Index

Show enters and exits. Hide enters and exits.

00:00:01agardinergotta run to a meeting... bbiab
00:01:51cremesbrixen: check your email
00:02:09drbrainmspec -tr is broken when nothing fails
00:02:22drbrainmspec/bin/mspec-run:35:in `exit': no implicit conversion from nil to integer (TypeError)
00:02:32drbrainbin/mspec -tr spec/ruby/1.8/core/io/eof_spec.rb reproduces this for me
00:05:03brixenthat's odd, nil.to_i => 0
00:05:37brixenah, didn't add it to mspec-run, just mspec-ci
00:05:48moofbong leaves the room.
00:07:00ruenil.to_i is by definition not implicit, brixen :D
00:09:25brixencremes: no luck on the pw
00:10:27cremesbrixen: odd... i just did 'su - brixen' and used that pword successfully...
00:10:43brixencremes: ok, le'me try again
00:11:36boyscout1 commit by Brian Ford
00:11:37boyscout * Make sure exit gets a number in mspec-run script.; b1b0af2
00:12:57cremesbrixen: my mistake... i had allowed only myself and dbussink as remote logins... just added you to the allow list
00:13:14brixenahh, ok
00:14:59AndrewO_ enters the room.
00:16:39brixencremes: all good, thanks!
00:16:50cremesbrixen: my pleasure!
00:16:56evanok, i've going to carve Hash out of the VM entirely
00:17:03evanit's going to exist only in Ruby land
00:17:12evanthe VM will exclusively use LookupTable
00:17:21cremeswow, Hash is going to be a turtle...
00:17:42evanand LookupTable is only going to allow symbols as keys
00:18:11wmoxam leaves the room.
00:21:30AndrewO leaves the room.
00:23:43ezmobiusevan: talked with tom finally, his meeting lasted for 5 hours
00:23:47ezmobiusnow he's on a plane
00:24:05evanyikes, ok.
00:24:08rubuildius_ppcBrian Ford: b1b0af241; 1573 files, 5491 examples, 18955 expectations, 0 failures, 28 errors; http://pastie.caboo.se/154548
00:24:26ezmobiusso iguess we'll have to try for tomorow?
00:24:46evanwe'll have to confirm with rue
00:24:48evanrue?
00:24:54rubuildiusBrian Ford: b1b0af241; 1573 files, 5490 examples, 18983 expectations, 0 failures, 5 errors; http://rafb.net/p/3uqBJs78.html
00:25:57rueIf it is getting too late, then early or late tomorrow, I will be out most of the day probably
00:27:07evanwell
00:27:14evanshould we just have the call without tom then
00:27:58evanlike right now
00:28:20ezmobiusim game
00:28:30evanrue?
00:29:00ezmobiusevan: , rue http://www.perlfoundation.org/parrot/index.cgi?mod_parrot http://www.smashing.org/talks/future_of_mod_perl.pdf
00:29:04ezmobiusinteresting reading
00:29:27rueFine by me, sounded like Tom wanted in though
00:30:08ezmobiuswe could have a quick call now, and then another with tom later
00:30:24ezmobiusor whatever you gusy think is best
00:30:35evanlets have a short call now
00:30:41ezmobiusk calling in
00:30:42boyscout1 commit by Eric Hodel
00:30:43boyscout * Add library to read/write ar(5) files; 77fdbe4
00:31:15rueBe there in a bit
00:31:52evanrue: go ahead and call in
00:32:01AndrewO enters the room.
00:36:31ragnard leaves the room.
00:38:23AndrewO_ leaves the room.
00:41:14grazamass: hi, can i ask a question about something you fixed back in Oct?
00:41:35cremes leaves the room.
00:42:17d2dchat enters the room.
00:42:41masshmm?
00:43:23defunkt leaves the room.
00:43:38masswhats up graza
00:43:55rubuildius_ppcEric Hodel: 77fdbe404; 1573 files, 5491 examples, 18955 expectations, 0 failures, 28 errors; http://pastie.caboo.se/154562
00:44:40grazaundefined symbols _getcontext etc when compiling on Tiger
00:44:40grazai'm getting that error now... :(
00:44:40grazahttp://pastie.caboo.se/pastes/153965
00:44:56rubuildiusEric Hodel: 77fdbe404; 1573 files, 5490 examples, 18983 expectations, 0 failures, 5 errors; http://rafb.net/p/SbLjFM18.html
00:45:54kofno enters the room.
00:46:20drbrainwhat is up with rubuildius?
00:47:05masson ppc graza?
00:47:15grazamass: yep
00:47:47masshmmm
00:47:54drbrainah, all the MD5/SHA1 problems are endian issues
00:47:59drbrainlook at failure 28
00:48:11drbrainBasicSocket#getsockopt gets a socket option ERROR
00:48:13masssilly ends
00:48:19drbrainExpected "\000\000\000\001" to equal "\001\000\000\000":
00:49:55massgraza: do you have a export MACOSX_DEPLOYMENT_TARGET=10.4 in config.mk ?
00:50:49drbrainevan: any insight into the rubuildius_ppc endian problem?
00:51:54grazamass: sure do
00:52:00masshmmm
00:52:43AndrewO leaves the room.
00:52:50evandrbrain: the spec perhaps expects a certain endian-ness?
00:53:12drbrainevan: look at that last failure
00:53:23massgraza: someone changed PortableUContext.h since me, I think that might be whats going on but my C-fu is not up to snuff right now, a bit flu-like here :(
00:53:26cored enters the room.
00:53:34massmy last ver was 8802fb9fd0e33b9ba0befc362013a6d0dea89b92
00:54:06headius enters the room.
00:54:18drbrainevan: also, I know for sure that "d41d8cd98f00b204e9800998ecf8427e" is the MD5 for an empty string
00:54:28drbrainbut, it's being compared against something entirely different
00:55:01grazaok, thx mass. i'll keep looking here
00:55:28mass__DARWIN_UNIX03 is new in leopard, so it should be undefined previously
00:55:51massbut c2e2786d changed it to be a comparison against 0 , which I _think_ will never be true
00:55:55massbut my C fu is lacking :(
00:56:05masswaves hello to evan while he's stopped in the channel :)
00:56:11evan:D
00:56:15evanon the phone with rue and ezmobius
00:56:18wvl leaves the room.
00:56:52massright now this flu is really mild, I'm trying to decide how much I want to push myself if it risks me getting sicker and having more symptoms
00:56:58masskinda a "how stupid do I want to be" think :)
00:57:22massbut I've been relaxing excessively since Sunday, and am getting restless
00:58:41massoh, and evan - still no updates on my getcontext bug reported against leopard. I'll poke them after I see 10.5.3 dev seeds
00:58:47evanok
00:58:48evani saw that
00:58:49evanthanks
00:59:02evanwe might be ditching using ucontext in the near future
00:59:10massI think defiler has done a lot more experimentation on the ways of its stupidity anyway :)
00:59:15massreally? fun
00:59:37nicksieger enters the room.
01:00:00massI guess its main purposes are signal handling and userland threading
01:01:11massby having a nil object rather than setting references to null/0, you remove the need to have custom sigsegv handling
01:04:38RyanTM leaves the room.
01:06:50jtoy enters the room.
01:08:41RyanTM enters the room.
01:11:48evandrbrain: whats up with line line 116 of ar.rb?
01:12:20drbrainso the ljusts are all aligned
01:12:37evanit just looks so... odd
01:12:42evanbecause ljust is being called on name, yes?
01:12:44drbrainit does
01:12:50drbrainyes
01:12:57evanok
01:14:43skaar leaves the room.
01:15:04vintrepid leaves the room.
01:15:58drbrain120 has a stupid extra space
01:20:21lopex leaves the room.
01:20:31agile leaves the room.
01:21:06Fullmoon enters the room.
01:21:30rueAway for a bit
01:21:45evank
01:23:37jrun_ enters the room.
01:28:18dysinger leaves the room.
01:29:28drbrainevan: http://rafb.net/p/sj5XPv91.html
01:29:43drbrainevan: those should make `git stash` not fill up any more
01:29:57drbrainand, should properly switch back to the topic branch at all times
01:30:02drbraindo they look right?
01:31:08headius leaves the room.
01:33:30Defilerbinary is awesome
01:33:40DefilerI love the properties of integers expressed in binary
01:37:16drbrainDefiler: can you look at the PPC failures for rubuildius?
01:37:22drbrainI strongly suspect endian issues
01:37:41DefilerI don't have access to a PPC box, but I am happy to look at the error messages
01:38:34DefilerHuh., Digest::MD5#== should be equal to appropriate object that responds to to_str
01:38:39Defilerin particular is a really strange failure
01:38:45drbrainDefiler: look at failure 28
01:39:19DefilerIndeed.
01:39:31drbrainand, I know that for failure 24, "d41d... is the MD5 of an empty string
01:39:48crafterm leaves the room.
01:39:52drbrainso the expected value seems completely off
01:39:56DefilerOh, yeah.. looking at the code for getsockopt, it doesn't do anything to attempt to handle endianness
01:40:13DefilerOn the other hand, I'm not sure it is meant to.. we really need MRI 1.8.6 CI
01:40:20DefilerSo we can tell when we check in a retarded spec
01:40:27drbrainyeah
01:41:10DefilerDoes IBM still sell POWER workstations?
01:41:23DefilerOoh yes
01:41:36technomancy leaves the room.
01:41:49Defilerhttp://www-03.ibm.com/systems/intellistation/power/hardware/285/index.html
01:41:53drbrainstarting at $5,999*
01:41:57DefilerI smell a business expense
01:43:57brixenDefiler: there are only 2 MRI spec failures on socket on ppc: Socket#pack_sockaddr_un packs and unpacks => string contains null byte
01:44:04brixensame for Socket#sockaddr_un
01:44:31DefilerOK, so apparently 1.8.6 normalizes those results to little endian?
01:44:37DefilerUnless there are lame guards
01:45:10DefilerNope. No guards
01:45:22brixenthere are no Digest failures
01:45:43DefilerI know 'i' is the pack option for Integer..
01:45:47drbrainI'm wondering if it's an FFI thing
01:45:52Defiler..but does anyone remember if that specifies endianness?
01:46:06drbrainI think i is native endianness
01:46:29drbrain... well, MD5 uses pack/unpack in places too
01:46:35RyanTM leaves the room.
01:46:47DefilerOdd that that spec doesn't fail, then
01:47:52dysinger enters the room.
01:48:44jeremydurham enters the room.
01:49:19rueIt could be magic
01:53:44cjheath leaves the room.
01:54:15mad_phoenix enters the room.
01:54:35sudothinker leaves the room.
01:57:47jeremydurham leaves the room.
01:57:47jeremydurham_ enters the room.
02:00:53DefilerLooking at the matzruby code, it just does whatever the platform version does
02:02:02DefilerWhich is (hopefully) what we do
02:02:23DefilerSo if we have an endianness problem here, it might be at the FFI level, which I really hope isn't true
02:04:14headius enters the room.
02:07:03VVSiz_ enters the room.
02:11:19claudio_ leaves the room.
02:12:06hornbeck leaves the room.
02:14:36headius leaves the room.
02:15:20mad_phoenix leaves the room.
02:19:24eventualbuddha leaves the room.
02:19:25wvl enters the room.
02:25:05VVSiz leaves the room.
02:27:21rueYay
02:27:22graza leaves the room.
02:28:04graza enters the room.
02:29:06dysinger leaves the room.
02:29:22cored leaves the room.
02:31:10nicksieger leaves the room.
02:31:39bburcham enters the room.
02:31:44bburcham leaves the room.
02:35:52sudothinker enters the room.
02:37:31crafterm enters the room.
02:41:18agile enters the room.
02:46:22_VVSiz_ enters the room.
02:48:10rueThis new lightbulb is totally messing me up
02:48:58rueI got one of the 150W-equivalent spiral Reveal lamps. It looks *exactly* as if all the windows were open during daytime when I look into the living room
02:49:05rueI keep checking the time
02:50:53_mutle enters the room.
02:51:07AndrewO enters the room.
02:51:34brixenrue: that's not a reason to stay awake 24 hrs a day :P
02:52:32rueTell it to my sleep regulatory glands
02:52:46ruePlus I did sleep from 7-1 today
02:52:49gdagley leaves the room.
02:54:06brixenwell, I know there are people whose biochemistry basically makes them be awake at night
02:54:14squeegyrue: i have a couple pretty intense full spectrum bulbs or whatever they are
02:54:40squeegyone is a bulb, another is an entire desklamp type thing with a special fixture
02:54:43VVSiz_ leaves the room.
02:54:50brixenbut I've also read that disturbed sleep will impaire your body's ability to regulate blood sugar by 25% in as few as 2 days
02:55:19squeegyinteresting
02:55:33brixenan insulin insensitivity is a pre-condition for diabetes
02:55:40djwhittheh, if that's the case I don't think I've had well regulated blood sugar for about 7 years
02:55:40squeegyclasses make keep me on a regular schedule
02:55:46squeegyjeez
02:55:53squeegyclass make me keep a regular schedule.
02:56:07squeegyand that still isn't right.
02:56:28brixenheh, sometimes fingers just rebel against that whole typing thing
02:56:46squeegyindeed...
02:57:06djwhitthe's doing some kind of turing test I think
02:57:20djwhittinjecting random mistakes into his messages
02:57:26brixenheh
02:57:26djwhitttrying to see if we can still decifer them
02:59:47squeegyheh
03:00:33djwhitt*decipher
03:07:00mutle leaves the room.
03:07:01nemerle enters the room.
03:15:50rueOh no, it is spreading
03:16:12squeegy?
03:16:16srbaker leaves the room.
03:17:34tongueroo leaves the room.
03:17:54mfp leaves the room.
03:19:23rueThe turing misspelling virus
03:19:57squeegyi must have got infected today!
03:20:00squeegyrun away!
03:20:35graza leaves the room.
03:20:39graza_ enters the room.
03:22:06KirinDave leaves the room.
03:24:45nemerle_afk leaves the room.
03:25:13geekounet leaves the room.
03:31:48bitbckt enters the room.
03:33:29srbaker enters the room.
03:34:06hornbeck enters the room.
03:38:21_mutle leaves the room.
03:38:24mutle enters the room.
03:39:10mad_phoenix enters the room.
03:44:50graza enters the room.
03:44:51graza_ leaves the room.
03:53:02mad_phoenix leaves the room.
03:53:46ezmobius leaves the room.
04:05:38jrun_ leaves the room.
04:05:39graza leaves the room.
04:06:05graza enters the room.
04:12:44gdagley enters the room.
04:18:07graza_ enters the room.
04:18:08graza leaves the room.
04:20:25Fullmoon leaves the room.
04:33:40bitbckt leaves the room.
04:35:24bitbckt enters the room.
04:39:43tbmcmullen enters the room.
04:42:04graza enters the room.
04:42:05graza_ leaves the room.
04:50:09AndrewO leaves the room.
04:54:43graza_ enters the room.
04:54:44graza leaves the room.
04:55:08dysinger enters the room.
04:59:05binary42 leaves the room.
05:03:59hornbeck leaves the room.
05:15:57graza enters the room.
05:15:58graza_ leaves the room.
05:24:48kofno leaves the room.
05:31:34graza_ enters the room.
05:31:35graza leaves the room.
05:44:08graza enters the room.
05:44:09graza_ leaves the room.
05:48:08ruetarcieri: What were your thoughts on Rev + Rubinius?
05:50:18RyanTM enters the room.
05:55:51ruejero5_: Do you not have a commit bit?
05:57:01graza_ enters the room.
05:57:02graza leaves the room.
05:57:33tizianobis enters the room.
06:05:05tizianobis_ enters the room.
06:05:40tizianobis_ leaves the room.
06:07:28_ADS enters the room.
06:16:17GMFlash leaves the room.
06:16:23therealadam enters the room.
06:16:23GMFlash enters the room.
06:17:37jero5rue: i'm on a fresh linux install, and i'm too lazy to setup the ssh stuff again. or i just want a filter between me and shotgun. :)
06:18:25boyscout1 commit by Brian Ford
06:18:26boyscout * Added mspec-tag script. Misc MSpec fixes.; 30ff4c0
06:18:57tizianobis leaves the room.
06:19:01graza_ leaves the room.
06:19:32RyanTM leaves the room.
06:19:39geekounet enters the room.
06:19:42graza enters the room.
06:19:44tbmcmullen leaves the room.
06:26:12tbmcmullen enters the room.
06:27:25tbmcmullenargh. Is there a quick way to tell whether Evan gave me the commit bit yet? I can't tell if I don't have the bit or if I'm just screwing up setting something up.
06:28:57bennyI guess if you can check out as a git:committer you can commit
06:29:44bennythat's just a guess though
06:29:52rubuildiusBrian Ford: 30ff4c009; 1573 files, 5490 examples, 18983 expectations, 0 failures, 5 errors; http://rafb.net/p/spM8q972.html
06:30:28rubuildius_ppcBrian Ford: 30ff4c009; 1573 files, 5491 examples, 18955 expectations, 0 failures, 28 errors; http://pastie.caboo.se/154649
06:31:11brixentbmcmullen: does your .git/config have url = git@git.rubini.us:code under [remote "origin"] ?
06:32:39tbmcmullenbenny: can't check out... it prompts me for a password, then tells me its wrong
06:32:44tbmcmullenbrixen: checking...
06:33:25tbmcmullenbrixen: I don't even have a "remote origin" section... sooo... yeah.
06:33:59headius enters the room.
06:35:00brixentbmcmullen: if you did git clone git://git.rubin.us, do rake git:committer in that dir
06:36:01tongueroo enters the room.
06:36:34tbmcmullenbrixen: I will pull it down again and try that. Thanks for the help.
06:36:46brixenok
06:37:13brixentbmcmullen: you can try git clone git@git.rubini.us:code
06:37:28tbmcmullenbrixen: that was what I tried initially, and it failed
06:37:35brixenah, ok
06:37:38brixenI'd ping evan then
06:38:03tbmcmullenbummer, okay then... I guess I don't commit new specs tonight. heh.
06:40:57boyscout4 commits by Adam Gardiner
06:40:57boyscout * Switch SendSite.create to use primitive, new stables; 8bd0fc6
06:40:57boyscout * Switch to using SendSite#at primitive; 26a91a4
06:40:57boyscout * Define new primitives for SendSite; cde6bc3
06:40:58boyscout * Remove unused lookup_func parameter from send_site_create; cb58797
06:41:12agardinerwhew! that was harder than it should have been!
06:41:19nicksieger enters the room.
06:41:59headiusevening friends
06:42:36agardinerlater...
06:42:39agardiner leaves the room.
06:44:23crafterm leaves the room.
06:47:31enebo enters the room.
06:53:09rubuildius_ppcAdam Gardiner: 8bd0fc604; 1573 files, 5491 examples, 18955 expectations, 0 failures, 28 errors; http://pastie.caboo.se/154659
06:53:56brixenhmm, seekdir with a value not returned by telldir appears to be undefined (i.e. platform specific)
06:54:06brixenI don't think we should spec that
06:54:44rubuildiusAdam Gardiner: 8bd0fc604; 1573 files, 5490 examples, 18983 expectations, 0 failures, 5 errors; http://rafb.net/p/F7cTC496.html
07:00:52_ADSbrixen: could we spec something like 'seek with a position outside of the range of positions returned by tell'
07:03:49brixen_ADS: it doesn't really make sense to, since the behavior is really undefined
07:04:08brixenthe man page for seekdir says "seekdir() should be used with an offset returned by telldir(3)"
07:04:27brixenand I get different values on different linuxes (e.g. ubuntu and gentoo)
07:05:00brixenif someone's code does that, their problem, not really the implementation's concern
07:05:19_ADSdifferent values? I thought almost all implementations just returned the byte offset.
07:05:29contextcompiler / architecture / 32/64 bit
07:05:42_ADSbut I see your point.
07:06:07_ADSoh, wait this is seekdir, not fseek. nevermind :)
07:08:26contextbrixen: umm read the man page dude
07:08:35context The telldir() function returns the current location associated with the named
07:08:39context directory stream. Values returned by telldir() are good only for the lifetime of the
07:08:42context DIR pointer (e.g., dirp) from which they are derived. If the directory is closed and
07:08:46context then reopened, prior values returned by telldir() will no longer be valid.
07:08:50contextREAD SLOWLY
07:09:53contextoops
07:10:44contextbrixen: from how the man page words it, you shouldn't be expecting anything static and consistent
07:13:27brixencontext: yup, I think we're on the same page there
07:14:38contextsrry sounding harsh, im use to completely dumb people in #c on efnet having troubles with hello world :/
07:15:10contexti get into a moody mindset :x does bad things
07:22:41yipstar leaves the room.
07:25:05srbaker leaves the room.
07:25:26graza leaves the room.
07:28:44cavalle enters the room.
07:32:39TheVoice leaves the room.
07:33:04srbaker enters the room.
07:40:44lstoll leaves the room.
07:42:28srbaker leaves the room.
07:42:43_ADSrbx -e "p 1<<(2**28)"
07:42:53_ADSdon't try this at home
07:42:59_ADSit's been running for 20 minutes
07:43:10evanug.
07:43:29evanthe sendsite code is busted bigtime.
07:43:31evandamnit adam.
07:45:58brixenthe File.truncate problem on ppc looks like it's related to FFI, I make a primitive and it worked ok
07:46:17brixenI don't see a problem with the attach_function signature
07:46:55tongueroo leaves the room.
07:48:56brixenFFI.type_size returns the correct size for :off_t (8) based on a C sizeof call
07:54:04enebo leaves the room.
07:55:14enebo enters the room.
07:55:41srbaker enters the room.
07:56:09_ADS leaves the room.
07:58:20srbaker leaves the room.
07:59:28KirinDav enters the room.
08:00:21dbussinkmorning
08:00:48VVSizmorning!
08:00:55imajes leaves the room.
08:01:40defunkt enters the room.
08:01:40dbussinkbrixen: having ffi problems?
08:02:17dbussinkbrixen: there is probably still something broken there, they are always a bit problematic
08:02:18brixendbussink: not sure, was trying to track down the File.truncate issue on ppc
08:02:28brixenI made a primitive and it worked fine
08:02:42brixenI don't see anything blatantly wrong with the FFI signature for truncate
08:02:45dbussinkbrixen: ah ok, how do you debug it? gdb?
08:03:16brixenyeah, only way I think, but I haven't done that yet
08:04:03brixendrbrain: do you recall the issue with using ruby YAML?
08:04:17dbussinki would love to have something that can output the parameters that get passed to the c function easily, makes ffi debugging a lot more pleasurable
08:05:42evanoutput how?
08:06:57dbussinkdunno, maybe the generated code could do that if a flag is set, something like that
08:07:22dbussinkbut the problem might also be my limited gdb skills ;)
08:13:52benburkert leaves the room.
08:14:31thehcdreamer enters the room.
08:25:58ragnard enters the room.
08:29:01hassox_ enters the room.
08:34:06maduyb__ enters the room.
08:34:08wifelette leaves the room.
08:37:53wycats leaves the room.
08:39:36headius leaves the room.
08:40:04hassox__ enters the room.
08:43:49enebo leaves the room.
08:44:42hassox leaves the room.
08:48:55hassox_ leaves the room.
08:49:36hassox enters the room.
08:51:17dewd leaves the room.
08:58:41dbussinkhmm, i get compile warnings on machine.c
08:58:50hassox_ enters the room.
08:59:29evanyeah, that was my mistake.
08:59:32evani'll fix them tomorrow.
09:00:09nicksieger leaves the room.
09:03:26wycats enters the room.
09:03:28dbussinkbut is sendsite broken atm? i saw you commenting on that?
09:03:47evannot sure what adam did
09:03:54evanbut you have to do a 'rake clean build'
09:04:50dbussinkah, so only build time problems?
09:06:00maduyb__ leaves the room.
09:07:06hassox__ leaves the room.
09:09:43hassox leaves the room.
09:10:35cavalle leaves the room.
09:12:19KirinDave leaves the room.
09:12:24wycats_ enters the room.
09:12:34imajes enters the room.
09:14:54dbussinkbrixen: there
09:14:56octopod enters the room.
09:15:13dbussinkbrixen: there's definitely a ffi problem, it segfaults on bsd on the freeaddrinfo
09:19:35dbussinkbrixen: the ptr in ffi_call is 0x0
09:21:03therealadam_ enters the room.
09:25:35wycats__ enters the room.
09:26:16claudio_ enters the room.
09:33:57therealadam leaves the room.
09:34:37headius enters the room.
09:37:30wycats leaves the room.
09:43:06wycats_ leaves the room.
09:43:07claudio_ leaves the room.
09:44:54Arjen_ enters the room.
09:45:21w1rele55 enters the room.
09:46:45langenberg enters the room.
09:48:04antares enters the room.
09:53:40cavalle enters the room.
09:59:47w1rele55 leaves the room.
10:09:07imajes leaves the room.
10:10:13Arjen_ leaves the room.
10:10:21headius leaves the room.
10:11:08claudio_ enters the room.
10:11:23claudiohi
10:11:53lstoll enters the room.
10:18:31antaresclaudio_, hi there
10:20:30Arjen_ enters the room.
10:21:39headius enters the room.
10:21:39geekounet leaves the room.
10:24:31VVSizbrixen: online? (maybe not...) :)
10:25:00VVSizI'm probably not smart enough to figure out how to use the tag comment to mark tests as failed..
10:27:23VVSiztag comment --> tag command
10:27:54VVSizI'm doing something like: bin/mspec tag --add fail --fail spec/ruby/1.8/core/io/read_spec.rb
10:30:01antaresVVSiz, I think the answer lies in tag runner spec ;)
10:30:34geekounet enters the room.
10:30:36VVSizantares: :) sure, I can read the code and finally figure it out, but that's not how it shoul be :)
10:30:42antaresmspec/spec/runner/tag_spec.rb lays down parser behaviour pretty well
10:31:01antaresVVSiz, then someone should create a page on it at LH
10:34:20imajes enters the room.
10:38:02antaresVVSiz, it seems that --add 'Array#[]' --fail [PATH] is the way it should work
10:38:52VVSizlet's say I'd like to add 'fail' tag to a number of specs
10:39:11chris2 enters the room.
10:39:18VVSizit seems that I should use --add 'fail' and --fail switches
10:39:23VVSizand path at the end
10:39:45VVSizdoing that I get: Missing --action-tag or --action-string
10:39:50VVSizwhat the hell is that :)
10:41:17antares leaves the room.
10:46:23VVSizwhoa, the tags dir, where mspec stores tags, is in gitignore :)
10:53:06boyscout1 commit by Vladimir Sizikov
10:53:07boyscout * New IO.read specs.; d522af8
11:00:02ragnard leaves the room.
11:00:34geekounet leaves the room.
11:02:08Arjen_ leaves the room.
11:03:14rueVVSiz: Nice catch
11:03:43rubuildius_ppcVladimir Sizikov: d522af83d; 1573 files, 5491 examples, 18955 expectations, 0 failures, 28 errors; http://pastie.caboo.se/154712
11:04:01VVSizrue: no wonder I was confused, invoking the tag command over and over and nothnig was changed... :) (but actually, it was changed)
11:04:46rubuildiusVladimir Sizikov: d522af83d; 1573 files, 5490 examples, 18983 expectations, 0 failures, 5 errors; http://rafb.net/p/29AczD64.html
11:05:12VVSizrue: I haven't pushed .gitignore changes, since I'm not sure whether it is intentional or not, to filter out tags, and somebody might rely on that, dunno.
11:08:35geekounet enters the room.
11:12:16Arjen_ enters the room.
11:28:45antares enters the room.
11:30:17halorgium leaves the room.
11:30:21langenberg leaves the room.
11:31:47jtoy leaves the room.
11:37:36agardiner enters the room.
11:42:33antaresVVSiz, is there a way to view the list of fail tags?
11:43:14VVSizyou mean, a list of tests tagged with fail tag?
11:47:09halorgium enters the room.
11:49:24antaresVVSiz, yup
11:49:40antaresor should one cat those text files by hand?
11:49:59VVSizhmm, I don't really see a better way than just cat those files
11:50:13VVSizwe should aks brixen, it would be good to have some kind of reporting capability
11:51:04lstoll leaves the room.
11:54:24langenberg enters the room.
11:58:07kofno enters the room.
12:04:08graza enters the room.
12:09:00antaresVVSiz, I agree
12:09:13antaresVVSiz, there's a convention in class-to-spec-filename mapping anyway
12:14:44ragnard enters the room.
12:16:21ctennis leaves the room.
12:20:35ragnard_ enters the room.
12:22:18graza_ enters the room.
12:33:39chris2 leaves the room.
12:36:42graza leaves the room.
12:37:11ragnard leaves the room.
12:47:30graza enters the room.
12:47:31graza_ leaves the room.
12:47:39agardiner leaves the room.
12:48:23ragnard enters the room.
12:48:23ragnard_ leaves the room.
12:50:48langenberg leaves the room.
13:03:15skaa1 enters the room.
13:11:13imajes leaves the room.
13:16:53graza_ enters the room.
13:16:54graza leaves the room.
13:24:36w1rele55 enters the room.
13:29:42graza enters the room.
13:29:43graza_ leaves the room.
13:35:32zimbatm enters the room.
13:42:19graza_ enters the room.
13:42:20graza leaves the room.
13:47:17crafterm enters the room.
13:47:25crafterm leaves the room.
13:48:20RyanTM enters the room.
13:51:50hornbeck enters the room.
13:53:33pietia leaves the room.
13:54:37mad_phoenix enters the room.
13:56:35cremes enters the room.
13:57:17cremesDefiler: i have created an account on my osx ppc for you; ping me for details
14:02:49pietia enters the room.
14:03:37djwhittcremes: you in the US?
14:05:57cremesdjwhitt: yes, chicago, gmt -6
14:07:15djwhittcool, I thought I might be the only US person here not on PST
14:07:42djwhitt(I'm in the philadelphia area)
14:07:45AndrewO enters the room.
14:09:29mad_phoenix leaves the room.
14:19:34geekounet leaves the room.
14:19:44geekounet enters the room.
14:21:52cremesdjwhitt: nice... i am an early bird plus i start work at 7am so i'm used to being one of the "few" awake on this channel
14:23:23djwhittcremes: yeah, I'm definitely not an early bird (usually not around till 9-10 EST), but that still puts me ahead of most people on PST
14:23:51dbussinkjust live in CET, that makes you lonely during the day ;)
14:24:49cremesdbussink: no thanks! :)
14:25:15djwhittyeah, heh, could be worse, in a previous job I worked with several people in India
14:25:33dbussinkcremes: what's wrong with europe? ;)
14:26:15cremesdbussink: why nothing, of course! for me it's a nice place to visit but i wouldn't want to live there <grin>
14:27:11dbussinkcremes: and still don't want to live there? :P
14:27:32dbussinkthere is more difference between two neighboring countries here than there is in the entire us ;)
14:28:07djwhitteh, I don't know about that, I feel like I'm in a different country when I visit the south
14:28:15djwhittbut I guess we still all speak the same language
14:29:02djwhittthere's the midwest, the south, northeast and westcoast those are the US "countries"
14:29:07cremesdbussink, djwhitt: to each their own; i like it here but do enjoy visiting foreign lands; i need to collect more stamps on my passport soon
14:29:35dbussinkcremes: just have a new one, so i have to start from scratch again
14:30:01dbussinkcremes: not that i had a lot of stamps, only 1 from the usa, free travel in europe has it disadvantages
14:30:16cremesdbussink: any neat privacy invasion biometrics in it or rfid?
14:30:33cremeswe have that coming to our passports soon (if not already)
14:30:42djwhittI think we already have rfid in the new ones
14:30:50cremesugh
14:30:57dbussinkyeah, here too i think
14:31:06dbussinkit has an additional small logo on it
14:31:23probablycorey enters the room.
14:32:04djwhittdbussink: where are you located in europe?
14:32:10dbussinkdjwhitt: netherlands
14:33:49dbussinkdjwhitt: that nasty liberal country :P
14:34:13djwhittoh horror!
14:34:14djwhitthehe
14:35:01cremesdbussink: i visited finland a few years back; loved the lapinkulta beer and lonkero!
14:35:24cremesdbussink: not that you even know what those are... ;)
14:35:40jtoy enters the room.
14:35:45dbussinkcremes: nope, finnish beers are not very famous
14:36:12dbussinkcremes: belgian and german beers are better known
14:36:14cremesdbussink: i guessed as much after i hit <send>
14:36:47cremesah yes, belgians... my favorite is chimay; i pretty much like anything made by a monk!
14:37:02dbussinkcremes: hehe, i did drink grolsch and heineken when i was in the usa though
14:37:36dbussinki'm still appalled by the fact that people said that heineken had a strong tast
14:38:07dbussinktaste
14:38:08kofno_ enters the room.
14:38:16cremesdbussink: i'm sorry; there are lot's of excellent small batch beers available now; not all americans drink domestic crap made by busch/budweiser/coors/etc
14:38:37kofno leaves the room.
14:38:38ruivaldo enters the room.
14:38:50dbussinkcremes: i know
14:38:53cremesi actually live about a mile away from the main usa school for training brewmasters (siebel institute)
14:39:03dbussinkbut the concept of light beer, as in, why?
14:39:11cremesonly god knows!
14:40:16imajes enters the room.
14:42:01mosfet enters the room.
14:43:40mosfetHi,
14:44:42cremeshowdy
14:44:56mosfetIs rubinius an official project ?
14:45:00mosfetis it a commercial
14:45:14mosfetI don't understand its use
14:45:25dbussinkit's an open source project
14:46:00dbussinkthere's no direct commercial interest
14:46:04djwhittit's a project to build a better virtual machine for Ruby
14:46:22murphee enters the room.
14:46:46mosfetand in which state is it ?
14:46:50mosfetalpha beta ?
14:46:53murphee leaves the room.
14:47:04djwhittheh, possibly alpha or pre-alpha
14:47:12probablycorey leaves the room.
14:47:16djwhittcan run a lot of Ruby, but definitely not ready for production
14:47:39cremesalpha = not feature complete; beta = feature complete but buggy; rubinius is definitely alpha
14:47:56probablycorey enters the room.
14:49:05crossblaimiphone enters the room.
14:49:18crossblaimiphonehi
14:50:08djwhitthi
14:51:00ruivaldo leaves the room.
14:53:08cored enters the room.
14:56:26FFighter leaves the room.
14:57:12imajes leaves the room.
14:57:41anonuser enters the room.
14:57:58rubymaverick enters the room.
14:59:56anonuserMorn'
15:00:11dbussinkafternoon :)
15:00:18ragnard leaves the room.
15:02:39cremesdbussink: what's the verdict on osx ppc failures? ffi?
15:02:50dbussinkcremes: yeah, we blame ffi :)
15:02:58dbussinkcremes: it results in failures on other systems too
15:03:17cremesffi is used for interfacing to external C libs, right?
15:04:15binary42 enters the room.
15:11:38dbussinkcremes: to c functions in general yeah
15:11:50dbussinkbut there is still something that goes wrong there
15:12:56cremesdbussink: hmmm... is ffi the final api or is there a plan to replace it? i'm just wondering if it is worthwhile to expend effort on fixing it right now
15:13:32dbussinkwe really want the ffi interface to work ok
15:14:02dbussinkmakes it very easy to interface with c, a lot easier than for example with subtend (what mri uses to interface)
15:15:18djwhitta lot of the libraries for file io etc use ffi I think so unless you want to use primitives for all of them you need working ffi
15:15:26djwhitt(I think still figuring this stuff out)
15:16:06dbussinkwe don't want primitives for everything, only the really low level stuff
15:16:27djwhittyeah, makes sense
15:16:35dbussinkso yeah, a proper ffi is a must have
15:17:10djwhittand it's just super cool :)
15:17:45dbussinkdrbrain wrote a nice blog article on it
15:17:46dbussinkhttp://blog.segment7.net/articles/2008/01/15/rubinius-foreign-function-interface
15:24:51crossblaimiphone leaves the room.
15:25:05srbaker enters the room.
15:27:36cremesi will check out that article
15:30:51yipstar enters the room.
15:36:16agile leaves the room.
15:45:44graza_ leaves the room.
15:48:33rao45 enters the room.
15:51:36langenberg enters the room.
15:57:49antaresdbussink, hey there
15:57:58antaresdbussink, how about reviewing GC docs? :)
15:58:22antaresdbussink, or should I ask other folks?
15:58:32dbussinkantares: yeah, i'm here
15:58:54dbussinkantares: others are probably more into the details
15:58:56antaresdbussink, I think I'll put a pastie link here soon
15:59:25antaresdbussink, so if you can review it would cool
16:04:09be9 enters the room.
16:06:40dbussinkantares: just pastie it up and probably poke some other people too
16:10:03KirinDave enters the room.
16:11:17jayWHY enters the room.
16:12:27w1rele55 leaves the room.
16:13:46langenberghi guys
16:18:09dbussinklangenberg: hi
16:18:14dgtized leaves the room.
16:18:39langenbergYou won't believe this, I just had a lecture about Agile development at school.
16:19:10langenbergFirst you learn the waterfall model, next thing you know they are actually talking about agile methods.
16:20:38tbmcmullen leaves the room.
16:20:45dbussinkhehe
16:22:39dbussinkbut most useful things i know about software development processes comes from experience, not theory
16:24:46wmoxam enters the room.
16:28:34jtoy leaves the room.
16:29:48macournoyer enters the room.
16:31:17benburkert enters the room.
16:36:11dgtized enters the room.
16:36:40benburkert leaves the room.
16:36:45dodecaphonic enters the room.
16:37:20langenbergIs Rubinius broken on Intel Leopard?
16:37:27langenbergI'm unable to build.
16:38:16langenbergOh, can't run shotgun/rubinius.
16:38:18enebo enters the room.
16:40:38dbussinklangenberg: you did a distclean?
16:41:11langenbergrake distclean && rake build?
16:41:43dbussinklangenberg: yeah, you and also do rake build distclean :)
16:41:53dbussinklangenberg: the other way around of course
16:42:01FFighter enters the room.
16:46:39langenbergdbussink: after distclean: shotgun/rubinius:98:in `exec': No such file or directory - rubinius (Errno::ENOENT)
16:49:46agile enters the room.
16:50:35srbaker leaves the room.
16:51:37w1rele55 enters the room.
16:53:44brixenVVSiz: hey, sorry about that, just pushed mspec-tag last night. docs coming
16:54:26VVSizbrixen: thanks!
16:54:54VVSizit seems that currently the tag command requires one of the action arguments, which i'm not sure why
16:56:13brixenVVSiz: the use case is, you want to run several specs but tag only a set, or invoke the debugger on only a certain spec
16:56:23brixenVVSiz: so I made all actions require a specific filter to match
16:56:23srbaker enters the room.
16:56:43brixenVVSiz: whereas, the others -e/-E -p/-P -g/-G are for filter which specs run
16:56:54brixenso there's two things: which specs are run, which specs trigger actions
16:57:02VVSizheh, that's different from what I typically do :)
16:57:02brixencombining them was too limiting
16:57:17srbaker leaves the room.
16:57:21VVSizlet's say I modified two spec files, then i need to exclude only those specs from those files
16:57:40VVSizI don't really care how they are named. All I know that they are in those two files
16:58:12brixenright, and actually the actions match by default
16:58:27brixenI'll fix the runner, it's kicking you out on tag when it shouldn't
16:59:16w1rele55 leaves the room.
16:59:22VVSizI'd like to be able to do something like this:
16:59:23VVSizbin/mspec tag --add 'fails' --fail spec/ruby/1.8/core/io/read_spec.rb
16:59:34brixenbtw, to run anything with a tag: bin/mspec -g fails for instance
16:59:45brixenVVSiz: yeah, that will work, I'll fix it
16:59:50VVSizcool!
17:00:44VVSizas for -g, I'm not sure it really works
17:01:04srbaker enters the room.
17:01:10brixenit looks for tags and runs spec with descriptions matching those tags
17:01:16VVSizbin/mspec -g fails spec/ruby/1.8/core/fixnum/ --> 33 files, 72 examples, 247 expectations, 0 failures, 6 errors
17:01:49VVSizwhat I'd like to be able to is to execute *ONLY* failing specs (those that are marked with fails tag), not everything else
17:01:58brixenyeah, I'll check that
17:02:50VVSizalso, someone on the channel asked how to print out list of all specs with particular tag
17:02:54brixencurrent conditional is: if (include is empty or spec is in include) or (exclude is not empty and spec is not in exclude)
17:03:11VVSizI'd also like to know if it's possible (without grepping the tag files manually)
17:03:35brixenwell, you should get that by running the tag, but you want to know without running the tag?
17:04:00VVSizyeah, like all specs tagged with fails
17:04:14brixensure, ok
17:04:56VVSizrunning them is not optimal, since the output is filled with errors and exceptions (not to mention than some exclusions might actually break run)
17:05:07brixenyep
17:05:16brixenrue is going to add some reporting/verifying stuff
17:05:29VVSizgreat stuff!
17:05:30brixenbut we could add a general --pretend flag that does everything but execute
17:06:49brixenlangenberg: is your build still broken? I did rake clean build on os x intel and it's running ci
17:07:18langenbergI did a clean checkout, it's working again.
17:08:10brixencool
17:10:09boyscout5 commits by Brian Ford
17:10:10boyscout * Symbols as Fixnums is long deprecated. We don't spec it.; cfd51af
17:10:11boyscout * Exclude spec/tags dir from .gitignore.; 3180d25
17:10:12boyscout * RbYAML is not in Ruby standard lib. Move specs for it to spec/library/rbyaml.; a8bd2a1
17:10:13boyscout * Restrict specs in spec/ruby/1.8 to current stable 1.8 version.; 56b454a
17:10:14boyscout * Fix Dir#pos=/#seek specs. We shouldn't spec undefined platform behavior.; 22e01d1
17:12:04jayWHY leaves the room.
17:14:56antares leaves the room.
17:16:10brixenVVSiz: ok, -e/-E and -p/-P are working as expected, and -G fails seems ok. I'll figure out -g
17:16:42pietia leaves the room.
17:17:06VVSizcool
17:17:35VVSizheh, that .gitignore with 'tags' in it was giving me a hard time today... :)
17:17:50brixenheh, I bet I know what it is, there's no tag, so it never adds an entry to the include list, so the list is empty, so the specs aren't filtered
17:17:53brixenhmm
17:18:08brixenVVSiz: doh, yeah me too when I pushed all the mspec changes, it didn't push a bunch of tags
17:18:19brixenand I looked in that damn file multiple times!
17:18:32brixenI was looking for stuff like .txt
17:18:34brixenheh
17:18:41VVSiz:)
17:19:01Defilercremes: yo
17:19:35brixenVVSiz: yep: it "does not create a MatchFilter if no tags were found" do
17:19:52cremesDefiler: give me your email addr in exchange for login details to my osx ppc box
17:20:46Defilercremes: wilson@supremetyrant.com
17:20:48brixenVVSiz: I suppose -g must be made to explicitly mean 'only if'
17:22:52cremesDefiler: check your email... ping me if you have a problem getting in
17:22:58rubuildius_ppcBrian Ford: cfd51af48; 1572 files, 5474 examples, 18929 expectations, 0 failures, 28 errors; http://pastie.caboo.se/154869
17:24:54rubuildiusBrian Ford: cfd51af48; 1572 files, 5473 examples, 18957 expectations, 0 failures, 5 errors; http://rafb.net/p/K69WGA63.html
17:26:49Defilercremes: Excellent.. thanks very much
17:27:02cremesDefiler: you are welcome
17:27:46dbussinkDefiler: going ffi bughunting? :)
17:28:00dbussinkDefiler: there are segfault issues with ffi on freebsd too
17:28:02DefilerGetting set up to do so, though I may not get to it today
17:28:13DefilerFirst I have to learn about these fancy sockets
17:28:15Defiler:)
17:28:54tongueroo enters the room.
17:29:05tongueroo leaves the room.
17:35:34thehcdreamer leaves the room.
17:37:47benburkert enters the room.
17:38:19DefilerIs it just me, or do AF_INET and AF_UNIX sockets have little in common?
17:38:29benburkert_ enters the room.
17:38:38DefilerI'm not sure having a file descriptor is enough to put those two things in the same class hierarchy
17:39:41lopex enters the room.
17:42:41ragnard enters the room.
17:43:11KirinDave leaves the room.
17:44:00technomancy enters the room.
17:48:42jtoy enters the room.
17:48:43srbaker leaves the room.
17:49:02wmoxam_ enters the room.
17:50:01wmoxam leaves the room.
17:50:44pietia enters the room.
17:52:48wmoxam_ leaves the room.
17:52:55rueDefiler: Yeah well
17:53:22rueUNIX ain't perfect :P
17:53:42rueI was complaining about the ridiculous File < IO scheme in Ruby earlier
17:54:01wmoxam enters the room.
17:56:28enebo leaves the room.
17:56:54Defilerrue: Yeah, it just seems inappropriate
17:57:14Defilerrue: Any design that forces you to think twice before calling 'super' is busted, IMO
17:58:21boyscout1 commit by Brian Ford
17:58:22boyscout * Fix TagFilter to only include specs matching tag.; 1262b7f
17:58:40headius leaves the room.
17:58:54brixendown to only 4 fails specs on mri for spec/ruby
17:58:56benburkert leaves the room.
17:59:41antares enters the room.
18:01:41brixenDefiler: are you getting an illegal instruction running specs on 64bit? I was getting it on the ey slice
18:02:42djwhittI am
18:02:52djwhittusing Gentoo on amd64
18:03:02Defilerbrixen: Yes
18:03:04brixenVVSiz: do you get a failure in mri for the io/reopen_spec
18:03:17brixendjwhitt / Defiler: ok, so it's not just me :)
18:03:24Defilerbrixen: ffi gives me SIGILL on my EY slice
18:03:25hooopy enters the room.
18:03:39brixenDefiler: ok, in a socket spec, right?
18:03:56maduyb__ enters the room.
18:04:21hooopyAre you guys trying to target 1.8.6 and not a superset of its functionality?
18:04:39DefilerNo. :)
18:04:44hooopycool
18:05:02hooopyBecause I just notice rubinius supports zero width negative look behind assertions
18:05:03Defilerdrbrain: $stdout question for you, if you have a sec
18:05:24Defilerhooopy: We are using the regex library that 1.9 uses, so we have (roughly) the same feature-set
18:05:44hooopyoh... will you be re-implementing that in Ruby?
18:05:54DefilerNope
18:05:56DefilerIt is nice
18:06:17hooopybut, but i thought you were going to make this self hosting
18:06:22DefilerIt is already
18:06:41hooopywell i guess i took that to mean that as much of the libraries as possible would be written in ruby
18:06:41Defilerself-hosting doesn't mean you have to write an operating system or anything
18:07:13DefilerIf someone can come up with a good reason to rewrite the regex library in Ruby, then maybe somebody will
18:07:39Defileroniguruma has had a lot of work done to it, though, and we would probably spend the first 8 months fixing bugs he has already fixed
18:07:45rubuildius_ppcBrian Ford: 1262b7fdb; 1572 files, 5474 examples, 18929 expectations, 0 failures, 28 errors; http://pastie.caboo.se/154900
18:09:41rubuildiusBrian Ford: 1262b7fdb; 1572 files, 5473 examples, 18957 expectations, 0 failures, 5 errors; http://rafb.net/p/it6zKg10.html
18:10:16VVSizbrixen: about reopen, one sec
18:10:46dewd enters the room.
18:11:26VVSizbrixen: nope, reopen is clean
18:11:30Defilerhooopy: On the other hand, though..
18:11:46Defilerhooopy: We do at least call the Ruby methods for the various regexp match, callback, etc.
18:11:48gensym enters the room.
18:12:02Defilerhooopy: So if you change String#=~, String#gsub will use it
18:12:46brixenVVSiz: what platform?
18:13:02VVSizUbuntu Linux 7.10 32bit
18:13:05KirinDave enters the room.
18:13:07TheVoice enters the room.
18:13:09hooopyam I right in thinking using oniguruma would be an exception to a rule?
18:13:22hooopya rule stating that "do as much as you can in Ruby, then C"?
18:13:25brixenVVSiz: ok, this is what I get on os x: http://pastie.org/154905
18:13:27pauldix enters the room.
18:13:27KirinDav enters the room.
18:14:04djwhittso why does the regexp benchmark give such poor results?
18:14:35brixendjwhitt: that's a good question, there is a lot of ruby code to interact with the primitives
18:14:46brixena good place for someone to investigate :_
18:14:49brixener :)
18:15:12VVSizbrixen: weird, what platform? MacOS?
18:15:17Defilerhooopy: We don't really have rules. heh
18:15:21djwhitthmm... not a bad idea
18:15:34brixenVVSiz: yeah 10.5.2
18:16:19GMFlash leaves the room.
18:16:20Defilerhooopy: If there is a rule, it is this.. Write it in Ruby unless it would take longer to write it in Ruby
18:16:27GMFlash enters the room.
18:16:32hooopypoints to MRI
18:16:34rueHa, looks like Ola will be in SF next week
18:16:39rueNot that I will be there or anything
18:16:49DefilerWhat about MRI?
18:17:21VVSizbrixen: also, seems to be working on Win32
18:17:58brixenVVSiz: yeah, weird, works on 64bit gentoo, also
18:18:17brixenVVSiz: but I have 2 failures for sysread and sysseek
18:18:33Defilerdjwhitt: We are creating Ruby objects up the wing wong in the regexp benchmark
18:18:33brixenspec'ing this stuff that is so platform dependent is really a pain
18:18:55Defilerdjwhitt: Because relatively little of the work happens in the regex library, mostly in the wacky Ruby wrappers around it
18:19:00VVSizbrixen: finally!! somebody else also having those 2 annoying failures for sysreead and sysseek! :)
18:19:16brixenVVSiz: heh, so they fail on 7.10?
18:19:24VVSizpastie: for brixen
18:19:33headius enters the room.
18:19:36evanmorning.
18:19:38evanbrixen: got your email
18:19:42brixenmorning
18:19:43brixenok
18:19:44djwhittDefiler: ok, I'm actually running it with profiling right now just to see for myself
18:19:45evanwe'll have the call later this week
18:19:55brixenevan: sure, fri works
18:19:56pastiebrixen: http://pastie.org/154907 by VVSiz.
18:20:02evanbrixen: ok
18:20:16brixenVVSiz: exactly those two, 7.10, right?
18:20:17enebo enters the room.
18:20:23VVSizbrixen: yep, failures are on Ubuntu 7.10. yes.
18:20:24Defilerdjwhitt: Cool
18:20:35be9 leaves the room.
18:20:46brixenVVSiz: hmm, how can we do this without ping-ponging back and forth on these platforms?
18:21:12hooopyVMWare
18:21:37tarcierisomething like FreeBSD tinderbox?
18:21:41tarcieria build farm that does CI
18:21:47VVSizI use virtualbox to run Vista, XP, Ubuntu, but can't run MacOS, damn! :)
18:22:25VVSizfor JRuby excludes, I sort of maintain them on Linux platform, and headius handles macos excludes
18:22:32dbussinkbrixen: you saw my messages about segfaults on freebsd
18:22:43brixendbussink: yeah, I think so
18:22:51brixendbussink: in socket specs?
18:23:09wifelette enters the room.
18:23:20VVSizas for ping-ponging... yeah, I think it's inevitable, since the docs are not very detailed, unfortunately
18:23:33dbussinkbrixen: yeah, socket was problematic, for ffi functions like freeaddrinfo
18:23:40brixenVVSiz: yeah, seems almost nothing docs all these differences
18:24:04brixenevan: would an ey slice with vmware instances for multiple flavors of linux be possible?
18:24:41VVSizfurthermore, MRI behavior CHANGES from version to version too. for example, MRI 1.8.5 used to pass rubicon's io tests, and MRI 1.8.6 fails reopen test
18:25:07brixenVVSiz: yep, it's really frustrating
18:25:12dbussinkbrixen: i know that xen is pretty easy to set up
18:25:27d2dchat leaves the room.
18:25:39brixendbussink: but doesn't xen require a modified kernel?
18:26:02evanbrixen: no
18:26:02dbussinkbrixen: yeah, but it's usually available for every distribution
18:26:14evanbrixen: likely not
18:26:16VVSizbrixen: headius made a major IO rewrite in JRuby to match MRI sources as close as possible, since it's practically impossible to have MRI behavior without having similar sources :)
18:26:22dbussinkbrixen: unno what you can or can't do with an ey slice
18:26:54brixenevan: dedicated machine?
18:27:02evanwhat we should probably do is talk with EY about investing in a dedicated machine
18:27:05evan:D
18:27:08brixenk
18:27:22nicksieger enters the room.
18:27:23evanperhaps 4 core, 8G of memory
18:27:31evanwe could run a number of VMs on that
18:27:33brixenVVSiz: yeah, I finally bit the bullet and ported jruby fnmatch ;)
18:27:42brixenevan: that would be really cool
18:28:25brixenVVSiz: the other issue is this, perhaps we should aim for a baseline behavior and below that there are no specs, so you're in undefined territory
18:28:42brixenVVSiz: for example, if sysread from buffered IO will fail on some platforms, no one should use it ever
18:29:23brixenbecause your code is inherently non portable in a way that is not really remedied by using a different library, for instance
18:30:03Defilerevan: EY can buy my used one and I will get a new one :)
18:32:04evanDefiler: heh
18:32:26headiusgod morning
18:32:52headiussysread from buffered IO should fail everywhere, as far as I can tell
18:33:02headiusthey have a hard check in the C code for it
18:33:17brixenheadius: seems to pass on ubuntu 7.10 :)
18:33:18brixenfun
18:33:34headiusa sysread after a read?
18:33:37VVSizbrixen: well, it's not strictly sysread after buffered read
18:33:42VVSizit's a sysread after reopen
18:33:47sudoer enters the room.
18:33:49brixenright
18:33:59VVSiza reopen of a file that also was with sysread
18:34:00brixenbut the failure I get is sysread from buffered IO
18:34:26VVSizso it seems that Ubuntu is right there. the question why sysread after reopen fails on MacOS with that weird error message :)
18:35:02VVSizbut I agree that some highly-platform-dependent behaviors, not cleary specified (if at all) should not be in rubyspecs
18:35:36VVSizI tried to write specs for backtraces, but finally moved them to jruby proper, since there are no hard specs for that, and every impl. doing thing a little differently :)
18:35:51headiusthat could be a bug...reopen should reset/flush buffer
18:36:12headiusof course there's so many paths through the IO methods who knows
18:36:49ruetarcieri: Hey--I do not recall if I asked about it earlier but had you done any work or planning on Rev/Rubinius interaction?
18:37:12brixenVVSiz: oh, did you have a question about bignum_value helper?
18:37:46VVSizbrixen: yeah, that one. I wasn't sure why it should be really in helper. It took me some time even to figure out where it's coming from :)
18:37:57VVSizI mean, in global helper
18:38:08brixenwell, it's because bignums are used in other specs
18:38:18hornbeck leaves the room.
18:38:24brixenand since the range of bignums is platform/impl dependent, this is more general
18:38:50brixenI need to write one for epsilon_value too, so the be_close matchers are more correct
18:39:35brixenVVSiz: you notice all use of 0xffffffff in bignum specs and elsewhere, they're not even spec'ing bignum interactions on jruby for instance
18:39:54d2dchat enters the room.
18:39:57ruetarcieri: I have a vague, possibly fabricated recollection that you were of the