Camelia, the Perl 6 bug

IRC log for #parrot, 2008-11-30

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:02 Whiteknight joined #parrot
00:06 dalek r33356 | kjs++ | trunk:
00:06 dalek : [imcc] remove ^#line syntax. Use .line instead. I think setline and setfile are handled in the lexer, not ops?! Needs more cleanup. For now #line can't be used anymore, which is good.
00:06 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33356
00:09 bacek joined #parrot
00:09 AndyA joined #parrot
00:13 Limbic_Region joined #parrot
00:36 dalek r33357 | Whiteknight++ | call_conv_redux:
00:36 dalek : Creating second branch to help with debugging
00:36 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33357
01:10 Whiteknight I just got failures in t/compilers/tge/grammar.t in a fresh checkout
01:10 Whiteknight is this a known issue?
01:11 kid51 I see that r33355, a little over an hour ago, is tagged [tge]
01:12 kid51 diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33355
01:15 kid51 That test passed for me at r33354 on both Linux and Darwin.
01:16 Whiteknight as of 33356, it fails for me on Linux
01:20 kid51 Recommend contacting kjs re 33355 or 33356
01:20 kid51 purl seen kjs
01:20 purl kjs was last seen on #perl 159 days, 9 hours, 47 minutes and 29 seconds ago, saying: yo  [Jun 23 15:33:30 2008]
01:21 kid51 Better file RT
01:21 jimmy joined #parrot
01:21 cotto joined #parrot
01:27 jimmy Will docementation support multi-language officially on both parrot and perl6? i.e. php.net ,cakephp.org
01:28 jimmy err. it's documentation
01:31 jimmy Will docementation support multi-language officially on both parrot and perl6? i.e. php.net ,cakephp.org
01:31 jimmy Will docementation support multi-language officially on both parrot and perl6? i.e. php.net ,cakephp.org
01:32 jimmy sorry, i'am cleaning the keyborad
01:35 masak 你好, jimmy.
01:36 jimmy 好啊,masak
01:36 masak :)
01:36 jimmy 今天是星期天,没想到你会在这里
01:37 masak jimmy: I haven't gone to bed yet.
01:37 masak too much fun programming Perl 6.
01:38 jimmy ah, when here was morning and there was night.
01:39 masak 对.
01:40 jimmy i want parrot and perl5 can officially support multi-language.
01:43 cotto jimmy, I'll mention that at the next #parrotsketch, if nobody else does.
01:43 cotto botsnack
01:43 purl :)
01:44 jimmy thanks cotto. it's important.
01:44 jimmy sorry, not perl5, should be perl6
01:44 cotto I agree.  Thanks for mentioning it.
01:45 cotto seen pmichaud
01:45 purl pmichaud was last seen on #parrot 1 hours, 51 minutes and 17 seconds ago, saying: (and I wrote them in PIR, so not easily "todo")
01:45 jimmy seen jimmy
01:45 purl jimmy was last seen on #parrot 0 seconds ago, saying: seen jimmy
01:45 cotto pmichaud, ping
01:45 cotto jimmy++ #documentation i18n
01:47 jimmy My experience is that perl is not widely used for  lack of official multi-language
01:50 cotto thanksgiving++
01:53 chromatic That was a 56k leak in Rakudo.
01:53 dalek r33358 | chromatic++ | trunk:
01:53 dalek : [src] Fixed a memory leak where the memory allocated for dynop libraries never
01:53 dalek : had a corresponding free.  Now it gets freed only in the parentmost interpreter
01:53 dalek : during global destruction.
01:53 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33358
02:02 chromatic Now Rakudo's "Hello, world!" only leaks 100k.
02:02 dalek r33359 | chromatic++ | trunk:
02:02 dalek : [src] Fixed another dynops memory leak, this time for the dynops func table.
02:02 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33359
02:03 masak chromatic++
02:07 chromatic That leaves 3 other reachable memory allocations I don't have the energy to explore and ~230 context-related leaks I don't have the heart to read.
02:07 bacek joined #parrot
02:07 contingencyplan joined #parrot
02:10 tewk_away pmichaud: can you nopaste the tests?
02:59 tak joined #parrot
03:08 apeiron joined #parrot
03:33 cotto coverity?
03:33 purl coverity is a commercial tool for Automated Error Prevention and Source Code analysis, See,  http://www.coverity.com/main.html or it has been used to measure the quality of the LAMP stack and other major source projects
03:37 Zaba joined #parrot
03:43 pmichaud tewk: http://www.pmichaud.com/perl6/subflags.t is what I have so far.
03:43 pmichaud tewk: if you're working in a branch we could check it in there
03:44 pmichaud or, as soon as :nsentry is parseable, we can check it in then and I can do skips.
03:45 Hinrik joined #parrot
03:51 cotto pmichaud, is the code from your lolcode demo available (esp the past->lolcode translator)?
03:52 pmichaud not yet -- have to refactor PCT::HLLCompiler first
03:52 pmichaud it doesn't work without some patches to PCT::HLLCompiler and NQP
03:52 pmichaud (but after the refactor it'll work just fine)
03:53 cotto so you're planning on releasing it eventually
03:53 pmichaud oh, definitely.
03:53 cotto great
03:53 pmichaud it's one of the reasons I want to get HLLCompiler refactored soon.
03:53 cotto i.e. before 1.0
03:53 pmichaud i.e., before December release if I can help it.
03:54 pmichaud gotta run -- bbl
03:54 cotto bye
03:56 Coke whee, segfault.
03:56 purl segfault is http://xkcd.com/371/
04:02 elmex_ joined #parrot
04:02 * Coke wonders if feather is wicked slow or if it's just coke's net cx.
04:11 Zaba_ joined #parrot
04:14 * Coke hates chasing segfaults. :|
04:16 chromatic Try chasing segfaults on a platform you don't run.
04:17 davidfetter joined #parrot
04:18 * Coke tries going about this with a bisect instead of reducing the sample segfault code.
04:21 Coke Infinoid++
04:21 Coke EROODE--
04:33 Coke who is azawawi?
04:35 davidfetter cousin of bawitdaba iirc
04:36 Coke sounds like a kid rock song.
04:37 Coke (he's the most likely reason feather's a bit slow. just curious.)
04:44 Infinoid what did I do?
04:49 Zaba joined #parrot
04:50 Coke svn-bisect
04:51 Coke I'm in the middle of rewriting my tool, but am blocked by a bug in Iterator.pm
04:51 Coke so I'm using yours in the meantime.
04:52 Infinoid oh, so you're the one who downloaded it.  thanks :)
04:52 Coke you can't fool me, no one knows who downloads things off the cpan. :|
04:55 Zaba_ joined #parrot
05:03 Zaba joined #parrot
05:09 Zaba_ joined #parrot
05:16 * Coke tries out the force unleashed demo
05:19 bacek joined #parrot
05:21 Zaba joined #parrot
05:23 jimmy_ joined #parrot
05:25 Coke ah, it's tewk's fault.
05:26 tewk Coke: what broke?
05:26 tewk all tests passed on linux
05:29 Coke it causes a segfault when running tcl.pbc's library/tcltest/tcltest.tcl
05:29 Coke opening a ticket now.
05:29 Coke (will include backtrace)
05:29 Coke (r33351)
05:30 Coke (this is on feather, a linux box)
05:30 MariachiElf joined #parrot
05:32 Zaba_ joined #parrot
05:32 cotto (why are you whispering?)
05:35 Coke trac 10
05:36 * Coke hates treading water.
05:37 tewk I'll look right now.
05:40 tewk Coke do you have your own sub PMC?
05:40 tewk :q
05:41 bacek joined #parrot
05:42 Coke tewk: yup.
05:42 Coke I subclass .Sub in PIR
05:42 Coke languages/tcl/src/class/tclproc.pir
05:43 tewk The layout of struct Parrot_sub has changed, I don't know if that effects you, I'll look
05:45 tewk do you have any checked in pbc that isn't being rebuilt?
05:45 tewk The Parrot_sub in your segfault looks corrupt.
05:46 Coke I was doing a realclean in parrot -and- tcl before each build.
05:47 Coke tcl doesn't check in .pbc files.
05:47 Coke (atm)
05:49 Coke Are you relying on the struct existing?
05:49 tewk The backtrace looks like three deep nested exceptions being thrown.  fail_if_type_exists
05:50 Coke (just wondering if you're deailing with PIR subclasses)
05:53 tewk does this run the whole test suite.
06:01 tewk Sorry coke, but -G just finished successfully for me :(
06:13 bacek joined #parrot
06:24 Coke does what run the whole test suite?
06:24 Coke no. it just loads the library file and stops.
06:24 tewk Yeah I figured that out.
06:24 Coke (works with -G). Doesn't surprise me.
06:24 tewk ./parrot -G finished successfully for me, does it work for you?
06:25 chromatic Sounds like a PMC gets reclaimed early.
06:25 chromatic Or a Tcl Sub isn't created as a constant PMC.
06:25 Coke I'm in the middle of testing something else now, can't double check.
06:25 Coke no, the tcl subs aren't constant.
06:27 chromatic If they're not constant, you have to be very sure that they get marked.  (Being in a NameSpace doesn't guarantee that.)
06:27 Hadi joined #parrot
06:27 Hadi left #parrot
06:28 Coke ... it doesn't?
06:29 chromatic Not if the NameSpace is constant.
06:29 Coke and how would one make a NS constant?
06:29 Coke (I presume they are not, by default.)
06:30 chromatic PMC *ns = constant_pmc_new(interp, enum_class_NameSpace);
06:31 Coke most of my NS manipulation is from PIR.
06:31 dalek r33360 | tewk++ | trunk:
06:31 dalek : [nsentry] ns_entry -> nsentry tokenizer change
06:31 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33360
06:31 Coke I hesitate to say all, but I think it may be that.
06:35 * Coke tries to rip out src/*pic* and sees that it's used. Coke reverts.
06:37 chromatic Yeah, all of the computed cores use parts of it.
06:39 Coke If we have a "fulltest" target that now DTRT, can we add an nqp-test component to "make test" ?
06:41 Coke tewk: with -G, I get a panic, out of memory.
06:41 chromatic It's debatable that fulltest DTRT, but I was thinking about that this morning and I'm not sure that it can get any better than it gets right now, not without rewriting chunks of TAP::Harness.
06:41 chromatic ... but adding NQP would be good.
06:42 Coke chromatic: this might be a simpler segfault: http://rt.perl.org/rt3/Tic​ket/Display.html?id=59968
06:42 Coke I can duplicate that on feather (first changing P0 to $P0)
06:44 jimmy http://www.parrotvm.org/svn​/parrot/revision?rev=33360 has nothing?
06:44 tewk I've got 4gig, it finishes here.
06:44 chromatic Coke, I believe that feather is haunted.
06:44 chromatic jimmy, the SVN::Web server is a little slow.  The commit message showed up in my email.
06:45 jimmy chromatic: no, not slow.
06:46 jimmy it displayed no changes
06:46 tewk purl: message pmichaud nsentry should parse now.
06:46 purl Message for pmichaud stored.
06:48 jimmy chromatic: it seems that SNV::Web does not work normally
06:48 * tewk sleeps
06:49 chromatic Coke, I thought it was a missing .get_results in the exception handler, but that's not it.
06:50 Topic for #parrotis now Parrot 0.8.1 "Tio Richie" Released | http://parrot.org | 594 RT
07:14 Zaba joined #parrot
07:18 iblechbot joined #parrot
07:34 jimmy rakudo: <Hello>.say
07:34 polyglotbot No output (you need to produce output to STDOUT)
07:36 jimmy_ joined #parrot
07:38 Theory joined #parrot
07:41 bacek joined #parrot
07:42 jimmy___ joined #parrot
07:58 apeiron_ joined #parrot
08:03 Hadi joined #parrot
08:03 Hadi left #parrot
08:17 bacek joined #parrot
08:21 Alias joined #parrot
08:39 Theory joined #parrot
08:45 Zaba joined #parrot
09:04 nadim__ left #parrot
09:14 Zaba joined #parrot
09:16 iblechbot joined #parrot
09:35 Ademan joined #parrot
09:41 dalek r33361 | simon++ | trunk:
09:41 dalek : Function signatures used by SQLite
09:41 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33361
09:52 lathos What's Perl 6 for $classname->method() ? (Holding the name of a class in a string and calling a class method?)
10:04 jimmy joined #parrot
10:04 jimmy l
10:19 MariachiElf joined #parrot
10:37 alvar joined #parrot
11:16 TimToady lathos: any identifier component of a name may be symbolically replaced by ::($name), which is the explicit symbolic reference syntax in p6
11:17 TimToady so, ::($classname).method should work for that
11:38 jimmy_ joined #parrot
11:39 Zaba joined #parrot
11:53 lathos TimToady: Lovely, thanks. (Assuming Rakudo supports it.) I'm working with Tim B on DBDI
11:55 Alias TimToady: Quick question, does your @cpan.org account work?
12:13 Zaba joined #parrot
12:20 Zaba_ joined #parrot
12:33 rob joined #parrot
13:05 kj joined #parrot
13:40 kid51 joined #parrot
13:52 Hadi joined #parrot
13:54 Hadi left #parrot
14:09 ruoso joined #parrot
14:12 bacek joined #parrot
14:13 Whiteknight joined #parrot
14:20 dalek r33362 | kjs++ | trunk:
14:20 dalek : [pdd] remove the part about resuming an exception from pdd19: it's not a part of the PIR language. Also, it was also written in better detail in pdd23. Add a section 'Exception Object Interface' to pdd23, of which resuming is one subsection. Retrieving the message is another section.
14:20 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33362
14:21 dalek r33363 | kjs++ | trunk:
14:21 dalek : [pdd23] woops. A head2 should have been a head3.
14:21 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33363
14:24 Theory joined #parrot
14:24 lathos leh.
14:24 lathos % ../../perl6 -e 'class A { method b () { say "I was called" } }; our $x = "A"; ::($x).b()'
14:24 lathos Method 'b' not found for invocant of class 'Failure'
14:24 lathos Puts paid to that for the time being, then.
14:33 dalek r33364 | kjs++ | trunk:
14:33 dalek : [lib] instead of using the 'global' keyword, use the 'find_global' op. Yes, that's deprecated as well, but this is the first step in removing the 'global' keyword from IMCC. The whole find_global/store_global deprecation is in fact a separate issue, and a larger task.
14:33 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33364
14:34 Zaba joined #parrot
14:36 dalek r33365 | kjs++ | trunk:
14:36 dalek : [t] more 'global' -> 'find_global' conversions.
14:36 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33365
14:37 gmansi joined #parrot
14:41 dalek r33366 | kjs++ | trunk:
14:41 dalek : [t] {find_global,global} becomes get_global.
14:41 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33366
14:44 Hadi1 joined #parrot
14:44 Hadi1 left #parrot
14:48 Whiteknight Is anybody else getting TGE test failures?
14:49 kj Whiteknight: yes, me too
14:49 kj not sure why that is.
14:50 Whiteknight okay, I'm just making sure that I'm not crazy
14:50 Whiteknight plus I've got other work to do, and don't want to stall out on test failures if they are known failures that I didn't cause
14:50 kj I've been working on imcc yesterday, but I don't think that's the cause; ran make test, and that was fine.
14:50 lathos make distclean and rebuild worked for me.
14:51 kj is distclean different from realclean?
14:51 kj ehm, i'm sure it is :-) I meant to ask in what way..
14:51 Whiteknight i've never even heard of make distclean
14:53 masak joined #parrot
14:54 kj Whiteknight: what's your error message, somthing about a attribute 'gmin'?
14:55 Whiteknight i dont know,
14:55 Whiteknight i have to look at it closer
14:56 Whiteknight yeah, "Cannot find attribute gmin"
14:56 kj yeah me too. The funny thing is, there's no mentioning of 'gmin' in the source
14:56 kj so there's something messed up somewhere
14:57 Whiteknight good
14:57 kj oh wait
14:57 Whiteknight (something messed up somewhere)--
14:57 kj there is.. I was looking at something else.
14:57 kj doing fresh co and make test...
14:58 * kj can't wait for 256 processor systems...
15:00 jimmy does lexical var mean part var ?
15:00 Alias Erlang founder reported his projections for 10-years-from-now server gear
15:00 Alias 1Ghtz x 1000 cores wide
15:01 kj that would be nice
15:01 jimmy I want to translate it to chinese.
15:01 jimmy I think it is the same meaning
15:01 kj jimmy: what do you mean?
15:02 jimmy lexical meaning using another word
15:02 kj locally scoped?
15:03 jimmy kj: is it the same meaning?
15:03 jimmy for lexical?
15:04 apeiron joined #parrot
15:04 kj ehm. well, in my understanding, yes.
15:04 jimmy I am translating some docs to chinese.
15:04 jimmy thanks kj
15:05 kj you're welcome.
15:06 kj jimmy: have a look at http://www.perlmonks.org/?node_id=95813
15:07 jimmy perl is especial
15:08 kj http://gigamonkeys.com/book/variables.html
15:08 kj common lisp :-)
15:08 kj I think the name 'lexical' comes from the fact that the scope is limited which is easy to see
15:09 kj for instance, within a { and } or do and end keywords (in e.g. Lua, etc.)
15:09 kj it's the opposite of dynamic scope
15:10 kj but I've never quite understood that...
15:10 jimmy hmm, the problem is not here.
15:11 kj lathos: just did a fresh co, tge is still failing for me
15:11 kj (win32)
15:12 pmichaud kj:  tge is failing build or failing test?
15:12 jimmy it is difficult to select a native word to conform to native people.
15:12 kj pmichaud: oh sorry, failing test
15:13 pmichaud lathos:  I'm reworking lexical handling in rakudo a bit -- when that's done I'll get interpolated namespaces   ::($x)   to work.
15:14 masak kj: HOP contains an excellent introduction to the difference between lexical and dynamic variables. http://hop.perl.plover.com/
15:14 kj masak: thnx, will have a look
15:15 pmichaud masak: I almost have    .subst( $regex, $closure, :g )    working.  Just have to fix up lexicals a bit.
15:15 masak woot
15:18 * masak goes to track down the test failures in S29-num/abs.t
15:21 Whiteknight pmichaud+
15:21 Whiteknight or pmichaud++
15:24 tetragon joined #parrot
15:26 dalek r33367 | Whiteknight++ | call_conv_redux:
15:26 dalek : [call_conv_redux] Update Parrot_build_sig_object_from_varargs
15:26 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33367
15:31 kj pmichaud: I'm looking into RT#48705; about doc. of 'chain' pasttype. Would it suffice to copy/paste the description from compilers/pct/past/compiler.pir into pdd26?
15:35 dalek r33368 | Whiteknight++ | call_conv_redux:
15:35 dalek : [call_conv_redux] Rename Parrot_pcc_invoke_sub_from_sig_object to Parrot_pcc_invoke_from_sig_object
15:35 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33368
15:47 dalek r33369 | Whiteknight++ | call_conv_redux:
15:47 dalek : [call_conv_redux] update commit_last_arg_sig_object and introduce the PARROT_ARG_INVOCANT flag
15:47 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33369
15:57 pmichaud kj: sure.
15:57 pmichaud (rt #48705)
15:58 pmichaud fwiw, I get the tge/grammar.t failure also on my system.
15:58 pmichaud I'll try a fresh co in a second.
15:59 kj pmichaud: I tried a fresh co already. It seems there's a problem with imcc.
15:59 dalek r33370 | pmichaud++ | trunk:
15:59 dalek : [rakudo]: spectest-progress.csv update: 227 files, 4653 passing, 48 failing
15:59 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33370
16:00 kj I'll try to check with an older version
16:01 kj pmichaud: (rt #48705) also adding pasttype('return'): can a PAST::Op( :pasttype('return') ) take any number of children, meaning all of them are returned?
16:03 pmichaud kj: no.
16:03 pmichaud at least, I don't think that's the case.
16:03 pmichaud eventually it will probably be able to do that.
16:03 kj pmichaud: ok, but for now only a single return value?
16:04 dalek r33371 | Whiteknight++ | call_conv_redux:
16:04 dalek : [call_conv_redux] update set_context_sig_params
16:04 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33371
16:04 pmichaud currently only a single return value.
16:04 kj ok,thanks
16:04 pmichaud we'll wait for the calling conventions branch to be updated before we can update PAST to match.
16:06 Whiteknight I'm working on that branch at full-speed today
16:06 Whiteknight hopefully should't be too too long now
16:06 tewk pmichaud: don't know if you got the message from purl, but the lexer was looking for ns_entry, which I fixed.
16:07 pmichaud I got the message from purl, yes.  underscores are evil.
16:08 tewk it was a left over that I missed. fixed now.
16:08 dalek r33372 | kjs++ | trunk:
16:08 dalek : [pdd26] add pasttype descriptions for 'chain' and 'return'.
16:08 dalek : resolving RT#48705.
16:08 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33372
16:15 nopaste "pmichaud" at 72.181.176.220 pasted "issame fails for String PMCs (?!)" (15 lines) at http://nopaste.snit.ch/14754
16:15 pmichaud every so often I run into a Parrot bug that totally confounds me.
16:21 kj pmichaud: issame wants to get something from $P1, but it's null.
16:21 kj is that a bug?
16:21 pmichaud kj:  issame is supposed to simply determine if two PMC registers are the same PMC.
16:22 pmichaud it just compares addresses (at least according to the docs)
16:22 kj but $P1 is null..
16:22 pmichaud so obviously they have a different address
16:22 kj well, "no address" (or no valid address)
16:22 kj it's a question whether issame should allow null operands
16:23 kj I'd say a not-null check shouldn't be part of issame; it would slow down the op; if you want that behavior, add an 'if' instruction :-)
16:24 pmichaud I think that "issame" should simply tell me if two pmc registers are in fact the same PMC.
16:24 masak the reverse.t and abs.t regression occurred in r33263. investigating further.
16:24 pmichaud masak:  I think the regression is in core parrot.
16:24 masak it is.
16:24 pmichaud masak: it has to do with multidispatch changes
16:24 masak yes.
16:25 kj yeah, I seeyour point. But it's important to decide whether ops should accept null operands
16:25 pmichaud some ops do;   for example,   isnull does
16:25 pmichaud in fact, issame accepts null operands (see nopaste coming up)
16:25 kj because otherwise it wouldn't make sense :-)
16:25 nopaste "pmichaud" at 72.181.176.220 pasted "issame accepts null operands" (15 lines) at http://nopaste.snit.ch/14755
16:26 pmichaud it just doesn't accept a null operand for the String type.
16:26 kj mm. weird.
16:26 kj ah I see
16:26 jan joined #parrot
16:27 kj if the operands are in fact the same (both null), then that works out fine
16:27 pmichaud but that doesn't explain 'get_global main'
16:27 kj but otherwise, if they're not equal, both are passed to VTABLE_is_same
16:27 pmichaud ....why is "issame" a VTABLE function?!?
16:27 pmichaud that seems totally bizarre to me.
16:27 kj you're right.
16:27 pmichaud I guess because we want PMC registers to be able to lie about their identity...?
16:28 tewk or does issame mean isequal here?
16:28 pmichaud the docs are pretty clear that "issame" identifies PMC identity by address.
16:29 kj it's an == comparison
16:29 kj in src/ops/cmp.ops
16:29 pmichaud =item B<issame>(out INT, invar PMC, invar PMC)
16:29 pmichaud =item B<issame>(out INT, in STR, in STR)
16:29 pmichaud Sets $1 to 1 or 0, depending on the identity of the 2 objects.
16:29 pmichaud The identity is the PMCs memory address.
16:29 kj if that's false, it falls back to vtable_is_same
16:30 tewk :q
16:30 pmichaud we don't need issame to do a == comparison -- we already have  C<iseq>  C<cmp>,  C<cmp_num>,   C<cmp_str>,   etc.
16:30 kj i'd say it's an optimization?
16:30 pmichaud optimizing... what?
16:30 kj you don't need to call vtable_is_same, if it's in fact the same object
16:31 pmichaud I'm saying that is_same should check object identity, and that's it.
16:31 pmichaud checking object value is the purpose of other opcodes (and we have plenty of them)
16:31 pmichaud if those other opcodes want to optimize by checking identity, great.
16:31 tewk I agree
16:32 tewk grepping to see who uses issame
16:36 tewk grep is_same *.pmc
16:36 tewk grep is_same src/pmc/*.pmc
16:36 dalek r33373 | Whiteknight++ | call_conv_redux:
16:36 dalek : [call_conv_redux] update count_signature_elements
16:36 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33373
16:37 masak pmichaud, chromatic: commenting out line 265 (added on r33263, then as line 260) of src/pmc/namespace.pmc makes reverse.t and abc.t pass again. I don't know enough to fix it, but that line's the cause.
16:40 pmichaud the && !old line?
16:41 masak aye.
16:41 masak that was one of the two changes in r33263.
16:42 pmichaud well, I don't understand enough about namespace/multisub guts to know why that's there... so I'll leave it for chromatic++ or jonathan++ to work out.
16:42 pmichaud there's a ticket for it
16:42 pmichaud so you could comment there
16:42 masak what's the ticket number?
16:44 tak joined #parrot
16:55 * Coke wonders what Alias's secret is.
16:56 Alias :)
16:56 Alias I've lost track of who knows and who doesn't
16:59 Coke Oh, well you've already told me, but I need you to privmsg me so I don't... forget.
16:59 * Coke pokes Alias.
16:59 Alias heh
16:59 Alias I'm going to shut up from now until release, just in case something goes wrong
17:00 Coke hurm. can I play the grant committee card? =-)
17:01 Alias The reason I dropped the teaser is SPECIFICALLY because SOMEONE from the grant committee tried to publically embarrass me into submitting my report :)
17:01 Coke oh, that's probably ambs. I'm the -good- cop!
17:01 Alias Consider this payback and go beat them up
17:01 * sjn wonders if the sekrit is what he thinks it is
17:01 Alias sjn: Quite possibly
17:02 sjn it is a very, very, very cool sekrit :)
17:02 Alias In fact, almost certainly yes
17:02 Alias heh
17:02 Alias YOU SHALL NOT PASS (what it is on to Coke or anyone else) :)
17:04 sjn not ok # SKIP no annoucement until osdc.au
17:04 sjn bah, that came out weird
17:05 Alias Worked for me
17:05 sjn no test comment?
17:06 sjn test name
17:06 Alias I got the point :)
17:06 sjn anyhoo
17:06 Alias Coke: And I couldn't have done it without Europe
17:06 Alias (Although perhaps not in the way you think) :)
17:08 Alias sjn: I think he's sulking :)
17:08 * sjn wonders if anyone can help me make a list of rakudo/parrot devs who would be interested in coming to Oslo in April
17:09 Coke grab a page on the wiki and send out an email to the parrot-dev list?
17:10 * sjn has to do something a bit more sneaky before this :)
17:10 Coke ->
17:11 * sjn has already asked jonathan, and he was interested then
17:30 ambs joined #parrot
17:32 ambs left #parrot
17:36 ruoso joined #parrot
17:40 register joined #parrot
17:45 kjs_ joined #parrot
17:53 dalek r33374 | Whiteknight++ | call_conv_redux:
17:53 dalek : [call_conv_redux] update Parrot_mmd_build_type_tuple_from_sig_obj
17:53 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33374
18:01 Limbic_Region joined #parrot
18:20 masak pmichaud: using curlies is fine -- that TODO ticket is mainly of a code-aesthetic nature, and because I like to turn not-yet-implemented features into TODO tickets :)
18:20 pmichaud masak: yes, I'm in full agreement.
18:21 pmichaud I like having the todo tickets; when there's a reason that they're not done or a blocker than I convert them to "stalled" with an explanation.
18:21 masak aye, that's good.
18:21 masak & # nom
18:22 register is there a search engine for the irc logs?
18:23 pmichaud http://irclog.perlgeek.de/search.pl?channel=parrot
18:23 register thx
18:27 Aisling joined #parrot
18:28 Aisling joined #parrot
18:35 masak -gq tests rock!
18:35 masak :/
18:55 pmichaud purl: message tewk  the version of http://www.pmichaud.com/perl6/subflags.t doesn't produce _any_ output for me (kinda bizarre).
18:55 purl Message for tewk stored.
19:15 moritz re
19:28 allison joined #parrot
19:30 dalek r33375 | Whiteknight++ | call_conv_redux:
19:30 dalek : [call_conv_redux] update Parrot_pcc_invoke_from_sig_object
19:30 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33375
19:32 register what is the runloop level??
19:33 Whiteknight register, I think it has to do with the level of recursion
19:33 Whiteknight but I may be thinking about the wrong thing
19:33 register it makes sense
19:34 moritz I get a failure in t/compilers/tge/grammar.t, is that known?
19:35 register mmm
19:36 kjs_ moritz: could you try svn up?
19:36 dalek r33376 | kjs++ | trunk:
19:36 dalek : [tge] undo the change from #line into .line. This made a tge test fail.
19:36 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33376
19:36 kjs_ and try again?
19:37 moritz kjs_: passes agian, thanks
19:37 kjs_ well it was my fault in the first place :-( Don't see why it didnt' work though
19:49 donaldh joined #parrot
19:52 Whiteknight moritz, I've been getting the same failure all morning
19:52 Whiteknight oh, is it resolved now?
19:52 kjs_ Whiteknight: yes, it's fixed
19:52 Whiteknight kjs++
19:52 kjs_ r33376
20:26 MariachiElf joined #parrot
20:35 davidfetter joined #parrot
20:52 bacek joined #parrot
20:52 bacek good morning from summer, people of winter
20:58 masak joined #parrot
20:59 * davidfetter just recently got to summer
20:59 davidfetter bacek, where are you?
20:59 bacek davidfetter: 30 km from Sydney CBD
20:59 tak joined #parrot
20:59 * davidfetter in sydney, near the osdc venue :)
21:00 bacek davidfetter: I'll try to go to OSDC...
21:00 davidfetter groovy
21:00 tetragon I would throw a snowball at all of you summer people, but unlike last week, there's no snow now
21:00 davidfetter that's ok, i go back to winter on the 9th
21:00 davidfetter fsvo "winter" ;)
21:01 * davidfetter lives in oakland, california
21:01 moritz it's -2°C here :)
21:02 cotto good time to make ice cubes
21:02 tetragon 2 here (and they claim light snow, but not where I'm sitting)
21:02 * bacek thinks that negative temperature should be banned
21:02 moritz bacek: just calculate in Kelvin, then ;)
21:03 bacek moritz: +300 Kelvin is good one
21:03 tetragon (It was -10 a few days ago)
21:03 * tetragon grumbles about the variance on autumn temperatures
21:04 moritz actuallly even Kelvin is a dumb, made-up temperature scale. There's one natural scalliing falling out of thermodynamics, but nobody uses it.
21:05 bacek interesting... TimToady is key-note speaker on Day 2. And there is no Perl on this day at all...
21:06 davidfetter moritz, true
21:06 * davidfetter thinks centigrade is at least as provincial as fahrenheit, and less useful
21:16 chromatic joined #parrot
21:26 dalek r33377 | fperrad++ | trunk:
21:26 dalek : [Lua]
21:26 dalek : - clean up clone()
21:26 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33377
21:27 dalek r33378 | fperrad++ | trunk:
21:27 dalek : [Lua]
21:27 dalek : - unTODO 2 tests
21:27 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33378
21:29 dalek r33379 | fperrad++ | trunk:
21:29 dalek : [Lua]
21:29 dalek : - move frexp, ldexp & modf to LuaNumber PMC
21:29 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33379
21:31 dalek r33380 | fperrad++ | trunk:
21:31 dalek : [Lua] mathx
21:31 dalek : - add infinity & nan
21:31 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33380
21:32 cotto In PGE, is there a way to specify a nested namespace as an attribute?
21:33 pmichaud ...as an attribute?
21:35 pmichaud afk # grocery shopping
21:35 cotto as in PAST::Block.new( :namespace('foo') )
21:37 pmichaud you mean in pct?
21:37 pmichaud PAST::Block.new( :namespace('foo','bar','baz') )    # works, I think
21:37 pmichaud :namespace will take an array.
21:38 pmichaud maybe  :namespace(('foo','bar','baz')) is syntactically cleaner.
21:38 register joined #parrot
21:38 dalek r33381 | fperrad++ | trunk:
21:38 dalek : [Lua] mathx
21:38 dalek : skip infinity & nan on win32
21:38 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33381
21:39 pmichaud in rakudo, we do the equivalent of:    my @ns := Perl6::Compiler.parse_name('Foo::Bar::Baz');
21:39 pmichaud and then use  :namespace(@ns)
21:39 pmichaud afk # grocery shopping
21:40 cotto pmichaud++
21:45 kid51 joined #parrot
22:07 alvar joined #parrot
22:12 alvar joined #parrot
22:26 dalek r33382 | Whiteknight++ | call_conv_redux:
22:26 dalek : [call_conv_redux] removing some unneeded cruft to get a better idea of what else has changed
22:26 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33382
22:27 rob joined #parrot
22:31 TiMBuS joined #parrot
22:35 Hadi joined #parrot
22:37 Hadi left #parrot
22:50 Infinoid can hash keys be consted?  Are they ever edited in place, or just added/removed?
22:50 chromatic Which type of hash?
22:51 Infinoid I'm looking at src/hash.c, there is an IMCC warning calling parrot_hash_put because the "key" argument isn't const.
22:51 Infinoid (I know, really low on the priority scale...)
22:51 Limbic_Region joined #parrot
22:53 dalek r33383 | allison++ | pdd22io_part2:
22:53 dalek : [pdd22io] A buffer flush is a full flush.
22:53 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33383
22:58 dalek r33384 | infinoid++ | trunk:
22:58 dalek : [CAGE] Fix a warning in src/nci_test.c, by adding "const".
22:58 dalek : src/nci_test.c: In function 'nci_vV':
22:58 dalek : src/nci_test.c:1170: warning: assignment discards qualifiers from pointer target type
22:58 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33384
23:07 masak chromatic: did you see my message about the line in src/pmc/namespace.pmc?
23:13 dalek r33385 | Whiteknight++ | call_conv_redux:
23:13 dalek : [call_conv_redux] deleting this branch, it didn't help me solve any problems
23:13 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33385
23:20 chromatic masak, I did.  I'm looking at it now.
23:20 masak woot
23:20 chromatic Infinoid, I think they can be const.
23:23 * Infinoid tries it to see how much stuff breaks.
23:31 bacek joined #parrot
23:34 dalek r33386 | chromatic++ | trunk:
23:34 dalek : [PMC] Reverted the check for an existing PMC in a NameSpace when adding a
23:34 dalek : MultiSub to a NameSpace with set_pmc_keyed_str.  I added this in r33263, but it
23:34 dalek : broke Rakudo.  See RT #60910, filed by Patrick Michaud.
23:34 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=33386
23:34 masak chromatic++
23:58 Khisanth joined #parrot

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

Parrot | source cross referenced