Camelia, the Perl 6 bug

IRC log for #parrot, 2008-10-15

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 nopaste "Infinoid" at 96.238.213.50 pasted "more info" (46 lines) at http://nopaste.snit.ch/14295
00:01 dalek r31964 | cotto++ | trunk:
00:01 dalek : [t] update tests to check that dynpmcs and dynops have test coverage
00:01 dalek : patch courtesy of Rafael Sanchez
00:01 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31964
00:01 Infinoid anything in the backtrace look likely?  I'm not really sure how to hunt this down further.
00:02 Infinoid driving home, back soon &
00:05 particle bah, missed him by this much
00:09 AndyA joined #parrot
00:11 dalek r31965 | Whiteknight++ | calling_conventions:
00:11 dalek : [Calling_conventions] A few changes that fix the segfault I was having earlier, but introduce a new segfault later in the build process.
00:11 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31965
00:12 Zaba_ joined #parrot
00:20 dalek r31966 | chromatic++ | trunk:
00:20 dalek : [OO] Avoided an exception in fail_if_type_exists(), where the code attempts to
00:20 dalek : stringify the name of a class if someone has already registered it.  Because
00:20 dalek : the "name" of the class may be a ResizableStringArray and because you can't
00:20 dalek : stringify an RSA in a same way, this exception would get lost in an exception
00:20 dalek : in the Default PMC's get_string() vtable entry.
00:20 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31966
00:22 chromatic 0.89% speed improvement in my benchmark.
00:22 chromatic At least it cleans up the callgrind output somewhat.  longjmp confuses it terribly.
00:24 Whiteknight longjmp is one of those esoteric holdovers that is only really useful for virtual machine developers
00:24 Whiteknight lucky us
00:26 chromatic Yes.  We rock.  Woo hoo.
00:27 Whiteknight I almost can't think of a single rational use for longjmp besides implementing an exception system
00:27 Whiteknight I can think of plenty of irrational ones :)
00:29 chromatic Avoiding stack smashing.
00:29 chromatic Cheap coroutines.
00:29 chromatic When you really really don't want to pay the overhead of your C and libc and processor ABI but you're too scared of FORTRAN.
00:44 dmknopp left #parrot
00:47 chromatic purl msg pmichaud See compilers/pge/PGE/Exp.pir:99 -- this seems to be the source of the multiple class registration attempts.
00:47 purl Message for pmichaud stored.
00:52 dalek r31967 | Whiteknight++ | calling_conventions:
00:52 dalek : [Calling_conventions] Fix it so it compiles again, without undoing any of the changes I've been making.
00:52 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31967
01:19 cotto would this be a good time to re-apply, test and commit the patch from #52778 (Are resizable arrays too Perlish?)
01:20 cotto all tests (core and perl6) pass
01:24 cotto (using string_from_literal instead of CONST_STRING for resizablestringarray.pmc)
01:24 cotto hmm.  bbl.  I'll commit then.
01:29 Coke whatever ticket I had tene work on earlier is closable now.
01:29 Coke if someone would like to make chromatic happy.
01:32 Coke Ah. I also am suffering from the TGE bug. (where PGE is now autosplitting on :: but TGE ain't.
01:32 * Coke digs up that ticket.
01:34 Tene Coke: !
01:34 Coke it's not a ticket, huh.
01:34 * Coke responds on the thread to parrotbug to make it a ticket.
01:34 Tene Coke: Can you describe what's happening?
01:35 Tene Coke: Or make a more restricted test case?
01:35 Coke ... PGE is autosplitting on :: but TGE isn't?
01:35 Coke so PGE puts things in the namespace "TclExpr"; "Grammar", but TGE is still looking at "TclExpr::Grammar" ?
01:35 Tene Coke: I don't understand TGE well enough yet to see where that needs to happen.
01:36 Tene If you can give me an example in the generated PIR, it will help me a lot
01:37 Coke .namespace [ 'TclExpr::PAST::Grammar' ]
01:37 Coke $P1 = subclass 'TGE::Grammar', 'TclExpr::PAST::Grammar'
01:37 Coke .return tree.get('dispatch', node, 'TclExpr::Grammar')
01:38 Tene I guess I need to check out TCL, then.
01:38 Coke eh, probably findable in pheme.
01:38 Coke I am obviously more familiar with tcl, though.
01:38 Tene I was having trouble figuring out pheme
01:38 Coke well, I'm here and available to blather about tcl. =-)
01:39 Tene I wonder if TGE needs to start producing code that uses P6object
01:39 Coke I hope not.
01:40 Coke but then I think of p6object as a perl 6 only thing.
01:41 Coke I don't see any examples in t/compilers/tge/*.t that rely on the :: split.
01:42 Tene No, there are none
01:42 Tene I was asking earlier for a failing test in t/compilers/tge
01:45 Tene Think you can make one, or do I need to actually read TGE?
01:49 Coke I probably can condense the failing bits of tcl into a test, but despair of doing so tnoight.
01:50 Tene That would make it much easier for me.  Think you can do it this week?
01:50 Coke probably.
01:50 purl Really? Probably? Are you Certain it's not certain? Are you sure it's unsure? I think you need to look harder.
01:50 Tene Fantastic.
01:51 Tene Coke++
02:01 jrockway_ joined #parrot
02:12 Zaba joined #parrot
02:29 Zaba_ joined #parrot
02:34 xiaoyafeng joined #parrot
02:57 particle1 joined #parrot
03:05 Zaba joined #parrot
03:06 bitlbee joined #parrot
03:07 jrockway joined #parrot
03:26 confound yesterday on #perl I suggested a languages/sendmail.cf directory for parrot
03:26 confound no one volunteered.
03:30 Infinoid isn't that ... m4?
03:31 Infinoid or was m4 invented to make sendmail.cf slightly less painful?  (I forget.  sendmail is responsible for years of therapy bills.)
03:39 Tene Mmm... sendmail...
03:51 notbenh joined #parrot
04:23 confound Infinoid: m4 was invented to make sendmail.cf slightly less painful
04:23 confound so you could say things like SEND_MY_MAIL_TO(the.right.place) instead of... whatever
04:36 Bzek joined #parrot
04:42 dalek r31968 | cotto++ | trunk:
04:42 dalek : [pmc] make resizable array PMCs not be too Perlish (closes #52778)
04:42 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31968
04:54 MariachiElf joined #parrot
04:54 MariachiElf left #parrot
04:54 MariachiElf joined #parrot
04:58 johbar joined #parrot
05:37 Zaba joined #parrot
05:41 chromatic joined #parrot
05:42 davidfetter joined #parrot
06:12 uniejo joined #parrot
06:16 chromatic t/pmc/resizablebooleanarray        (Wstat: 512 Tests: 23 Failed: 2)
06:16 chromatic Failed tests:  22-23
06:17 Zaba_ joined #parrot
06:23 grim_fandango_ joined #parrot
06:29 cotto that's odd.  It passes for me
06:30 moritz fails here too
06:32 Coke joined #parrot
06:37 cotto what happens when you run the test with perl instead of prove?
06:39 cotto that'll say exactly which part of the tests are failing
06:40 nopaste "moritz" at 89.13.205.64 pasted "resizablebooleanarray.t for cotto" (14 lines) at http://nopaste.snit.ch/14296
06:46 omega joined #parrot
07:01 cotto moritz, x64?
07:05 moritz cotto: no, 32 bit (Debian GNU/Linux 4.0)
07:06 cotto hmm.  I'm trying to make realclean and rebuild
07:10 cotto oh.  apparently I have a locally modified (as in fixed) version of that test I forgot to commit
07:11 cotto I need to clear out my working svn dir so I can find that kind of thing more easily
07:12 dalek r31969 | cotto++ | trunk:
07:12 dalek : [t] fix a test that broke after the previous commit
07:12 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31969
07:14 dalek r31970 | particle++ | trunk:
07:14 dalek : [rakudo] implement tagsets in 'is export()' trait
07:14 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31970
07:14 dalek r31971 | cotto++ | trunk:
07:14 dalek : [strings] add a TODO concerning the hashval caching bug
07:14 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31971
07:14 dalek r31972 | particle++ | trunk:
07:14 iblechbot joined #parrot
07:14 dalek : [rakudo] remove commented out code
07:14 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31972
07:15 cosimo joined #parrot
07:19 cotto chromatic++ for reporting that
07:19 cotto moritz++ for the same reason
07:19 cotto karma foo
07:19 purl foo has karma of 16
07:19 cotto foo++ foo++
07:19 cotto karma foo
07:19 purl foo has karma of 17
07:20 viklund joined #parrot
07:20 davidfetter karma davidfetter
07:20 purl davidfetter has karma of 3
07:20 davidfetter that's odd
07:21 cotto davidfetter++ #now we're even
07:22 * particle1 saw tom lane last weekend
07:22 davidfetter cotto, heh
07:23 davidfetter yeah, i heard about that. he's a good speaker :)
07:26 particle work in < 6 hours. bed &
08:16 barney joined #parrot
08:47 bacek joined #parrot
08:48 Zaba joined #parrot
08:53 tomyan joined #parrot
09:12 donaldh joined #parrot
09:43 Debolaz joined #parrot
10:36 kj joined #parrot
10:44 dalek r31973 | kjs++ | trunk:
10:44 dalek : [NEWS] added news for compilers/pirc
10:44 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31973
11:29 xiaoyafeng joined #parrot
11:39 Andy joined #parrot
12:19 * Coke wears his GSOC t-shirt to work today, presumably in violation of the dress code
12:38 iblechbot joined #parrot
12:43 Andy joined #parrot
13:01 Coke Tene: I gave you about 75% of what you need, I think.
13:30 grim_fandango joined #parrot
14:03 gryphon joined #parrot
14:10 pmichaud I don't quite understand why PGE's splitting namespaces on '::' has anything to do with TGE.
14:11 pmichaud oh, never mind, I do understand why.
14:15 Coke did languages get tested before that got merged back to trunk, OOC?
14:16 pmichaud many did, yes.
14:16 pmichaud tcl wasn't in the repo at that point, though.
14:16 pmichaud I don't know if pheme was tested.
14:16 Coke I was thinking more abou... right.
14:16 pmichaud the merge-to-trunk happened while I was on the plane ... I didn't know it was coming.
14:16 Coke I don't expect tcl to be tested, even when it -is- in the repository. =-)
14:17 pmichaud I think they pushed the merge quickly so that particle would have an up-to-date copy of trunk on his plane ride, or something like that.
14:18 particle yes, that's right, however my flight left hours before yours, so...
14:18 pmichaud ...it did?  what time did you leave?
14:19 Coke should probably add things that require a separate test run to the branch docs (pheme? nqp...)
14:19 particle 11:15am for airport
14:19 pmichaud oh, I was already in the air at that point :-)
14:19 pmichaud assuming you mean 11:15am PDT :-)
14:19 pmichaud my flight left 12:30pm CDT, which would be 10:30am PDT
14:20 particle oh, i got confused then on friday, sorry
14:20 * jhorwitz is sad pmichaud missed the lolcats by 10 mins
14:20 pmichaud me too.  but I'm very glad I made my flight.  I heard lolcats was the best talk of the set.
14:21 particle it was really funny
14:21 jhorwitz making your flight is a good thing these days.  :)
14:22 * Coke vaguely worries about his flight to the uk which is, what, six weeks out?
14:23 pmichaud speaking of which, do we have any more details about the Parrot summit?
14:23 pmichaud that's 4.5 weeks out or something like that
14:25 pmichaud ...so, what should we do about TGE?
14:25 particle i need some help with the wiki
14:25 Tene jhorwitz: what's this about lolcats?
14:26 particle i need to set up a survey&|spreadsheet&|table
14:26 particle something where folks can answer questions, like:
14:26 particle ~ are you planning on attending the summit
14:26 particle ~ if no, would you attend if your travel was funded?
14:26 Khisanth joined #parrot
14:27 particle ~ if yes, what are your expected flight expenses?
14:27 particle ~ do you mind sharing a room at the summit?
14:27 Tene Is the summit something that I should want to go to?
14:27 particle after i have answers to the survey, i can talk to the hotel
14:27 particle tene: yes
14:27 jhorwitz Tene: i did a lightning talk, "the lolcat history of perl6 and parrot".  50 slides of lolcats.
14:27 Tene jhorwitz: posting the slides?
14:27 particle tene: i'd like you there. it's for core parrot devs and hll devs
14:28 jhorwitz i'm working on transcribing the talk first --  the slides make no sense otherwise.  :)
14:29 * Tene needs to quickly get dressed and run to work.
14:29 particle drink more espresso!
14:30 lasermike026 jhorwitz: slacking again.
14:30 jhorwitz koko!
14:30 lasermike026 yo.
14:33 particle hey, it's like a little reunion.
14:35 * jhorwitz looks for a smiley with a cigar...
14:35 lasermike026 that's right, you work at home.
14:36 particle pmichaud: have a look at the code i stuck in last night and let me know if it needs changing
14:36 particle actions.pm ~850-900
14:36 pmichaud particle: I looked at it already -- it looks fine to me, although I think it could use some minor refactoring.
14:36 particle like what?
14:36 purl i think like is that generally frowned upon by ISPs?
14:37 particle i want to refactor other code based on this sample
14:38 pmichaud I wouldn't check for 'infix:,', for example.
14:39 particle just isa and pasttype?
14:39 pmichaud yes
14:39 particle ok, that's the one area i was a bit paranoid
14:39 pmichaud also, I think that the section that builds the binding code from @export_ns might need to be hoisted a little higher in processing
14:39 pmichaud since 'multi' will also have the effect of 'is export'
14:39 particle btw once i saw the light about introspecting past, the code flowed like a river
14:40 particle pmichaud: yes, i need to write tests for that first, then i'll refactor based on them
14:41 pmichaud so, that part might need to be in routine_declarator instead of routine_def
14:41 pmichaud which would mean that @export_ns acts like an attribute on the PAST::Block node somehow
14:41 particle that might help for subpackages, too
14:41 pmichaud yes.
14:41 particle fab. are subpackages working yet?
14:42 pmichaud not as far as I know.
14:42 particle ok. i'll test it.
14:43 pmichaud I'm a little surprised by
14:43 pmichaud my $tag := $_<named><value>;
14:43 pmichaud I would've just guessed $_<named>
14:44 particle doesn't $_<named> include the colon? i'll double-check
14:44 pmichaud are you talking about the PAST or the grammar?
14:44 pmichaud the named params don't normally include the ':'
14:45 particle grammar. ok, i'll look at the parse again
14:45 pmichaud my point is that it's not the parse at that point
14:45 pmichaud $_ is a PAST::Val node
14:46 pmichaud or, even more likely,   $_.named()
14:47 particle ah, there i thought "i need the string, i'll grab it from the parse"
14:48 pmichaud better is to grab it from the ast, I think.
14:48 particle so, yeah, you're right. should have used the past there
14:49 particle so, it's my feeling that any time you use something more than once, it should be bound to a variable to save instructions. does that make sense?
14:49 particle $_<foo><bar>[0] can get expensive
14:49 pmichaud I'm more interested in code clarity.
14:49 particle it's clearer with variables, too, i suspect
14:49 pmichaud yes.
14:49 pmichaud I was about to write "in general, binding expressions to a variable makes things clearer."
14:50 pmichaud otoh, binding to a variable that isn't reused is more expensive
14:51 Andy joined #parrot
14:51 pmichaud my $name := $aux<name>;
14:51 pmichaud ##  is export(...)
14:51 pmichaud if $name eq 'export' {
14:51 particle yes, correct on binding.
14:51 pmichaud I would've written that as
14:51 pmichaud if $aux<name> eq 'export' {
14:51 particle no, because eventually other traits will be processed
14:52 particle so $name will be reused
14:52 pmichaud in which case I'm hoping we get out of the if/then/else mode for that
14:52 particle yes please!
14:53 PerlJam There sure seems a lot of stair-step effect in that code.
14:54 particle i'm not fond of if $foo { for $foo {, but it's necessary in many places
14:54 PerlJam why?
14:54 PerlJam Do you sometimes get a boolean false that you don't want to iterate over?
14:55 pmichaud PerlJam: undef
14:55 pmichaud $foo := $<something>
14:55 pmichaud for $foo { ... }
14:55 PerlJam last unless $foo;  :-)
14:55 pmichaud where $<something> ends up being undef
14:56 pmichaud last doesn't help here, because we're trying to iterate over an undef.
14:56 pmichaud i.e., it breaks before even getting into the loop.
14:59 PerlJam So ... you guard the loop when you don't want to iterate over an undef, and you don't when you do want to iterate over an undef?
14:59 pmichaud PerlJam: the problem is that "iterate over undef" isn't defined in Parrot.
15:00 pmichaud $P0 = new 'Undef'
15:00 pmichaud $P1 = iter $P0    # error
15:00 particle ayep
15:00 PerlJam error == die a horrible death?
15:00 pmichaud "get_iter not supported in Undef" or somesuch.
15:02 pmichaud particle:                          my $export_ns_base := ( $?NS ?? $?NS ~ '::' !! '' ) ~ 'EXPORT::';
15:02 pmichaud looks odd.
15:03 pmichaud why the ternary?
15:03 particle if you're in the main package, you don't want ::
15:03 pmichaud uhhhh why not?
15:03 particle main package, no package
15:03 pmichaud well, technically we're always in a package.
15:03 particle use v6; say $?PACKAGE; # says what?
15:03 pmichaud 'Main'
15:04 particle that's not true now
15:04 pmichaud correct, it's not.
15:04 particle currently it's ''
15:04 pmichaud but even so, the leading ::'s wouldn't hurt.
15:04 particle ok then
15:04 pmichaud I would just do   $?NS ~ '::EXPORT::'
15:06 particle so how do we break out of if/elsif/else?
15:06 particle multimethods?
15:06 purl multimethods are the extension in the other direction
15:07 pmichaud probably.  I'm not going to worry about it just yet until I see how the other traits refactor.
15:08 Lorn joined #parrot
15:09 PerlJam perhaps it's just me, but someone seems to have cleaned up actions.pm in a good way.  It seems easier to read than when last I looked at it.
15:09 pmichaud I've been iteratively removing inline PIR sections and refactoring things, yes.
15:10 pmichaud but the stuff I've been doing hasn't been primarily aimed at readability -- more at correctness.
15:10 PerlJam well, the side effects are excellent  :)
15:17 PerlJam $?BLOCK.symbol('___HAVE_A_SIGNATURE')     heh, that looks like a hack waiting on something else to be implemented.
15:19 particle indeed, there are still ugly hacks
15:19 particle some due to nqp limitations, some due to human limitations
15:21 hercynium joined #parrot
15:22 uniejo joined #parrot
15:23 PerlJam That particular hack doesn't have a #TODO or #XXX associated with it.  Newish contributors to rakudo might propagate the hack through-ou code not realizing it's a hack.
15:23 PerlJam s/-ou/-out/
15:24 particle actions.pm needs a good once over
15:25 particle i keep hoping to do it, but adding features is sexier
15:25 particle so i do bit-at-a-time as i add features
15:25 particle i will respond to requests for clarity, though
15:25 PerlJam that's the sane way
15:30 pmichaud (propagate hack)   might --> have been
15:30 pmichaud a lot of the hacks come from trying to inspect the parse tree instead of the ast
15:36 dalek r31974 | julianalbo++ | trunk:
15:36 dalek : fix PMC_IS_NULL used instead of STRING_IS_NULL in oo.c
15:36 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31974
15:38 johbar joined #parrot
15:42 Zaba joined #parrot
15:45 particle my $export_ns_base := ~$?NS ~ '::EXPORT::'; # causes test failures
15:46 particle does that mean Perl6::Compiler.parse_name() needs fixing? looks like.
15:50 pmichaud it shouldn't -- I was using it w/o difficulty before.
15:50 pmichaud I know that it's supposed to be stripping out empty namespaces.
15:51 pmichaud and ignoring leading colons.
15:51 particle yes, that's what it should do. hrmm
15:51 pmichaud I wonder if $?NS isn't set.
15:51 pmichaud although if that's the case it should still stringify to ''
15:51 particle right
15:52 particle i even added prefix:~
15:52 particle maybe it's .named() that's causing the failures, i'm looking
15:52 pmichaud does it cause lots of test failures or just in the S11 tests?
15:52 particle i only ran s11
15:52 pmichaud the tests don't compile, or they run and fail?
15:52 particle t\spec\S11-modules\export.t               30    8  11-12 15-16 19-20 24-25
15:53 pmichaud it's possible that parse_name is broken.
15:53 pmichaud might need some 00-parrot tests for that.
15:54 particle it's also possible that $_.named() is causing the failures, since that defines the namespace names
15:54 particle the failures are during === =:= tests
15:54 pmichaud how could that be causing the failure?  you mean it's returning the wrong value?
15:54 particle yes
15:54 particle but it's more likely parse_name
15:54 pmichaud well, we already know it's non-null.
15:55 pmichaud I would do one change or the other and test
15:55 pmichaud rather than both (and not know which)
15:55 particle yep
15:56 particle already rebuilding with $?NS ternary
16:06 pmichaud afk, lunch
16:11 davidfetter joined #parrot
16:18 particle the trouble was in replacing $_<named><value>
16:23 rdice joined #parrot
16:35 pjcj joined #parrot
16:41 chromatic joined #parrot
16:50 Theory joined #parrot
16:50 dalek r31975 | julianalbo++ | trunk:
16:50 dalek : fix some signed-unsigned warnings
16:50 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31975
16:51 dalek r31976 | particle++ | trunk:
16:51 dalek : [rakudo] some refactorings suggested by pmichaud++
16:51 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31976
16:53 Tene particle: in that case, I'd love to go, but I don't anticipate having money available for travel.
16:53 particle tene: we may be able to fund your travel
16:53 particle that's why i need to set up the questionaire
16:54 particle but i can't figure out a good way with that wiki software
16:55 dalek joined #parrot
16:56 Tene particle: I'd also need to arrange for time off from work... what's the date?
16:56 sjansen joined #parrot
17:00 particle 15-16 nov
17:01 particle http://www.parrot.org/wiki/p​arrot-developer-summit-2008
17:02 Tene Hmm.
17:02 Tene If there's a class the week before, I might be able to get work to fly me there after teaching instead of home.
17:08 mj41_ joined #parrot
17:08 particle tene, jhorwitz; could you answer the questions on the wiki? http://www.parrot.org/wiki/p​arrot-developer-summit-2008
17:08 Tene Do I have write access to the wiki?
17:08 particle cool, that would make it easier
17:08 particle hrmm, it would be a problem if you don't
17:08 particle but i can fix that
17:10 particle you should be able to add yourself as a user
17:10 dalek r31977 | julianalbo++ | trunk:
17:10 dalek : fix a const issue in multidispatch that was creating lots of warnings
17:10 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31977
17:13 * Tene is waiting for registration email to greylist.
17:20 pmichaud okay, I'm on the list.
17:23 moritz when is the nex parrot dev event in europe? ;-)
17:26 PerlJam moritz: Take the survey  :)
17:27 PerlJam btw, I like "Parrot speaks your language"
17:28 chromatic Europe?  Europe still exists?
17:29 * NotFound plays 'The final countdown'
17:29 pmichaud Europe is the place that is telling the U.S. how to get out of its financial meltdown :-|
17:30 * pmichaud notes that "financial meltdown" and "the final countdown" sound a lot alike....
17:30 Infinoid uck.
17:30 PerlJam too many syllables in the former
17:30 pmichaud I cound five in both.
17:30 pmichaud *count
17:31 dalek r31978 | chromatic++ | trunk:
17:31 dalek : [OO] Fixed PMC_IS_NULL() check to check the right PMC (see r31966 and r31974).
17:31 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31978
17:33 pmichaud should updating parrot.org be included in docs/project/release_manager_guide.pod ?
17:34 pmichaud oh, it's there now
17:34 pmichaud just didn't see it with the other updates
17:35 pmichaud should we be using parrotcode.org or parrot.org as the primary url for Parrot?
17:35 pmichaud e.g., /topic
17:36 particle parrot.org i think, but migration is incomplete
17:36 particle the parrot.org homepage should say something about that
17:44 Ivatar joined #parrot
17:44 Zaba In perl6, is it possible to make an operator like 'something' first param 'delimiter' second param 'somethingelse'?
17:45 moritz I don't fully understand the question, but probably "yes"
17:45 moritz do you want to define an infix operator? Or a ternary?
17:45 TimToady at token is always allowed to parse sub-expressions as part of itself
17:46 TimToady *a
17:46 Zaba An operator | ; |, that'd take 2 params, x and y, used as | x ; y |..
17:47 TimToady well, that would be ambiguous with prefix:<|>
17:47 moritz but if you choose a differten symbol.. then yes.
17:48 TimToady (don't tell Zaba we've used up all the other symbols too :)
17:48 Zaba | ; | was only an example..
17:48 moritz TimToady: with the exception of ` and .ord > 127
17:48 moritz ;)
17:48 TimToady in fact, the ternary rule in STD simply treats ??<EXPR>!! as a "fat token" binary
17:48 moritz we've taken all the good ones, though ;)
17:49 Infinoid syntax highlighters love unary `
17:50 TimToady and basically all circumfixes in STD treat '(' ~ ')' <EXPR> as single tokens when viewed from above
17:50 moritz in Perl 6 syntax hilights you ;)
17:51 ruoso joined #parrot
18:31 Coke pmichaud: is 59006 resolved to your satisfaction?
18:32 pmichaud Coke:  yes, it is.  I'm a very satisfied customer there.
18:33 pmichaud Do you want to resolve it or shall I?
18:33 Tene Which airport would I fly into to get to mountain view?
18:33 pmichaud I think SJC is closest
18:33 pmichaud follow by SFO and OAK
18:33 Coke I can do it.
18:34 pmichaud Tene: see http://code.google.com/events/visitors/
18:34 Tene Clever.
18:35 pmichaud AA flies direct to SJC from DFW, so I'll probably go that route.  SWA also flies to SJC, but it's anything but direct (because we can't get direct from DAL to _anywhere_)
18:35 Tene SLC is a delta hub, which is nice
18:36 Tene Hmm.  Looks like I could fly for not much more than $200
18:36 pmichaud there's a $200 fare via Frontier Airlines from DFW to SJC (1 stop, I don't know where)
18:37 Tene What time will things start on Saturday?
18:38 Tene There, page updated.
18:38 Tene Need to consult with boss, though.
18:41 particle we have a room at google from 9-5 sat&sun
18:42 particle and they'll provide lunch
18:44 pmichaud I'm guessing arrive fri evening and (hopefully) depart sun afternoon.  But if we think we'll need time on Sun afternoon then perhaps I should move my flight to Monday.
18:44 pmichaud or maybe I can find a very late flight back.
18:44 pmichaud it's much easier to get late flights going east-to-west than vice-versa.
18:46 particle i can get 7p departure sunday
18:47 pmichaud yes, but you're flying north, not east :-P
18:47 particle we won't have the room after 5p, so there's no conference reason to stay
18:47 pmichaud but the question is really: do I need to plan to attend after 1pm sunday?
18:47 pmichaud if yes, then I either have to get a very late flight, or have another night stay
18:48 particle i think we can deal with that. can you put up a wiki page with travel constraints/questions?
18:48 particle it'll help in developing the schedule
18:48 particle but, really, it'll be a dozen people or so in one room
18:48 pmichaud well, let me see what the latest reasonable flight I can get on Sunday
18:48 particle so it shouldn't be hard to organize around travel
18:49 chromatic AHHH, filing new tickets doesn't help get our numbers down.
18:52 Tene Okay, work says they'll try to arrange for me to be home on an adjacent week, so the girl won't get upset, and if I'm traveling the adjacent week they'll get me a flight to or from there.
18:52 Tene chromatic: it raises the number of tickets but helps to lower the number of issues.
18:52 Tene The issues are there whether there's a ticket for them or not.
18:52 pmichaud I can get a 9:30p flight on Sunday, puts me back to DFW 5:05a on Monday :-P
18:53 Tene I still need to get another class in Dallas, so I can leave threatening notes and parrot feathers on pmichaud's doorstep.
18:53 chromatic Tene, ha ha only serious.
18:53 pmichaud Tene: you could send the via UPS or FedEx :-P
18:54 pmichaud s/the/them/
18:54 Tene chromatic: that ticket I filed about RPA::append, I can't find any notes left for me about how to reproduce or what I was using it for, and the commit you reference looks relevant, so I'd recommend closing that ticket.
18:55 Tene chromatic: I also recommend scolding me for not providing a way to reproduce.
18:55 * chromatic slaps Tene with a skyscraper
18:55 chromatic No PIR?  NO BUGFIX FOR YOU.
18:55 Tene Much better.  I can feel myself growing more responsible already.
18:56 chromatic Good, because lots of people fell out of the windows when I pulled the building out of the ground.
18:56 chromatic I do remain hopeful that some of the width and signedness warnings cleanup will accidentally fix some 64-bit issues.
18:57 Tene Oh?  Want me to check?
18:57 Tene I think they showed up with the jit core, right?
18:57 particle (tene's work)++
18:57 Tene OR, I could give you ssh to my laptop, if you think you can reproduce.
18:57 dalek r31979 | coke++ | trunk:
18:57 dalek : [codingstd] remove commented out code, resolve RT#43739
18:57 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31979
18:58 particle ...gorram 64bit mess...
19:01 NotFound The way to fix 64 bit issues is to have more people working in 64 bit machines. For example, send me a good one ;)
19:01 pmichaud particle:  I just edited the existing attendee survey page
19:01 chromatic I don't remember them being 64-bit specific.
19:01 * moritz has a 64bit machine with a 32 bit kernel on it
19:01 pmichaud I'll probably switch back to 64-bit here when the next release of kubuntu appears this month
19:01 moritz might be time to re-install from scratch
19:02 Infinoid I've got a debian/x86_64 box I can provide shell access on, should anyone need it
19:02 NotFound A system, I mean, not just a machine.
19:03 Infinoid I do most of my work on a gentoo/x86_64 laptop, but that one moves around a lot, so shell access is problematic.
19:04 NotFound Infinoid: It is actually showing some 64 specific ticket?
19:05 * Infinoid does a build on debian/x86_64 to find out
19:05 Coke purl, 685-46-633
19:05 purl 6
19:05 Infinoid NotFound: got a 64 specific ticket for me to check for?
19:06 NotFound Infinoid: don't remember any ATM. Some failing test?
19:07 Infinoid #59876 has been failing for me on gentoo, should be able to reproduce it on debian too
19:07 moritz Coke++ # closing tickets like hell
19:08 cosimo_ joined #parrot
19:10 Lorn joined #parrot
19:12 dalek r31980 | coke++ | trunk:
19:12 dalek : [mmd] add passing test from RT#41291  (multi dispatch on null)
19:12 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31980
19:13 Coke moritz: just the easy ones! =-)
19:15 Infinoid NotFound: yep, reproduced #59876
19:15 NotFound Infinoid: If I understand well, in that ticket what is failed is the test, not parrot.
19:15 Infinoid I bet it's an easy fix, I just haven't looked at the details.
19:16 NotFound I don't think that a word size of 8 in a 64 bit machine is a failure at all.
19:17 Infinoid agreed
19:17 NotFound What surprises me is that someone expect that wordsize 2 may work.
19:18 Infinoid I haven't tried ot use one of those architectures this century
19:18 Infinoid s/ot/to/
19:18 NotFound Unless the machine has CHAR_BIT > 8, but in that case a lot of things will fail.
19:19 Infinoid I know toshiba has made some 128-bit MIPS processors... haven't managed to get my hands on one yet.
19:20 dalek r31981 | Whiteknight++ | trunk:
19:20 dalek : [PMC] Update documentation for stmvar.pmc, as per RT#43261
19:20 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31981
19:23 dalek r31982 | julianalbo++ | trunk:
19:23 dalek : implement sysinfo CPU_ARCH, RT#59722
19:23 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31982
19:36 peepsalot joined #parrot
19:48 Andy joined #parrot
19:49 Coke HEY DID ANYONE ELSE GET THEIR GOOGLE T SHIRT!? huh!?
19:50 bacek joined #parrot
19:50 Tene Oh!  That must be what that is.
19:50 Tene I heard I got a squishy package from google, but I haven't gone to pick it up yet.
19:50 * moritz never got one
19:51 Tene I never participated in any gsoc-related stuff.  I feel like I must have cheated to get one.
19:51 chromatic It's a t-shirt.
19:51 chromatic Odd design, but black and free.
19:51 Tene Perhaps I'll just leave it in the box, so it remains mysterious.
19:51 pmichaud "I mentored GSOC and all I got was some broken code and this t-shirt."  :-P
19:52 * moritz mentored GSOC, got lots of good code but not t-shirt
19:52 moritz I think that combination is preferable ;)
19:52 NotFound moritz: Maybe the shirt is the consolation prize for the others.
19:53 moritz NotFound: heh, never thought of it that way ;)
19:55 Tene I could ship my shirt to moritz
19:55 Infinoid oh, speaking of GSOC, did we ever hear back from Senaka Fernando regarding harmony GC and parrot?
19:55 NotFound Tene: fax it.
19:55 Infinoid I answered a bunch of questions back in may-june and then heard nothing.
19:56 particle i don't remember anything from senaka after that
19:56 * Infinoid bugs him on yahoo
19:57 NotFound Did you think that if we call it 'Bit recicling' instead of 'Garbage collection' we've got more donations from enviromentally concerned organizations?
19:58 cjfields joined #parrot
20:00 cjfields Coke: got my shirt as well (mentored for NEScent)
20:01 cjfields ...goes to show anyone can get one  :)
20:01 TimToady phone
20:01 cjfields (well, as long as they do the work)
20:03 particle chromatic: are you joining in?
20:05 chromatic Yes.  Would you believe my broker called at 12:53 pm?
20:05 Coke ... except for the part where you have a broker.
20:21 cosimo_ left #parrot
20:28 gaz joined #parrot
20:34 ab5tract joined #parrot
20:40 dalek r31983 | particle++ | trunk:
20:40 dalek : [rakudo] minor action grammar refactorings
20:40 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31983
20:43 particle pmichaud: so, how do i know the name of the package(s) just used, so i can grab ::EXPORT::... ?
20:43 particle should i be introspecting the ast in use_statement?
20:45 particle probably need to look in $sub after it's compiled
20:59 Whiteknight joined #parrot
21:08 Zaba_ joined #parrot
21:16 PacoLinux joined #parrot
21:18 TiMBuS joined #parrot
21:28 Zaba joined #parrot
21:50 Zaba joined #parrot
22:10 donaldh joined #parrot
22:29 particle1 joined #parrot
22:57 Zaba_ joined #parrot
23:01 DietCoke joined #parrot
23:01 Whiteknight I've got two tests failing on my system, is that known?
23:02 Whiteknight I haven't run make test on trunk in a while, so I dont know
23:03 Whiteknight actually I've got four failing tests, but two of them are codingstd tests that I broke
23:04 Coke the smolder tests have been failing for days.
23:05 Whiteknight no, I've only got one failing test, pcre.t
23:06 Coke http://smolder.plusthree.com/app/public_p​rojects/report_details/6115#first_failure
23:06 tetragon joined #parrot
23:08 dalek r31984 | Whiteknight++ | trunk:
23:08 dalek : [Core] Test for NULL pointer dereferencing in src/stacks.c, as per RT#43851
23:08 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31984
23:09 dalek r31985 | Whiteknight++ | trunk:
23:09 dalek : [Core] codingstd fix on the stuff I just wrote.
23:09 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31985
23:10 dalek r31986 | Whiteknight++ | trunk:
23:10 dalek : [PMC] Codingstd trailing whitespace fix for some documentation I added earlier today.
23:10 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31986
23:11 ruoso joined #parrot
23:12 chromatic Coke, do you have GMP?
23:13 Coke maybe?
23:13 Coke (that's not my smolder report)
23:14 chromatic Okay.  Someone doesn't have GMP.
23:14 Coke ... are we requiring GMP?
23:14 * tetragon wonders if anyone has looked at RT#59874
23:15 * Coke wonders if there's some way to auto-URL that in this discussion.
23:15 chromatic tetragon, that's the one we're discussing.
23:16 cjfields joined #parrot
23:16 tetragon I'm preparing a fresh tree at the moment, but that build is still around
23:16 Coke sadly, I think that started failing after allison fixed pow for me. =-)
23:18 tetragon And no, I don't have GMP installed.  Even when I had my iBook (which had GMP), parrot couldn't find the thing as it was a framework build instead of a Unix build.
23:19 chromatic Okay, one moment please.
23:21 dalek r31987 | chromatic++ | trunk:
23:21 dalek : [t] Skipped arithmetic overflow test if Parrot didn't link against GMP (RT
23:21 dalek : #59874, Seneca Cunningham).
23:21 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31987
23:21 chromatic tetragon, would you mind upgrading to r31987 and running prove t/op/arithmetic.t?  No rebuild necessary.
23:22 tetragon The test skipped, and nothing in that file failed.  Does the check have any effects on 64-bit?
23:22 tetragon My 64-bit build passed the old test without any issues
23:23 chromatic 32-bit integers will overflow.  64-bit integers are fine.  I suppose the test is more of 32-bit overflow than anything else.
23:23 tetragon (although it failed the wordsize test in t/native_pbc/header.t (RT#59876))
23:24 tetragon So the test will be skipped even on my system where it would pass
23:25 chromatic For now.  If someone wants to modify it to check for 64-bitness, even better.
23:26 chromatic The conditional will be something like skip( ... ) if $PConfig{intvalsize} == 4 && !$PConfig{gmp};
23:26 dalek r31988 | Whiteknight++ | trunk:
23:26 dalek : [Core] Remove rotate_entries function from src/stacks.c, which isn't used anywhere. Doing this as per a suggestion in RT#43851
23:26 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31988
23:26 chromatic If you test that on 32- and 64-bit builds and it makes sense, let me know and I'll make that change.
23:27 Whiteknight two tickets today!
23:27 tetragon Skips on 32-bit
23:28 Coke anyone else have an opinion on closing out the "self contained languages" ticket?
23:28 tetragon Doesn't skip on 64-bit
23:28 Whiteknight Larry met Professor Aho? I'm so jealous!
23:28 chromatic Does it pass on 64-bit?
23:28 tetragon Yes
23:29 Coke of Professor Aho? eh. I've met larry. it's not that special.
23:30 chromatic still?
23:30 Coke chromatic: I wonder if #38432 is related to the exceptional issues we've been seeing elsewhere.
23:31 dalek r31989 | chromatic++ | trunk:
23:31 dalek : [t] Refined intval overflow test to skip only with 32-bit INTVALs without GMP,
23:31 dalek : per Seneca Cunningham.
23:31 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=31989
23:32 chromatic Coke, I wonder the same.
23:32 Tene Can anyone tell me how to build a jit-capable parrot?
23:33 chromatic perl Configure.pl --jitcapable
23:33 Tene chromatic++
23:41 Coke an intrepid bug-admin could probably combine several of the open storable tickets for parrot. I suspect some duplication.
23:41 Tene Hm.  I still get a segfault on 64-bit trying to use jit for perl6
23:41 Whiteknight Alfred Aho was one of the authors of AWK, co-author of the dragon book, and did a few other things
23:42 Coke ... it's a joke, son.
23:42 Coke aho?
23:42 purl i heard aho was the A in AWK?
23:42 * Coke pats purl onna head.
23:42 Tene Coke: how's that TGE test coming?
23:42 Coke ... did you see my email?
23:42 Whiteknight damn sons, oh no he di'n't
23:42 chromatic Bumping to Storable 2.18 seems the most reasonable solution to me.
23:42 Tene Coke: I didn't see any email from you
23:42 Coke it was attachd to the ticket.
23:42 Tene Coke: unless it was to a list...
23:43 chromatic They'll need a crane.
23:43 Tene I guess this would be easier if I had a short-term memory.
23:43 Tene There's a ticket?
23:45 Tene Ah, 59912
23:45 Coke http://rt.perl.org/rt3/Tic​ket/Display.html?id=59912
23:45 Coke ... curse my metal body, I wasn't fast enough.
23:45 Tene It's okay.  It's shiny enough.
23:46 Tene Was chromatic's patch there applied?
23:46 Coke not sfaik. I haven't tested it yet, either.
23:47 chromatic I haven't applied it.  Allison said she'd take a look.
23:51 Coke http://rt.perl.org/rt3/Tic​ket/Display.html?id=45965 ... I could probably whip up a patch that removed the leading __ from the slot names. Any interest ?
23:53 chromatic Yes please.
23:56 nopaste "tene" at 67.137.148.11 pasted "this might be another part of the fix needed for TGE (Coke++)" (19 lines) at http://nopaste.snit.ch/14300
23:57 Tene Hm, I'm not sure if that shoudl be new or if that should be a get_class/new pair
23:57 Tene no time to test right now, need to run home.
23:58 chromatic Tene, that seems pretty sane.  Mostly it needs to mirror what parsegrammar() does.

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

Parrot | source cross referenced