wda enters the room.
21:46:15jartz leaves the room.
21:50:25wycats leaves the room.
21:52:49nicksieger leaves the room.
21:57:06nicksieger enters the room.
21:59:30dysinger enters the room.
22:06:49rue<