Camelia, the Perl 6 bug

IRC log for #parrot, 2008-06-18

Parrot | source cross referenced

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:00 * tetragon is trying to get an Intel OS X 10.5 user to say it works
00:00 * tetragon has PPC OS X 10.5
00:01 japhb Can any Mac OS X people take a gander at RT #55950?  Packy appears to be getting a weird error trying to link with his libz, but he says that it goes away if he disables OpenGL in Configure, which mystifies me.
00:01 tetragon Just as soon as I prepare today's tree
00:01 dalek r28494 | Whiteknight++ | trunk:
00:01 dalek : [src/gc] Add some basic function-level documentation to src/gc/smallobject.c as per RT#48260. More is needed, but I've covered the basics.
00:01 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28494
00:01 japhb tetragon: prepare today's tree?
00:02 tetragon japhb: I keep one tree always clean, and every day (that I do a build) create a copy that gets built on
00:03 japhb tetragon: why?  Trying to avoid having to do a fresh checkout ever?
00:03 tetragon Fresh checkouts are slow on my box
00:03 japhb Ah.
00:03 apeiron tetragon, My box with the Macports perl works fine.
00:03 apeiron (5.10)
00:04 tetragon One of the networks I spend time on took a couple hours to do typical checkouts
00:06 japhb ewww
00:09 tetragon That's the network my Intel box is on
00:09 tetragon I grew used to carrying trees and patches across on my laptop
00:09 TiMBuS joined #parrot
00:10 AndyA joined #parrot
00:12 tetragon Looks 55950's reporter is having some issues with darwinports
00:13 japhb tetragon: Ah.  Wasn't this something that kid51 was just working on?
00:13 japhb (As in, over the last couple days?)
00:17 tetragon I'm thinking more in terms of the library side
00:18 * DietCoke hides out in his room, briefly.
00:19 tetragon That's a plain old linker error
00:20 japhb So what does that have to do with OpenGL?  Or do you think he ate a red herring?
00:20 tetragon He may also have an OpenGL install through darwinports
00:21 DietCoke the error I saw this weekend was gl related.
00:21 tetragon But that ticket doesn't have the real details that I would need
00:21 japhb Ah.  Hmmm.
00:21 japhb DietCoke: hmmm what?
00:21 DietCoke packy's error.
00:21 japhb DietCoke: Packy is the one we're talking about.
00:21 DietCoke The last time I saw it (not the one in the ticket, on his console), it looked opengl related.
00:21 DietCoke japhb: Yes. That's why I chimed in.
00:21 japhb ah, IRConfusion
00:22 tetragon The one in the ticket looks like he managed to screw up his darwinports tree
00:24 japhb Can darwinports supports be configured away?  Like '--without-darwinports' or somesuch?
00:25 tetragon japhb: sudo rm -rf /opt/local
00:25 japhb LOL
00:25 DietCoke that's helpful! =-)
00:25 japhb I see.
00:25 DietCoke you could, say, have him eschew adding those library directories. (by explicitly specifying the lib and include paths to configure)
00:26 DietCoke I am wondering why you say it looks like he screwed up his darwinports, though.
00:27 DietCoke you think a permissions issue on the libz1.dylib?
00:27 tetragon Or it doesn't exist there
00:27 DietCoke ok. *must* it exist there?
00:27 japhb DietCoke: wouldn't the darwinports dirs just be added back in by the _handle_darwin_for_macports hacks?
00:27 tetragon That is where darwinports install it
00:27 DietCoke yes, but does it -always- install it?
00:27 DietCoke or just based on certain packages?
00:27 contingencyplan joined #parrot
00:28 japhb *lightbulb*
00:28 tetragon How do you use darwinports again?
00:28 tetragon I have a minor amount of stuff from it left on my box, and that includes libz
00:28 DietCoke "port"
00:29 japhb I bet I know why he thinks it's OpenGL related -- because config::auto::opengl is one of the four steps that calls _handle_darwin_for_* .  Thus, with OpenGL disabled, he might not trip over that problem.
00:29 tetragon My box can't find the manpage
00:30 DietCoke There. There's an excellent reason for us to rip those methods out.
00:30 japhb You'll get no argument from me.
00:31 DietCoke I added a note about his config options. Those were the only options that worked on osx ppc for some time.
00:31 DietCoke (modulo optimize)
00:31 tetragon Which options?
00:31 purl Which options are you talking about?
00:31 DietCoke tetragon: 'port'
00:31 japhb Is there any sane reason for a Mac OS X user to *not* use the Apple-supplied OpenGL and GLUT frameworks if they want OpenGL?
00:31 tetragon I normally don't pass Configure.pl any options
00:31 DietCoke I would not claim to support such a config.
00:31 DietCoke tetragon: do you have osx ppc?
00:32 tetragon Yes
00:32 DietCoke what os rev?
00:32 tetragon 10.5.2
00:32 DietCoke that may be why.
00:32 tetragon It's what most of my builds have been on
00:32 DietCoke I've been with parrot since 2001 or so, and those options were the only way I ever found to build on older boxen.
00:32 tetragon And what are those options?
00:33 DietCoke ... they're in the ticket. =-)
00:33 DietCoke CC="gcc-4.0" CX="g++-4.0" perl Configure.pl --
00:33 DietCoke > > cc="$CC" --cxx="$CX" --link="$CX" --ld="$CX"
00:33 DietCoke I was adding 'ccache' in there as well at one point.
00:33 DietCoke could probably be resolved by a clever edit of the hints file.
00:34 DietCoke ISTR the link/ld were the important ones.
00:34 tetragon My 10.4 box doesn't need that
00:34 DietCoke again, what version of 10.4? is it post-introduction of intel?
00:34 DietCoke could also be tied to the version of XCode.
00:35 tetragon Have you set the environment variable to set the OS X version to build for?
00:35 DietCoke he submitted a patch for that, yes.
00:37 DietCoke I should note, I don't build on osx ppc much at all anymore. I primarily use strawberry perl, linux, or osx/86
00:37 DietCoke But osx/ppc (10.3, 10.4) used to be my primary dev env.
00:37 tetragon And he hasn't played with gcc-select (IIRC) to change the system to defaulting to gcc 4 instead of gcc 3?
00:38 magnachef_ joined #parrot
00:38 DietCoke perhaps he has. that was irrelevant to the issue of needing to specify the link/ld flags.
00:38 tetragon (My 10.4 box is not where I am)
00:38 DietCoke and I am not where packy is.
00:39 japhb OK, lemme see if I understand the concept of the _handle_darwin_for_* calls -- each time we try to autodetect a library that could be system or could be a fink or darwinports install, we check to see if the header file for that library exists in the fink/ports locations.  If so, we add compile/link flags for fink/darwinports.  Correct?
00:39 tetragon My box didn't have any difficulties building
00:39 tetragon Just the usual stack of deprecation warnings
00:39 DietCoke tetragon: I understand this. Please believe me when I tell you ours did.
00:40 DietCoke japhb: we seem to just add library path and include path info.
00:40 DietCoke rather than adding it once when we do the intial detection. instead, we are there saving it off into separate config vars.
00:41 japhb So this sounds like our handling of fink/darwinports tries to be too magical.  Shouldn't we just be giving the user a Configure.pl option or two that allows them to decide to add those lib/include paths or not?
00:41 DietCoke my recommendation on this was to just, if we find those directories, add them to the normal -L and -I listings. then eschew any further work down the road.
00:41 japhb DietCoke: nod
00:42 DietCoke I think adding those paths if on OS X is not entirely evil.
00:42 dalek r28495 | Whiteknight++ | trunk:
00:42 dalek : [src/gc] Add some basic function-level documentation to src/gc/resources.c as per RT#48260. More is needed, but I've covered the basics.
00:42 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28495
00:42 DietCoke But I would have little problem with forcing users to specify non-standard paths. it's just that these are semi-standard.
00:42 japhb But there should be a separate config::auto step for detecting them, that the user can override, rather than each random library step running the detection on its own in a non-overrideable fashion
00:43 japhb bah, have to go AFK.  Will backlog.
00:43 DietCoke japhb: yes yes yes
00:44 DietCoke tetragon: replied to packy's ticket with a request to retest sans the extra options now that we have the DEPLOY var set by default.
00:44 DietCoke ... trying to, but rt sux.
00:45 tetragon DietCoke: I'm trying a build with libz.1.dylib removed from the remains of darwinports
00:49 tetragon Hrm... the error in the ticket wasn't even during the build of OpenGL-related code
00:54 tetragon And my box didn't have darwinports in the include list for things like subproxy.bundle even though Configure.pl found it
00:56 DietCoke tetragon: no, but the supposition is that not disabling opengl support causes that dir to be checked and added.
00:56 DietCoke ... which then borks something later down the line.
00:56 DietCoke Iunno.
00:56 tetragon And I didn't disable OpenGL support
00:59 tetragon My parrot build didn't link againsts anything in darwinports
01:00 tetragon With the exception of my ICU build, everything comes with OS X 10.5
01:02 tetragon (I don't have readline)
01:12 japhb tetragon: My reading is that he had *something* GL related in his darwinports.  config::auto::opengl noticed it, and added darwinports to the paths.  But the code that does that adds the top level of darwinports, and it gets added as default flags for compile and link of all of parrot.  So if there is some non-OpenGL-related breakage in his darwinports, it would show up in a different compile step.  So enabling OpenGL was just a trigger, b
01:12 japhb ecause it enabled linking to darwinports, which blew his compile to heck.
01:13 tetragon I'm part of the way through testing that theory (using GMP instead of OpenGL)
01:13 japhb nodnod
01:13 japhb cool
01:14 tetragon I have a GMP build that parrot cannot currently use as the configure step doesn't understand frameworks
01:15 tetragon For the opengl step to trigger, all that would be required is the presence of /opt/local/include/GL/glut.h
01:15 japhb sounds right.
01:16 tetragon I'm just using GMP instead of OpenGL for this test
01:16 japhb nod
01:22 tetragon Well, I'm still not getting the error from ld (probably because nothing that parrot would touch here makes use of (the excised) libz from darwinports)
01:23 tetragon But it does result in /opt/local/foo being included in the c++ lines
01:24 tetragon Hrm...
01:25 tetragon My understanding of the lines in config::auto::opengl is that it will /never/ use any fink or darwinports OpenGL install, always the system frameworks
01:26 tetragon But it will still add the fink or darwinport file locations if it finds GL/glut.h
01:30 DietCoke ... that doesn't sound right.
01:30 tetragon Which results in parrot using other libraries from darwinports
01:30 tetragon Ones that don't have _handle_foo lines in their steps
01:32 tetragon OS X doesn't supply OpenSSL 0.9.8, but it's being detected through the handle_foo of GMP
01:44 dalek r28496 | coke++ | trunk:
01:44 dalek : remove old XXX comment; doesn't hurt to have the coda here.
01:44 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28496
01:50 dalek r28497 | coke++ | trunk:
01:50 dalek : [codingstd]
01:50 dalek : Fix some codingstd tests and a typo.
01:50 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28497
01:54 dalek r28498 | Whiteknight++ | gsoc_pdd09:
01:54 dalek : [gsoc_pdd09] Creating a basic GCAccess DYNPMC that we can use to access the GC internals from high-level code. This is probably just temporary for testing stuff.
01:54 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28498
01:56 tetragon DietCoke: What doesn't sound right, what the step does on OS X?  Unless you explicitly disable it (or possibly on the server variant of OS X, haven't checked there) it will always pass using the system framework GL.
01:58 tetragon Sticking a copy of GMP into darwinports gets Configure.pl to detect gettext, something that is only installed via darwinports on my box
02:04 tetragon Removing that set of GMP, but creating an empty GL/glut.h causes parrot to still find darwinports installs of libraries
02:04 tetragon s/parrot/configure.pl/
02:10 DietCoke tetragon: I'd recommend working more directly with packy to try to resolve the issue.
02:10 jjuran joined #parrot
02:11 DietCoke I am simplifying the FAQ.
02:11 DietCoke anyone care to proffer a one paragraph summary of why we're using register over stack?
02:12 Whiteknight I could take a stab at it, what's it for?
02:12 dalek r28499 | Whiteknight++ | gsoc_pdd09:
02:12 dalek : [gsoc_pdd09] A few small fixes, updates, and a new method for the GCAccess PMC.
02:12 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28499
02:12 DietCoke "the faq"
02:12 DietCoke docs/faq.pod
02:12 DietCoke There's way too much old cruft in it, trying to cleanse it.
02:12 Whiteknight let me throw down a basic draft now, we can improve on my words later if needed.
02:13 DietCoke danke. I may simply remove the section entirely and then you can proffer a suggestion and add it back in. =-)
02:14 Whiteknight i could make a commit on it within 5 minutes if you want, or I could ship you a patch or something
02:16 Whiteknight I'm basically just revising and improving the stuff that is already there
02:20 Whiteknight actually, a lot of the stuff there was pretty good already, I think. Room for some improvements, but the gist is nice
02:43 japhb tetragon: OK, sounds like our hunches were correct.  And that the whole way of handling fink and darwinports is fu.
02:44 japhb The problem is that each config step thinks locally and acts globally.
02:45 japhb Simply doing a global detection and application of fink or darwinports would be wrong too.
02:47 japhb A user needs to specify their preferred order for looking for libs and includes, and we need to respect that.  We can still do per-step detection of whether fink, darwinports, or system actually has support for a given package or not, but that only should act to knock that option out of the running if it doesn't support that package.  And the compile/link options need to DTRT per lib.
02:47 japhb Bleah.
02:48 japhb Configure and build are hard.  Let's go shopping.
02:50 Whiteknight (go shopping)++
02:51 tetragon Bah, shopping is hard, let's do math
02:51 Whiteknight math++
02:51 japhb heh
02:51 * Whiteknight is easy to please
02:51 * tetragon vastly prefers math to shopping
02:52 japhb I'd have to agree with tetragon, actually.  Math is much easier than shopping.
02:54 Whiteknight Depends on the math, i've never been a fan of partial diff-eq
02:55 japhb OK, here's another wrench in the works.  Compiling individual object files against different sets of headers than the link uses for libs would be bad.  So how do we respect the user's wishes per-package, when the whole thing gets linked into the parrot shared lib ...?
02:55 japhb grrr, unintentional resend
02:56 cotto_yapc joined #parrot
02:56 japhb Whiteknight: that depends on what the shopping equivalent is.
02:57 japhb Perhaps it would be finding something that doesn't look absolutely aweful on you at an expensive NY boutique?
02:57 japhb In which case, I submit that partial diffeq are still easier.
02:57 japhb er awful
02:58 Whiteknight touche
03:01 tetragon aweful, awful, what's the difference once you hit high fashion?
03:04 DietCoke Whiteknight: committed.
03:04 purl The chicken is involved, but the pig is *committed*.
03:04 DietCoke feel free to add back in a small note. I may edit it mercilessly, though. =-)
03:05 dalek r28500 | coke++ | trunk:
03:05 dalek : [docs]
03:05 dalek : Update the FAQ, eliminating a lot of old information, and update the tone
03:05 dalek : a bit.
03:05 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28500
03:07 Whiteknight DietCoke, okay, I'll add in some stuff. You want me to keep it small?
03:07 DietCoke yes please.
03:07 DietCoke faq is not the place for unabashed rambling. =-)
03:08 Whiteknight but rambling is gooooooooood.
03:15 tetragon left #parrot
03:15 DietCoke no no.
03:15 DietCoke if we need a rambling answer, it can go in docs/dev
03:15 tetragon joined #parrot
03:16 Whiteknight I'm looking at the diff, you were ruthless!
03:16 * tetragon notes that the Apple menu isn't the button that displays an 'X' when hovered over
03:16 dalek r28501 | coke++ | tcl_pct:
03:16 dalek : svn merge -r28489:28500 from trunk to the tcl_pct branch
03:16 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28501
03:18 japhb So noted, tetragon.  :-)
03:19 japhb Whiteknight: wow, you're not kidding
03:20 Whiteknight DietCoke, committed. Read and edit away
03:20 Whiteknight I'm also going to try to update docs/memory_internals.pod now
03:21 dalek r28502 | Whiteknight++ | trunk:
03:21 dalek : [docs] update FAQ to add a short, short, short blurb about why register-based is better then stack-based.
03:21 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28502
03:31 rblackwe joined #parrot
03:37 DietCoke Whiteknight: good, except I'd remove the "optimization line.
03:37 DietCoke I remain unconvinced that we can use the same optimization tricks with infinite registers.
03:37 DietCoke thanks for the writeup.
03:39 dalek r28503 | japhb++ | trunk:
03:39 dalek : [DOCS] Fix grammo in FAQ
03:39 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28503
03:39 japhb DietCoke: I think it's rather that the optimizations can be considered in their pure CompSci forms rather than having to face limits
03:52 spinclad for any number of registers 0..^n, they are bounded by n...
03:53 DietCoke (modulo the bug where we can only really have 99 registers. =-)
03:55 ank joined #parrot
03:55 japhb spinclad: I was thinking more that for us, register allocation becomes a pure question of optimal caching algorithms for a fully associative cache of unknown size, rather than algorithms for handling register files of fixed size
03:56 Whiteknight DietCoke, that's fine you can edit it as much as you want. I was just trying to paraphrase the stuff that was in the longer-version
03:58 DietCoke Whiteknight++
04:03 dalek r28504 | japhb++ | trunk:
04:03 dalek : [OpenGL] Use iter opcode instead of new 'Iterator'
04:03 dalek : * OpenGL.pir: change instances of "new 'Iterator', foo" to "iter foo"
04:03 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28504
04:12 tetragon joined #parrot
04:15 dalek r28505 | coke++ | tcl_pct:
04:15 dalek : [tcl]
04:15 dalek : First cut from the language shell creator. No tcl specific information yet,
04:15 dalek : but we have a shell.
04:15 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28505
04:23 dalek r28506 | cotto++ | trunk:
04:23 dalek : Updated kjs' tutorial #1 to add pointer to the documentation for commandline_*, addressing #54694.
04:23 dalek : Also did some minor rewording for clarity.
04:23 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28506
05:04 rblackwe joined #parrot
06:10 DietCoke joined #parrot
06:10 * DietCoke yawns.
06:10 * DietCoke seems to have been disconnected.
06:10 DietCoke piper?
06:10 purl i heard piper was the logbot for #parrot
06:10 DietCoke karma piper?
06:10 purl piper has neutral karma
06:10 DietCoke piper--
06:17 bacek_ joined #parrot
06:17 bacek_ hi everyone
06:17 purl Howdy, bacek_, you fantastic person you.
06:17 uniejo joined #parrot
06:18 bacek_ purl: you are still example of stupid AI
06:18 purl ...but purl is a buggy piece of shit or a flooterbuck or a douchebag or an autonomic handjob system, well beyond the need for human intervention or a retard or a trannybot or a fuckslut in GumbyBRAIN's mind or going alone. or almost an anagram of Donaudampfschiffahrtskapitaensmuetzenkordel or a perv or an auto-triage bot or a she or a smartass or the sixth beatle. or http://github.com\/dagbrown\/flooterbuck\/ or nvb...
06:22 cotto_yapc (ignoring piper)++
06:25 bacek_ probably we don't need piper anymore.
06:25 bacek_ moritz's bot is much better
06:41 Zaba joined #parrot
06:50 Psyche^ joined #parrot
06:54 iblechbot joined #parrot
07:27 masak joined #parrot
07:53 Ademan joined #parrot
07:59 bacek_ joined #parrot
08:52 Casan joined #parrot
10:16 zendman joined #parrot
11:47 smash joined #parrot
11:47 smash hello everyone
11:58 DietCoke joined #parrot
12:08 bacek msg pmichaud If you'll have any chance, please, take a look at http://bacek.homeip.net/cgit/parrot/tree/​languages/perl6/src/classes/List.pir?id=f​06c5247dcb39bd2f1ab8fd23a297f12ff0a3eaf (It's my 'draft' of lazy lists)
12:08 shorten bacek's url is at http://xrl.us/cpx8y
12:08 bacek yak.
12:08 bacek What is the problem with purl?
12:27 Zaba_ joined #parrot
13:14 cotto_yapc joined #parrot
13:27 dalek r28507 | Whiteknight++ | gsoc_pdd09:
13:27 dalek : [gsoc_pdd09] updating to trunk r28509
13:27 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28507
13:34 gryphon joined #parrot
13:48 cotto_yapc joined #parrot
13:58 magnachef_ joined #parrot
14:00 magnachef__ joined #parrot
14:04 iblechbot joined #parrot
14:08 kid51 joined #parrot
14:09 allison joined #parrot
14:32 Eevee rakudo patches go to parrotbug@parrotcode.org too, right?
14:33 pmichaud yes.
14:33 pmichaud or on the ticket that the patch is addressing.
14:34 Eevee didn't see one.  actually I'm wondering if I've misgrokked this problem; odd that nobody noticed
14:36 kid51 pmichaud: Which patches go to rakudobug@parrotcode.org?  And which to parrotbug?
14:36 dalek r28508 | pmichaud++ | trunk:
14:36 dalek : [rakudo]:
14:36 dalek : * Update docs/spectest-progress.csv to 2008-06-18.
14:36 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28508
14:39 Eevee kid51: google doesn't think that email exists
14:42 Eevee I know I keep crazy overthinking bugs I find and attempt to fix, but I'm wondering if this is just a general problem that parrot strings aren't getting promoted to Strs correctly
14:43 pmichaud string promotion to Str will be taken care of when we start using .HLL mapping.
14:44 pmichaud i.e., Parrot will take care of it for us.
14:44 Eevee oh, then this isn't really necessary at all
14:45 pmichaud I don't know how soon we'll be able to switch to .HLL .
14:45 pmichaud although having P6object helps.  I could go ahead and start a branch for it.
14:45 Eevee I think this is the third time you've told me a problem I found will be fixed by .hll mapping
14:46 DietCoke pmichaud: might be helpful for you to write up (if you havne't already) how you expect that to work in case someone else with cycles can get to it.
14:46 DietCoke (i fyou have expectations, you may not)
14:47 pmichaud I'm not sure how it's going to work.
14:48 dalek r28509 | Whiteknight++ | gsoc_pdd09:
14:48 dalek : [gsoc_pdd09] Fix arena sweep functions
14:48 dalek : * Use macros instead of function calls in inner loop
14:48 dalek : * Better macro support for common tasks
14:48 dalek : * Mark items on free list as GC_IT_CARD_FREE
14:48 pmichaud it's one of those dark icky corners of parrot that (unfortunately) will likely spread itself throughout the compiler tools
14:48 dalek : * Fix initial state flag in GC initializer function
14:48 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28509
14:49 Auzon joined #parrot
14:49 confound should I write a git-svn section for http://www.parrotcode.org/source.html?
14:51 pmichaud confound: yes.
14:52 pmichaud confound: even if it doesn't end up in source.html, I'm sure we'll want to make it available somewhere.
14:52 Eevee well that's a new one
14:52 Eevee set_pmc_keyed_str() not implemented in class 'RetContinuation'
14:52 pmichaud -G bug.
14:52 pmichaud (that's my bet.)
14:55 Eevee solving this np-complete problem in rakudo is hard
15:00 pmichaud the -G bug problem?
15:01 Eevee no; in the interest of making some practical use of rakudo, I'm trying to solve ita's overlapping movie title problem with perl 6
15:01 Eevee (fsvo "practical")
15:02 dalek r28510 | Whiteknight++ | gsoc_pdd09:
15:02 dalek : [gsoc_pdd09] Small fixes
15:02 dalek : * Fix loop logic for gc_it_enqueue_next_root
15:02 dalek : * Kill reference to non-existant gc_it_reset_cards function
15:02 dalek : * Turn tracing disambiguation function into a macro
15:02 dalek : * Add some information about GC_finish_FLAG handling
15:02 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28510
15:03 silug so who here at yapc is most familiar with the parrot build system?
15:04 silug for some reason, the rpm is trying to install stuff to /usr/runtime/.  that's not right, and it isn't immediately obvious where that's coming from.
15:11 confound pmichaud: ok.  where should I put it and how should I write it?
15:12 confound I mean, what format
15:12 Infinoid silug: I'm not at yapc, but ... various things get installed into $(PARROT_SRCDIR)/runtime, and so that path is hardcoded all over the place.  and last I heard, "make install" didn't work very well at all, and was considered unsupported
15:13 Infinoid the project is really geared toward running directly out of the source tree
15:13 Infinoid (for now.)
15:13 silug Infinoid: yeah, i know it is hardcoded...  the rpm build does a bunch of s,runtime/,%{_lib},
15:14 Infinoid sounds beautiful :)
15:14 silug and it was all working a year ago.  :)
15:14 silug (i haven't tried to build the rpm since 0.4.13)
15:14 Infinoid so, the rpm build is now missing some s,runtime/, stuff?
15:15 silug maybe.  but it is prepending /usr somehow
15:15 Infinoid isn't that just $(PREFIX)?
15:15 silug yeah, probably
15:15 silug grr...  should have realized it was that simple.
15:21 silug ooh, i see what's happening.
15:22 silug i think
15:25 Eevee joined #parrot
15:37 dalek r28511 | Whiteknight++ | gsoc_pdd09:
15:37 dalek : [gsoc_pdd09] Small fixes
15:37 dalek : * Assorted spelling, comment, and whitespace issues
15:37 dalek : * Add a few PARROT_ASSERT statements to double check
15:37 dalek : * Cut cruft out of Parrot_gc_it_pool_init
15:37 dalek : * Fix pointer problems in gc_it_add_arena_to_free_list
15:37 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28511
15:51 grim_fandango joined #parrot
15:53 wolverian joined #parrot
15:57 sjansen joined #parrot
15:59 NotFound Problem: I've made a build with --ccflags=-O3 and all parrot test pass, but in rakudo spectest_regression fails
16:03 NotFound perl6 make test also pass, forgot to say.
16:07 dalek r28512 | cotto++ | trunk:
16:07 dalek : added improved a gdbm test, making sure that the element count is consistent after a delete_keyed()
16:07 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28512
16:13 contingencyplan joined #parrot
16:14 dalek r28513 | jkeenan++ | trunk:
16:14 dalek : Applying patch submitted by Seneca Cunningham in http://rt.perl.org/rt3/Tic​ket/Display.html?id=52214.  Improve setting of flags in init::defaults, particularly re ppc vs i386.
16:14 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28513
16:14 dalek r28514 | Whiteknight++ | gsoc_pdd09:
16:14 dalek : [gsoc_pdd09] create basic versions of the write barriers for the new GC
16:14 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28514
16:15 japhb Grrrr ... why is purl constantly going offline?
16:17 IllvilJa joined #parrot
16:21 Whiteknight lazy
16:23 contingencyplan joined #parrot
16:30 dalek r28515 | Whiteknight++ | gsoc_pdd09:
16:30 dalek : [gsoc_pdd09] Fix root item handling to account for items added to the queue between increments
16:30 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28515
16:33 japhb Anyone happen to know donaldh's email?
16:34 moritz is he one of the soc students?
16:34 japhb I dunno ... he was in here earlier this week with some cygwin fixes among other things.
16:37 Whiteknight he's not in CREDITS eitehr
16:37 confound ]/w 22
16:41 smash Whiteknight: anyone applied any patch from him ?
16:44 Whiteknight I haven't, no
16:44 Whiteknight and I dont know of anybody else who has
16:44 japhb smash: We're working through one of his patches in RT right now.
16:44 japhb But he nopasted it, rather than making an RT
16:44 japhb (himself, I mean)
16:54 smash ok
16:56 smash if you apply it just remember to add him to credits
17:00 confound no one answered me earlier
17:01 confound if I wrote git-svn instructions for getting the parrot source, what format should they be in and where would I send them?
17:01 magnachef_ joined #parrot
17:02 moritz uhm, either the wiki that parrot uses, or POD + submit to docs/ (just guessing here)
17:02 japhb confound: docs/gettingstarted.pod?
17:03 japhb confound: (patch thereto, I mean)
17:03 confound does that end up at http://www.parrotcode.org/source.html?
17:03 confound oh, whatever.  gettingstarted.pod is probably fine
17:03 confound thanks
17:07 dalek r28516 | jkeenan++ | trunk:
17:07 dalek : Applying 2nd patch submitted in http://rt.perl.org/rt3/Tic​ket/Display.html?id=55856.  This should clear up any uninitialized value warnings in t/steps/auto_readline*.t.
17:07 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28516
17:08 smash confound: sorry, not sure.. but i think it's ok to send to parrot-porters
17:11 dalek r28517 | jkeenan++ | handledarwin:
17:11 dalek : Branch has been merged into trunk and is no longer needed at HEAD.
17:11 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28517
17:13 dalek r28518 | jkeenan++ | handledarwin-28441:
17:13 dalek : Branch to which this tag corresponded has been merged into trunk and tag is no longer needed.
17:13 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28518
17:15 barney joined #parrot
17:17 cjfields joined #parrot
17:28 cjfields pmichaud: checked backlog on #parrot, re: RT#55962, this is probably the same problem Eevee is describing
17:29 cjfields so it will have to wait for .hll mapping, my guess
17:31 ambs joined #parrot
17:36 GeJ joined #parrot
17:43 ambs where is purl?
17:48 smash ambs: sleeping
17:49 * smash off
17:59 Ivatar joined #parrot
18:00 dalek r28519 | kjs++ | trunk:
18:00 dalek : [chitchat]
18:00 dalek : + update layout a bit to make it easier for the eyes (IMHO)
18:00 dalek : + add some action methods
18:00 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28519
18:04 confound smthanks
18:04 confound oops
18:05 dalek r28520 | bernhard++ | trunk:
18:05 dalek : Remove trailing spaces.
18:05 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28520
18:05 Tene Bah, kjs isn't here for me to talk to.
18:05 iblechbot joined #parrot
18:07 toddr joined #parrot
18:09 packy joined #parrot
18:13 Zaba joined #parrot
18:15 cotto_yapc joined #parrot
18:16 dalek r28521 | bernhard++ | trunk:
18:16 dalek : [chitchat]
18:16 dalek : Add a MAINTAINER file.
18:16 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28521
18:21 toddr joined #parrot
18:22 toddr joined #parrot
18:39 ovid joined #parrot
18:40 ovid Cannot "make perl6" on OS X.  http://sial.org/pbot/31353  This is the latest subversion release.
18:42 ovid Is this the right room to ask this in?
18:44 pmichaud ovid: it's the right place, but many of us are at lightning talks at yapc
18:44 ovid Fair enough :)
18:44 NotFound ovid: Did you do make first?
18:44 ovid For parrot?  Yes.  Even submitted a patch for the failing test :)
18:45 NotFound No, in the languages/perl6 directory.
18:45 pmichaud looks like a parrot issue.try make clean perl6
18:45 confound you shouldn't have to
18:45 pmichaud "make perl6" ought to build a lot of ohter stuff first.
18:45 ovid Yes.  Similar failure.
18:45 NotFound If i remember well, sometimes make perl6 does not attemtp to rebuild the pbc
18:46 pmichaud "make clean perl6" has the same failure?
18:46 ovid make clean perl6:  http://rafb.net/p/QhcQS915.html
18:47 pmichaud it starts with "../../parrot"
18:47 pmichaud is that all of the output or just the last lines?
18:47 ovid Just the last lines.  There was a ton of stuff before hand that I trimmed.
18:47 toddr joined #parrot
18:48 ovid Should I paste it all?  Of course, if you're sitting in lightning talks, I understand if you want to pay attention to those.
18:48 pmichaud lightning talks are plagued by technical issues.
18:48 ovid Er, should "make clean perl6" be run from the top directory or from languages/perl6?
18:48 pmichaud from languages/perl6 is what I meant.
18:49 pmichaud But you might try 'make realclean' from the top directory and rebuilding parrot.  That really looks like a parrot issue to me.
18:49 confound ovid: do all the parrot tests pass?
18:49 ovid That's where I ran it from where it failed.
18:50 ovid confound: one TODO test passes and one coding standard test (trailing whitespace) fails.
18:51 ovid pmichaud: before you said "from languages/perl6", I started rerunning the "make clean perl6" in the top level directory and it ran a lot longer before getting teh same failure.  I'll try make realclean and rebuild parrot.
18:51 NotFound By the way, in the las build I have the message:  TODO passed: 157
18:52 ovid Same here.  It's a sprintf test.
18:54 toddr joined #parrot
19:00 magnachef_ joined #parrot
19:04 IllvilJa joined #parrot
19:06 malte joined #parrot
19:08 cotto_yapc joined #parrot
19:12 schmalbe joined #parrot
19:12 dalek r28522 | bernhard++ | trunk:
19:12 dalek : [Plumhead]
19:12 dalek : Add support for 'else' in NQP actions.
19:12 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28522
19:13 Zaba_ joined #parrot
19:16 schmalbe joined #parrot
19:16 dalek r28523 | bernhard++ | trunk:
19:16 dalek : [chitchat]
19:17 dalek : Set SVN properties for new file.
19:17 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28523
19:18 schmalbe Argh, 'svn commit' hangs again, but that might be a flakey connection
19:19 confound can I usefully send mail to parrot-porters@perl.org, or do I have to send to perl6-internals@?
19:19 moritz they are aliases
19:20 * confound figured
19:44 dalek r28524 | Whiteknight++ | gsoc_pdd09:
19:44 dalek : [gsoc_pdd09] Make pobject_lives much more streamlined and efficient so I dont have to worry about calling it more. Fix up gc_it_mark_children_grey to be better, and kill duplicate code between these two functions.
19:44 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28524
19:48 dalek r28525 | bernhard++ | trunk:
19:48 dalek : [Plumhead]
19:48 dalek : Add 'scalar_assign' to the NQP actions.
19:48 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28525
20:00 donaldh joined #parrot
20:10 dalek r28526 | bernhard++ | trunk:
20:10 dalek : [Plumhead]
20:10 dalek : Add 'array_assign' to NQP actions.
20:10 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28526
20:27 IllvilJa joined #parrot
20:33 dalek r28527 | bernhard++ | trunk:
20:33 dalek : [Plumhead]
20:33 dalek : Add support for string concatenation to the NQP actions.
20:33 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28527
20:35 cotto_yapc joined #parrot
20:37 toddr joined #parrot
20:37 ambs joined #parrot
20:47 Zaba joined #parrot
20:59 cjfields_ joined #parrot
21:12 IllvilJa joined #parrot
21:24 Zaba_ joined #parrot
21:27 bacek joined #parrot
21:33 Whiteknight holy cow, like 12 new rt tix in the past few hours
21:34 cotto_yapc woah
21:34 Eevee good!  depending on your pov
21:35 cotto_yapc at least 4 are non-bugs
21:35 * moritz submitted one or two TODO-bugs
21:43 jonathan Rakudo day tomorrow. :-)
21:43 * jonathan had hoped to get some hacking time on evenings too this week, but got sucked into @other_stuff.
21:44 slightlyoff joined #parrot
21:44 Eevee don't implement anything I could feasibly do  :P
21:44 slightlyoff left #parrot
21:44 Infinoid the trick is to lazily evaluate @other_stuff
21:45 Eevee then it might never end
21:45 cjfields joined #parrot
21:46 jonathan Eevee: Could you feasibly implement type-parametrized roles?
21:46 Eevee yes, in the same sense that I could solve a packing problem
21:46 Eevee so please feel free
21:47 jonathan Damm! I can't offload the hard bit. :-P
21:47 NotFound chr - The character specified by codepoint integer $2 in the current character set is returned in string $1.
21:47 NotFound What is the "current character set"?
21:47 Eevee utf8
21:47 Eevee always
21:48 Eevee dammit
21:48 Eevee (no idea)
21:48 NotFound Actually nothing, because that is not what the opcode do.
21:49 NotFound What it does is: < 128 - ascii , < 256 - latin1 , else utf8
21:50 Eevee jonathan: so what's the easy bit?  :P
21:50 jonathan For roles? I think I already did those... ;-)
21:51 jonathan Elsewhere - really depends what you consider easy and hard.
21:51 Whiteknight what are type-parameterized roles
21:51 NotFound And examples/shootout/mandelbrot.pir assumes that it will make a plain 1 byte encoding for 0-255.
21:51 Whiteknight ?
21:51 cjfields Eevee: did we ever work out the String issue with IO?
21:51 jonathan If you're into I/O, then there's plenty of filling out of methods there. Not to mention network I/O, which we curently have nothing on in Rakudo.
21:52 moritz Eevee: if you're looking for something easy to do, implement .method_call, which defaults to $_ as the invocant
21:52 moritz Eevee: it seems to be parsed already, and $_ already exists
21:52 Eevee cjfields: io.readline is spitting out native strings, pmichaud said .hll mapping will fix it, not sure if it's worth cluttering code to hackfix it in the meantime?
21:52 toddr joined #parrot
21:52 jonathan Whiteknight: role Foo[Bar] # a role Foo that also takes another type as a parameter do it, when you compose it
21:52 cjfields k
21:52 cognominal I just redefined a method in a grammar and did not see a warning/error
21:52 jonathan sldo/to/
21:53 moritz don't know if it's really easy though
21:53 Whiteknight johnathan, thanks
21:53 Eevee moritz: will have a look when I'm on something more powerful than a phone, hanks
21:53 cognominal is there a policy for that?
21:53 Whiteknight i'm not down with all this cutting-edge P6 lingo
21:53 cjfields The same problem exists with Match and split
21:53 cjfields filed an RT ticket yesterday on it
21:54 japhb jonathan: A while back a few of us were asking for being able to use modules from non-Perl 6 namespaces.  What's still blocking that?
21:55 Eevee cjfields: what exactly is .hll mapping anyway  :(  apparently it will solve all problems and cure cancer
21:55 Whiteknight .hll?
21:55 Whiteknight .HLL?
21:55 moritz it's called "magic" everywhere else ;)
21:55 cjfields yes .HLL
21:55 NotFound Eevee: not all, a bunch are expecting redoing of PIO.
21:56 japhb Eevee: tells what type a High Level Language wants to see instead of the low level Parrot types, IIRC.
21:56 Whiteknight purl, .HLL is going to bring democracy to Myanmar
21:56 Eevee plane taking off &
21:56 cjfields yes, that's correct
21:56 Eevee ah
21:57 cjfields "I have suffered from ED, now I take .HLL..."
21:57 Whiteknight HA!
21:57 Whiteknight .HLL?
21:57 Whiteknight purl is the worst
21:58 jonathan japhb: As far as I'm aware, pmichaud was looking into it - last I heard the issue was that we needed some standard way to ask a certain HLL to load a module and put stuff into a namespace.
21:59 japhb Yes, and we discussed it for a while, and he said something along the lines of "OK, everything sounds reasonable", and then I haven't heard anything since.
21:59 jonathan I may be wrong, but my impression was pm had taken this to work on, and seemed to have what was needed in his head better than I, so I left off it.
21:59 japhb nod
21:59 jonathan Yes, I remember that being the conclusion.
22:00 cjfields The 'mismapping' issues (getting a String instead of Str) seem to be common enough
22:01 jonathan japhb: OK, pmichaud and I should clarify whether that's something he really wants to do, or if it's a task for the taking. I'm certainly keen for it to be done.
22:01 cjfields perl6: my @foo = 'AB'.split(''); say @foo; @foo[0]++;
22:01 polyglotbot OUTPUT[AB␤increment() not implemented in class 'String'␤current instr.: 'postfix:++' pc 10973 (src/gen_builtins.pir:7331)␤called from Sub '_block11' pc 116 (EVAL_10:41)␤called from Sub 'parrot;PCT::HLLCompiler;eval' pc 806 (src/PCT/HLLCompiler.pir:469)␤called from Sub
22:01 polyglotbot ..'parrot;PCT::HLLCompiler;evalfiles' pc 1088 (src/PCT/HLLCompiler.pir:598)␤call...
22:02 jonathan We sure do need .HLL...
22:02 Whiteknight did the issue with string ranges get fixed since yesterday?
22:02 cjfields perl6: say 'A'..'Z';
22:02 polyglotbot OUTPUT[ABCDEFGHIJKLMNOPQRSTUVWXYZ␤]
22:02 cjfields yep
22:02 Whiteknight !!!
22:02 Whiteknight perl6++
22:02 cjfields pmichaud++
22:03 Whiteknight pmichaud++
22:03 packy Quick question: if I have a patch to the code that generates the makefile for Rakudo, should that be submitted to parrotbugs or to some bug tracker for Rakudo?
22:04 moritz packy: parrotbug@perl.org is fiine
22:04 packy (I found that MACOSX_DEPLOYMENT_TARGET wasn't defined while doing a "make perl6", so I added code to define it)
22:04 moritz packy: that will open a ticket
22:04 packy Coo.
22:04 Whiteknight fo sho
22:04 packy I'll submit it once I'm sure it compiles.  :)
22:04 jonathan packy: Yes, as moritz said. It sounds like it's a general fix that will help other languages too. :-)
22:05 moritz maybe that also fixes ovids problem from earlier today
22:05 packy Well... I had to edit the root.in file for perl6.
22:06 packy I think we'd have to edit the base file for each language, but I can do that.
22:06 jonathan Ah, OK.
22:06 Whiteknight packy++
22:06 dalek r28528 | Whiteknight++ | gsoc_pdd09:
22:07 dalek : [gsoc_pdd09] sprinkle "PARROT_ASSERT" liberally on everything.
22:07 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28528
22:07 moritz if you edit stuff in other languages, send to parrotbug@perl.org instead
22:07 Infinoid moritz =~ s/perl.org/parrotcode.org/
22:07 Infinoid (or does @perl.org work now?)
22:07 moritz parrotbug?
22:07 moritz where's purl?
22:09 Tene perl6: 'foo'.split('')[0].WHAT()
22:09 polyglotbot RESULT["Str"]
22:09 Tene perl6: 'foo'.WHAT()
22:09 polyglotbot RESULT["Str"]
22:09 moritz perl6: say 'foo'.split('')[0]
22:10 polyglotbot OUTPUT[f␤]
22:10 japhb Infinoid: did you ever get your OpenGL problems resolved?
22:11 Infinoid japhb: thanks for the reminder, now's a good time to try it
22:11 japhb excellent
22:12 Infinoid seems like just running one opengl app ruins it for everyone else.  time to switch back to a stable mesa.
22:12 japhb So glxgears broke it too?
22:13 Infinoid I ran /usr/lib64/misc/xscreensaver/glschool, which worked, but didn't work when I restarted it
22:13 japhb ah
22:14 pmichaud (Str vs String vs Perl6Str):  In general I'd prefer to wait for .HLL mapping than to add a lot of workarounds that we'll later have to take out.
22:14 pmichaud That said, if there are places where Parrot internals are leaking out that are likely to affect applications (as opposed to tests), then we can see about fixing those for now.
22:16 pmichaud packy:  In general, fixes for Rakudo should go to rakudobug@perl.org .  But if they go to parrotbug we'll switch them over (they both end up on the same rt system, although not the same queue).
22:16 jonathan pmichaud: Does getting .HLL working imply we need to worry about namespace stuff (hard/messy), or is there a way just to get the type mappings (much less messy)?
22:16 pmichaud ...namespace stuff?
22:16 jonathan As in, does a .HLL declaration automatically mean our set_hll_global and so forth starts referring to the perl6 HLL namespace rather than the parrot one?
22:17 pmichaud .HLL will cause rakudo's namespaces to be 'perl6', yes.
22:17 pmichaud we'll have to import the protoobjects and things we need from other namespaces.
22:18 pmichaud (or figure out how we want to reference them in other hll namespaces)
22:18 pmichaud since it affects almost the entire toolchain, I think it's something I prefer to do myself at this point.
22:19 japhb Which brings up my related question about using modules in other namespaces ...
22:19 pmichaud I see that as a different issue, actually.
22:19 pmichaud (maybe it's not a different issue, but thus far they don't have the same feeling to me.)
22:20 japhb It feels overlapping to me, at least.
22:20 pmichaud it maybe
22:20 pmichaud part of the difference:
22:21 pmichaud for most modules like, say, SDL or Postgres or OpenGL, those modules shouldn't have to be aware that they're being used by a separate HLL.
22:21 pmichaud however, the compiler tools (P6object, PCT, PGE) definitely have to be aware of the namespace in which they're being used
22:22 pmichaud if I use P6object to create a new class, it has to know how to create that class in the correct HLL namespace.
22:23 pmichaud (even though P6object itself is running in the 'parrot' hll namespace)
22:23 japhb nod
22:24 teknomunk joined #parrot
22:24 pmichaud anyway, it's on my list of things to address in the relatively near future.
22:24 pmichaud I expect it to occur before PGE changes, for example.
22:24 japhb great!
22:25 pmichaud on the issue of loading non-Perl6 modules, I'd like to see that happen within the next two weeks.  Certainly before OSCON.
22:26 pmichaud I was hoping to be able to have a discussion about it here at YAPC::NA, but the people involved were all busy (including myself), so I suspect it'll happen via the mailing list.
22:26 japhb When is OSCON this year?
22:28 pmichaud 3rd week July
22:28 pmichaud (but I'm on vacation for 5 days prior, so 2nd week July for HLL)
22:28 pmichaud certainly in time for the next Parrot release.
22:29 pmichaud ...unless we run across a huge stubmling block.
22:29 cjfields ...darned stubmling blocks...
22:29 japhb right
22:29 pmichaud sorry.  Gotta run, closing down here.
22:29 pmichaud bbl.
22:39 cotto_yapc joined #parrot
22:47 NotFound Can someone test my last patch to #39930 and check if it is appliable to #50092 ?
23:13 cjfields joined #parrot
23:41 tetragon joined #parrot
23:53 bacek_ joined #parrot

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

Parrot | source cross referenced