Camelia, the Perl 6 bug

IRC log for #parrot, 2009-01-12

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:04 Whiteknight no, docs/book/ is not out of date
00:04 Whiteknight at least, it shouldn't be
00:04 Whiteknight I work on it almost every day
00:04 Whiteknight although I could have missed certain changes
00:09 AndyA joined #parrot
00:10 riffraff Whiteknight, then I guess I shall report references to P2 being referred as the "self" register?
00:10 riffraff mh, ugly alliteration
00:11 Whiteknight it happens
00:11 Whiteknight When I started workig on the book it was very very out of date. there are lots of problems I've missed
00:12 riffraff of course
00:14 riffraff I did not mean to sound complaining, I just like to report stuff when I know it has a chance to be fixed :)
00:14 dalek r35428 | kjs++ | trunk/compilers/pirc/src (2 files):
00:14 dalek : [pirc] clean up some bits at the end of this productive day :-)
00:14 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35428
00:36 tetragon joined #parrot
00:46 kid51 joined #parrot
01:08 dalek r35429 | rurban++ | trunk/t/pmc:
01:08 dalek : Fix complex.t for MSWin32, tested on MSVC6 and mingw (XPSP2 libc)
01:08 dalek : - fix os_name to osname
01:08 dalek : - 2 failing sinh tests
01:08 dalek : - no failing inf test
01:08 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35429
01:10 TiMBuS is there a parrot op for unlinking files? it's specified in the PDD but there's no io opcode for it nor is it in the File pmc where 'rename' and 'copy' are found
01:13 TiMBuS perl 6 uses os.'rm'() hm.
01:22 tetragon_ joined #parrot
01:42 jrockway joined #parrot
01:43 nopaste joined #parrot
01:48 dalek r35430 | jkeenan++ | trunk/t/library (2 files):
01:48 dalek : Applying patch submitted by Geraud Continsouzas in
01:48 dalek : https://trac.parrot.org/parrot/ticket/160:  conversion on tests from Perl 5
01:48 dalek : (using Parrot::Test) to PIR.
01:48 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35430
01:50 omega joined #parrot
01:52 GeJ kid51: thanks
01:56 kid51 GeJ, Coke:  review what I wrote in ticket:  https://trac.parrot.org/pa​rrot/ticket/160#comment:4
01:56 kid51 msg DietCoke review what I wrote in ticket:  https://trac.parrot.org/pa​rrot/ticket/160#comment:4
01:56 purl Message for dietcoke stored.
02:45 Fayland joined #parrot
02:45 Zaba_ joined #parrot
02:51 Andy joined #parrot
03:15 dalek r35431 | pmichaud++ | trunk/languages/perl6/src/parser:
03:15 dalek : [rakudo]:  Throw an exception for bare "say" or "print" (RT #62028).
03:15 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35431
03:35 contingencyplan joined #parrot
03:45 baest joined #parrot
03:45 Zaba joined #parrot
03:45 Debolaz_ joined #parrot
03:47 gaz joined #parrot
03:49 Maddingue joined #parrot
03:57 dalek r35432 | pmichaud++ | trunk/languages/perl6/src/parser:
03:57 dalek : [rakudo]:  Missed a bare say in r35431 -- this fixes it (RT #62028).
03:57 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35432
04:31 Fayland joined #parrot
05:16 jimmy joined #parrot
05:31 cotto seen ovid?
05:31 purl ovid was last seen on #parrot 1 years, 13 days, 12 hours, 10 minutes and 58 seconds ago, saying: Just going by what pmichaud recommended.  [Dec 30 17:20:23 2007]
05:31 cotto 1 years ago
06:13 gmansi joined #parrot
06:17 moritz cotto: ovid was on #perl6 yesterday
06:21 cotto I figured
06:25 dtg joined #parrot
07:43 allison joined #parrot
08:03 namenlos joined #parrot
08:05 ask_ joined #parrot
08:10 krunen joined #parrot
08:10 elmex joined #parrot
08:12 UltraDM joined #parrot
08:21 iblechbot joined #parrot
09:05 alvar joined #parrot
09:23 masak joined #parrot
09:43 kj joined #parrot
10:10 galf joined #parrot
10:10 riffraff joined #parrot
10:37 tomyan joined #parrot
11:03 ruoso joined #parrot
11:34 donaldh joined #parrot
11:39 rob joined #parrot
11:40 bacek joined #parrot
11:55 dalek r35433 | cotto++ | trunk/t/pmc:
11:55 dalek : [t] pirify a PMC test
11:55 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35433
12:01 Lorn joined #parrot
12:18 jimmy joined #parrot
12:20 dalek r35434 | cotto++ | trunk/t/codingstd:
12:20 dalek : [t] make arg assert test's diagnostic output more useful
12:20 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35434
12:26 Casan joined #parrot
12:30 Lorn_ joined #parrot
12:52 dalek r35435 | cotto++ | trunk/t/codingstd:
12:52 dalek : [t] improve ordering of results from arg assert test
12:52 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35435
12:54 DietCoke joined #parrot
12:57 jonathan hi all
12:58 moritz hi jonathan
12:58 DietCoke ~~
12:59 masak \o
13:00 moritz /
13:00 moritz skewed multi-user smilies :-)
13:00 masak :)
13:02 jimmy yo
13:03 dalek r35436 | cotto++ | trunk/src/gc:
13:03 dalek : [gc] add some ASSERT_ARGS macros
13:03 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35436
13:04 * jonathan will hack on the extra annotations lookup stuff that Francois suggested, then get back onto Rakudo things.
13:05 * moritz wants line numbers in rakudo error messages!
13:05 jimmy jonathan++
13:06 * jimmy want documentation i18n and will apply himself to translation.
13:07 moritz I started to translated S13 and S05 to German, and failed miserably on both fronts
13:07 * jimmy want to know whether perl6 would support documentation i18n
13:07 moritz it was just too much to do, and too little understanding
13:08 moritz but it forced me to think about some aspects, which was surely a good thing
13:08 jimmy me too.
13:08 jimmy like ubuntu, I want it will be support i18n officially
13:09 * jimmy should say expect.
13:10 moritz jimmy: a good steps towards that is to standardize error message in classes and/or roles
13:11 moritz jimmy: then error messages can easily get text in different languages
13:12 jimmy moritz: I don't expect it's 18n in internally, that's not the problem to us, it's the documentation.
13:17 moritz jimmy: but if we want a culture of i18n, we better bring it to the core as well
13:18 elmex joined #parrot
13:19 jimmy yep, that is why we expect perl.org would be support it officially
13:19 jimmy s/be//g
13:21 Zaba joined #parrot
13:22 alvar joined #parrot
13:39 jonathan moritz: That will need to wait on pm doing Cursor refactor in PGE. :-) But hopefully that is soon.
13:39 * jonathan has bad stomach and just had to rest for a while :-(
13:48 DietCoke cotto: I can't test msvc.
13:48 DietCoke cotto: nevermind, that message is from september.
13:48 jimmy jonathan: Have a good rest
13:50 DietCoke msg kid51 Not much to review in the comment. I agree that someone should make sure that the conversion of tests was sane; perhaps that should have been done before the commit. <shrug>
13:50 purl Message for kid51 stored.
13:57 tetragon joined #parrot
14:02 AndyA joined #parrot
14:02 riffraff joined #parrot
14:03 Zaba joined #parrot
14:08 gryphon joined #parrot
14:16 Whiteknight joined #parrot
14:25 dalek r35437 | fperrad++ | trunk/languages/markdown:
14:25 dalek : [Markdown] standardization of the interface
14:25 dalek : - downcase language name
14:25 dalek : - now, generate PIR code that returns a string containing HTML
14:25 dalek : So, now usable from Rakudo :
14:25 dalek :    eval($source, :lang<markdown>)
14:25 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35437
14:27 * PerlJam wonders what it really means to eval markdown in rakudo
14:28 moritz PerlJam: construct an HTML string from markdown source?
14:30 dalek r35438 | fperrad++ | trunk/languages/lua (4 files):
14:30 dalek : [Lua] standardization of the interface
14:30 dalek : - downcase language name
14:30 dalek : So, now usable from Rakudo :
14:30 dalek :    eval(q<print 'hello'>, :lang<lua>)
14:30 dalek : but causes segfault
14:30 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35438
14:33 PacoLinux joined #parrot
14:33 PerlJam fperrad has an interesting idea of "usable"  :)
14:34 PerlJam fperrad++ though
14:37 masak that eval($code, :lang<other-lang>) is potentially a big deal.
14:38 moritz masak: have you seen Tene++s blog post about it?
14:38 masak no, not yet.
14:38 masak I'm behind on the firehose.
14:39 moritz http://blogs.gurulabs.com/stephen/200​9/01/parrot-speaks-your-language.html this is the one I'm reffering to
14:39 shorten moritz's url is at http://xrl.us/becavj
14:39 jonathan Gah! Am I being incredinbly stupid, or is there no way to know the current program counter from within a v-table method?
14:42 masak moritz: that is seriously cool.
14:42 * jonathan thinks there isn't. :-|
14:42 moritz masak: I can only second this :-)
14:43 masak anyone for thirds? :)
14:46 masak I'd just like to say, a propos merging massive branches, that November and Druid suffered quite badly this time. there was one previous branch where pmichaud and others actually checked against November before merging. in the event of another big branch merge, I think we should try and do that.
14:46 elmex joined #parrot
14:46 masak neither November nor Druid have really recovered yet from the rvar merge.
14:47 moritz masak: I tried to test Novemeber with that branch, which suffered from not being able to built master, even from Rakudo trunk
14:47 moritz it didn't build completely, that is
14:47 masak moritz: we also have/had the orthogonal problem of not being able to make, yes
14:47 masak moritz: maybe you have seen the discussion on the mailing list
14:48 masak moritz: fortunately, 'make test' has no dep on 'make'
14:48 masak so it doesn't suffer from this problem.
14:48 cognominal how come that jq and Stephen Weels blogs are not on parrot planet?
14:49 moritz dunno, but it should be changed :-)
14:49 cognominal ho, I am wrong for Stephen Weeks
14:49 moritz and jq didn't do much parrot before the recent burst of activity
14:49 kj cognominal: they're very new I think
14:50 kj i mean, the blogs, not the people
14:51 masak :)
14:52 dalek r35439 | fperrad++ | trunk/languages/dotnet/build:
14:52 dalek : [dotnet]
14:52 dalek : - fix deprecated opcodes : store/find_global
14:52 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35439
14:56 jonathan Wow. All Parrot tests are success on Win32!
14:56 jonathan rurban++ # fixing complex.t
14:56 Whiteknight rurban++
14:59 moritz jonathan: could you please test if 'NaN ~~ NaN' works in Rakudo on Windows?
15:01 dalek r35440 | jonathan++ | trunk (2 files):
15:01 dalek : [core] Add support for getting annotations of callers/outers to ParrotInterpreter PMC, plus tests.
15:01 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35440
15:01 jonathan moritz: Just a moment. :-)
15:03 moritz jonathan: no hurry
15:06 donaldh hi, the rakudo changes over the weekend seem to have broken things like NCI. HAs anyone else seen this?
15:06 donaldh The last working revision I had was 35300
15:06 jonathan donaldh: Prime suspect is the rvar merge - a small test case would be immensely helpful, if possible.
15:06 gmansi joined #parrot
15:07 donaldh jonathan: funny you should mention that, I'm trying to get one.
15:08 donaldh I thought I could use ext/SQLite3/t/test.p6 but I can't get that to work on #35300 either.
15:08 jonathan donaldh: What kinda breakage are you seeing?
15:09 donaldh SEGV, corrupted stack and sitting in sqlite3_step.
15:10 donaldh I'll try and get better info on a linux box.
15:11 donaldh (was testing on Cygwin)
15:11 jonathan That's...quite an impressive failure mode.
15:12 donaldh It appears that when it comes to NCI it either works or blows up spectacularly.
15:13 Whiteknight "Parrot: it either works or blows up spectacularly"
15:14 PerlJam careful, fud mongers may be lurking
15:15 donaldh agreed. I'm just playing with the sharp end of the stick.
15:15 contingencyplan joined #parrot
15:17 donaldh jonathan: ext/SQLite3/t/test.p6 works on rev #35300 on Linux
15:18 nopaste "donaldh" at 144.254.91.104 pasted "PATCH: fix ext/SQLite3/Makefile.PL" (12 lines) at http://nopaste.snit.ch/15288
15:18 AndyA joined #parrot
15:19 donaldh You'll need to patch Makefile.PL first.
15:20 AndyA joined #parrot
15:22 jonathan donaldh: Plesae can you make ticket?
15:23 jonathan (rakudobug)
15:23 donaldh sure thing
15:23 jonathan Thanks!
15:35 donaldh jonathan: what are the biggest changes from the rvar merge?
15:36 jonathan donaldh: It was mostly a big refactor of parameter passing and variable handling.
15:36 davidfetter joined #parrot
15:36 donaldh Does it use different PMCs?
15:36 jonathan Hmm.
15:37 jonathan I don't *think* there were huge changes in that sense.
15:37 jonathan Signature PMCs got a bit more important.
15:37 donaldh It looks like premature freeing to me.
15:37 jonathan Does -G make a difference?
15:38 donaldh Uh, actually no.
15:38 jonathan (Could still be a premature free, but not by the GC in that case.)
15:39 * donaldh is running valgrind just now
15:40 pmichaud I know why the IO stuff failed, though.
15:41 jonathan hi pmichaud
15:41 pmichaud (druid and november failing) -- yes, we probably should've tested against druid and november before merging.  At some point though, there will be too many applications for us to reasonably test against.
15:42 pmichaud rakudo will be moving to a release cycle in the near future anyway, so hopefully that will bring some order to the process.
15:42 moritz sometimes I think we should have one nearly-real-world application in the test sutie
15:43 jq kj: erm, in fact i'm very old on the parrot project - i was here circa 2002 :-)
15:43 moritz or at least something that loads modules, and actually does something more with them than just calling subs
15:43 jq but it's true that i was inactive since quite a long time
15:43 pmichaud masak: anyway, I pledge to have november and druid fixed sometime today.
15:44 pmichaud (if they can be fixed :-)
15:44 masak pmichaud: that's a big promise. but thanks.
15:44 PerlJam pm: wow.  How do I get a pledge that rakudo will implement the entire Perl 6 spec by some reasonably short time frame?  ;)
15:45 * jonathan curses a design decision that went against what he wanted a couple of years ago.
15:45 pmichaud pj:  give me lots of room on the "reasonably short" component?  ;-)
15:45 kj jq: (very old) I know; that's why I mentioned "the blogs, not the people"
15:46 kj but maybe your blog has been around too, just havent' seen it before ;-)
15:46 rdice joined #parrot
15:46 jq kj: no, it's new. the whole promoting perl outside perlers, you know
15:47 jq i don't like blogging, but if everyone can talk about perl, parrot and stuff, then i'm contributing this
15:47 kj yes. it'd be nice if your blog can be added to planet.parrotcode.org
15:47 rdice joined #parrot
15:47 jq kj: oh in fact it was moritz who said "jq didn't do much parrot before the recent burst of activity"
15:48 pmichaud (general note)  one change that occurred as part of the rvar branch merge is that .new() now initializes attributes to be copies of the values instead of binding them to the values.
15:48 jq sorry then, my first comment should have hilighted moritz :-)
15:48 pmichaud that's likely why IO.readline isn't working -- the IO object is getting a copy of Parrot's FileHandle instead of binding to it.
15:48 moritz jq: aye, and I meant to add "lately" somewhere in my sentence
15:48 kj that's just because moritz hadn't seen you before ;-)
15:49 moritz if somebody shows up on IRC and already has a commit bit, he must be an "old" one who made a pause
15:49 moritz newcomers don't commit right away
15:49 donaldh pmichaud, maybe there's more of that?
15:50 jq note that i'm not committing on parrot internals
15:50 kj jq: but you'd be welcome too ;-)
15:50 pmichaud donaldh: could be.  Since FileHandles are really NCI thingies, that could be related to what donaldh is seeing.
15:50 jq (not counting doing the small hands to hunt old idioms to help allisson merge a branch)
15:50 jq kj: my c is too rusty
15:51 kj start a replacement project, that helped me :-) I'm sure there's some component that needs cleanup
15:53 PerlJam "start a replacement project" seems to mean something different than what you want it to mean  :)
15:53 kj PerlJam: not sure what you mean there..
15:54 PerlJam kj: if someone gloms parrot to project, then it sounds like "start replacing parrot"
15:55 kj oooh ok. I meant a parrot component
15:55 kj I heard the GC could use some work :-P
15:56 * kj ponders what to do after PIRC's finished. Which will be in the not too distant future
15:57 PerlJam kj: I've heard the GC could use some work  ;->
15:57 barney joined #parrot
15:57 kj lol
15:57 kj well, who knows :-)
15:58 moritz kj: do you have a mechanism that drops pirc into imcc's position right now?
15:58 moritz that would be very valuable for testing
15:59 PerlJam A pir subroutine-level profiler would be very useful too.
15:59 pmichaud jonathan: (general comment)  it appears that NCI methods don't play well with ObjectRefs.
15:59 riffraff joined #parrot
16:00 kj moritz: nope not yet. There's a few things that I need to do to complete PIRC
16:00 kj and then test it like crazy
16:00 kj but I feel I'm nearly there.
16:00 kj after that, I'll start looking into how to make IMCC and PIRC proper alternatives
16:00 jonathan moritz: say NaN ~~ NaN # 1 (on Win32/MSVC)
16:00 moritz jonathan: thanks
16:00 Whiteknight kj, do you need any specific help with PIRC?
16:01 Infinoid good morning everyone
16:01 jonathan moritz: However, unfortunately
16:01 jonathan say NaN # 0
16:01 jonathan :-(
16:01 pmichaud jonathan: that's unchanged from before, ys?
16:01 pmichaud *yes?
16:01 kj Whiteknight: not at this point I think.. yesterday I solved the last piece of the bytecode puzzle (so far as I know), now just some final bits
16:01 jonathan pmichaud: Let me know, if you get some specifics...
16:01 moritz jonathan: that's unfortunate, but doesn't affect the test that I wanted to unfudge
16:01 jonathan pmichaud: I think so...I don't remember changes.
16:01 pmichaud jonathan: I'm just changing IO so that it dereferences any ObjectRef before calling the NCI method.
16:02 Andy joined #parrot
16:02 jonathan Though was there not a change recently to have ObjectRef and Perl6Scalar both at work?
16:02 pmichaud yes, there's a change to that too, but it's unrelated.
16:02 jonathan OK.
16:03 pmichaud Rakudo now distinguishes between scalar variables  (Perl6Scalar)   and simple references (ObjectRef).  The primary difference being that ObjectRef is now used primarily for attaching properties (e.g., readonly) or copy-avoidance, whereas Perl6Scalar introduces special semantics in e.g., flattening.
16:04 dalek r35441 | jquelin++ | trunk/languages/befunge (3 files):
16:04 dalek : supporting new io subsystem, courtesy of francois perrad++
16:04 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35441
16:04 jonathan Aha, OK.
16:04 pmichaud thus, an ObjectRef to something can still flatten in list context, while a Perl6Scalar never flattens.
16:05 pmichaud anyway, it means that this now works correctly:
16:05 pmichaud perl6:  sub foo(@a) { .say for @a; };   foo([1,2,3]);
16:05 polyglotbot OUTPUT[1␤2␤3␤]
16:05 pmichaud perl6:  sub foo(@a) { .say for @a; };   foo(([1,2,3],8));
16:05 polyglotbot OUTPUT[1 2 3␤8␤]
16:06 jonathan Nice.
16:06 moritz and the first one is actually valid?
16:06 pmichaud moritz: yes.
16:06 Whiteknight kj: has there been any thought put into making our parsers more pluggable?
16:07 kj Whiteknight: except for my random thoughts, don't think so.
16:07 Whiteknight the IMCC isn't encapsulated well, and maybe we should look for a better solution
16:07 kj I think it shouldn't be too much of a problem
16:07 kj it's just a matter of adding some extra indirection
16:07 pmichaud masak:  I think I have RT #62166 fixed -- just doing spectest on it now.
16:07 kj and having both PIRC and IMCC "implement" some interface
16:07 Whiteknight I'm sure PIRC will be awesome :) I just want to make sure we have flexibility later when we need it
16:08 kj if we can make both PIRC and IMCC shared libraries
16:08 masak pmichaud++
16:08 kj that'd be nice, so that you can load one or the other.
16:08 masak I'll re-test Druid after that.
16:08 * jonathan has the role composition bug fixed too - just waiting for spectest to finish.
16:08 pmichaud note that any program/application that was using IO probably was having difficulties.
16:08 kj However, that's more of a Makefile thing + export/import stuff than anything else. And that's where I'd need help
16:08 Whiteknight a standard parser interface would also let us write front-ends for completely different languages in Flex/Bison too, if we wanted
16:09 jonathan Ah, crap...
16:09 pmichaud jonathan: I was thinking about role stuff last night -- can I see the diff?
16:09 masak pmichaud: I'm still surprised that such a regression could happen.
16:09 pmichaud masak: we don't have many good IO tests.
16:10 nopaste "jonathan" at 85.216.157.73 pasted "patch" (33 lines) at http://nopaste.snit.ch/15289
16:10 jonathan pmichaud: You can spot the bug in it too. ;-)
16:11 pmichaud jonathan: basically, Parrot's treatment of roles is that all conflicts have to be explicitly managed by the caller -- there's no default?
16:12 jonathan Right.
16:12 pmichaud also, given allison's latest post on the subject, I'm guessing you should use    class.'methods'()   instead of  inspect class, methods
16:12 pmichaud because inspect will be heavyweight and return clones of things.
16:12 jonathan Also, Allison decided that the moment you say "add_role" the compositon takes place.
16:12 pmichaud ouch.
16:13 jonathan Exactly.
16:13 pmichaud so, Rakudo could delay composition.
16:13 jonathan I said it was against the Perl 6 spec at the time.
16:13 jonathan Well, thing is, we don't really add_method anything now, I don't think?
16:13 pmichaud that's the point of the '!meta_compose'  function -- to give us a way to delay things.
16:13 jonathan It all goes in the namespace?
16:13 pmichaud right, we're not using add_method now.
16:13 pmichaud I mean you could delay the adding of the role.
16:13 jonathan So we know the full list of methods "up front"
16:13 pmichaud correct.
16:14 pmichaud it's just an idea -- I agree we probably don't need to delay yet.
16:14 Whiteknight perl 6 question: What file extension are perl 6 files going to use? *.pl again, or something like *.pl6?
16:14 jonathan We could but we would still have to build the "resolves" list at some point.
16:14 pmichaud right, and it's convenient to do it now.
16:14 pmichaud s/now/immediately/
16:14 jonathan But agree, we could if we need to in the future store up roles we need to compose at the end.
16:14 jonathan In some prop attached to the class or something.
16:14 pmichaud exactly.
16:15 jonathan If we're not going to need add_method, we can get away with it.
16:16 * jonathan tries to make sense of why this fails...
16:16 pmichaud I never could understand what 'resolve_method' actually does.
16:16 jonathan It specifies the list of method names that we should ignore when composing roles.
16:17 pmichaud okay.  "resolve" sounds like a verb here.  Seems like it should be "resolved_methods"
16:17 jonathan (Which gets us the "class provides it" semantics.)
16:17 PerlJam Whiteknight: I've been using .p6 but I don't know how atypical that is.
16:17 pmichaud but it's okay as-is.
16:17 Whiteknight okay, thanks
16:17 jonathan pmichaud: Probably.
16:17 purl Really? Probably? Are you Certain it's not certain? Are you sure it's unsure? I think you need to look harder.
16:18 pmichaud argh, one failure in spectest.
16:20 gryphon joined #parrot
16:20 bricas joined #parrot
16:21 * jonathan does the same as he had that worked before, and re-spectests
16:21 masak Whiteknight: you don't need any extension.
16:21 * bricas seems to have read something about a parrot/perl6 ubuntu repository in the meeting logs, but no details -- anyone have them?
16:21 Lorn joined #parrot
16:22 Whiteknight bricas: allison was talking about that at one point, you probably have to ask her for details
16:22 Whiteknight masak: I don't *need* an extension, but it makes things a lot easier on Win32
16:22 masak Whiteknight: ah.
16:22 masak Whiteknight: I seldom have to deal with that OS.
16:22 Whiteknight masak: I wish I seldom had to, but I have to use it for work :(
16:23 masak Whiteknight: but I remember the DOS and Win 3.1 times, so yeah. :)
16:23 masak 'associate file ending'...
16:23 * masak smiles at long-gone pain
16:24 bricas Whiteknight: okay, thanks.
16:25 bricas oh, i see something now (after a bit of google-ing): https://launchpad.net/~allison/+archive
16:26 PerlJam It's kind of sad that building a .deb, .rpm, etc. isn't part of the release process.
16:27 Whiteknight PerlJam, I think it will be eventually
16:27 jonathan Does anyone else in Rakudo spectest see:
16:27 jonathan t\spec\S05-substitution\subst.............................ok 7/56 skipped: various reasons, 4/56 unexpectedly succeeded
16:27 jonathan TODO PASSED tests 22, 25, 35, 38
16:27 pmichaud jonathan: yes.  I think those are false passes.
16:27 jonathan Seen those for a while now...
16:27 jonathan Ah, OK.
16:28 pmichaud .subst needs some love again.
16:28 Whiteknight I don't suppose particle is around today?
16:28 dalek r35442 | pmichaud++ | trunk/languages/perl6/src/classes:
16:28 dalek : [rakudo]:  Adjust reference/NCI semantics of IO.  Fixes RT #62166.
16:28 moritz .subst needs specs, mostly
16:28 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35442
16:28 moritz I hope I get around to writing a mail to p6l this week
16:30 pmichaud masak: =$*IN should be fixed for you in r35442.
16:30 masak pmichaud: goodie. testing.
16:31 * bricas now has parrot 0.8.1 from allison's ppa. o/
16:31 dalek r35443 | jonathan++ | trunk/languages/perl6/src/builtins:
16:31 dalek : [rakudo] Fix role composition regression from rvar branch merge; methods in the class now resolve conflicts again.
16:31 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35443
16:34 pmichaud jonathan++ # nice fix
16:34 Whiteknight what's the best mailinglist to join to follow Perl6 and Rakudo?
16:35 moritz Whiteknight: perl6-all (that gets you p6l, p6c, p6u and p6-announce)
16:35 Whiteknight okay, thanks
16:36 * jonathan ponders why get_hll_global ['Perl6';'Grammar';'Actions'], '$?BLOCK' returns null...
16:37 jonathan I should be looking at @?BLOCK[0]...
16:37 pmichaud jonathan: yes, I got rid of $?BLOCK, since it should always be aliased to @?BLOCK[0]
16:38 pmichaud (as a package variable, that is.)
16:38 rob joined #parrot
16:38 pmichaud I didn't want to play the "make sure we always set/reset $?BLOCK appropriately" game.
16:38 jonathan That's fine, I just didn't realize it was gone. :-)
16:39 pmichaud note that @?MODULE, @?ROLE, @?GRAMMAR, @?CLASS are all gone also :-)
16:39 jonathan We just have an @?PACKAGE and the blocks are annotated with a package type, right?
16:41 rob sorry to jump in, mk_language_shell.pl creates a stub src/parser/grammar-oper.pg but theres no entry for it in config/makefiles/root.in is this a bug?
16:42 pmichaud it's done as part of the "compile grammar.pg" line
16:42 pmichaud grammar.pg and grammar-oper.pg are compiled as part of the same grammar
16:42 pmichaud i.e., it's made as part of building   src/gen_grammar.pir
16:43 pmichaud jonathan: yes, the blocks are all annotated with <pkgtype>
16:43 pmichaud okay, I have a _ton_ of errands I have to run today, so I'll be back later.
16:43 pmichaud masak:  let me know if Druid is working after the r35442 fix.
16:43 masak pmichaud: yes, sure. recompiling right now.
16:44 masak pmichaud: aye, it's working! \o/
16:45 pmichaud I'll look at November later today
16:46 pdcawley joined #parrot
16:48 masak pmichaud: I haven't dared do 'make' on November in a while, but a couple of days ago precompilation made large parts of the test suite fail. but even modulo that, there are some test failures which I've been addressing one by one during the weekend.
16:50 masak pmichaud++ # fixing Druid
16:53 Theory joined #parrot
16:57 * DietCoke chuckles at http://thedailywtf.com/Articl​es/Longjmp--FOR-SPEED!!!.aspx
16:58 jonathan lol
17:01 Whiteknight I can't beleiev somebody thought that was "faster" then a simple loop
17:01 masak I used to subscribe to The Daily WTF, but I had to stop, because it made me too depressed.
17:01 Whiteknight a clear sign that somebody thinks themselves to be cleverer then they actually are
17:03 hercynium joined #parrot
17:05 masak pmichaud: AFAICT, the testfails in t/storage are not your fault, but are actually November being broken. I suspect that the same is true for t/tags.
17:05 masak pmichaud: so, as a matter of fact, I don't think Rakudo needs any fixing wrt November. :)
17:05 jonathan What some people miss, is that for things like loops, the compilers these days - especially C ones - are *far* better at optimizing them than people, most often. So it pays to write simple code that the optimizer can deal with. :-)
17:06 masak (simple code)++
17:08 masak pmichaud: that said, ./test_wiki.sh gives an interesting parser error :)
17:12 namenlos this might be a stupid question: i got parrot installed, but got no clue, how to run perl6 scripts. could someone point me into the right direction?
17:13 cognominal c'est eux qui se genent pas, il font du yoyo entre mon fixe et mon portable
17:13 cognominal avec un numéro masqué bien sûr
17:13 jonathan cognominal: Oui. Merde.
17:13 masak namenlos: http://www.perlfoundation.​org/perl6/index.cgi?rakudo
17:14 masak cognominal: il font du yoyo.
17:14 jonathan ;-)
17:14 cognominal arf, wrong channel
17:14 * jonathan wonders if yoyo is the same in French as in English.
17:14 masak jonathan: aye.
17:14 cognominal I was complaining about direct marketting
17:14 jonathan Je ne parle pas le francais.
17:14 cognominal yes you do
17:14 masak jonathan: he said something was yoyo-ing between fixed and laptop
17:15 masak Je ne parle pas le francais either.
17:16 namenlos masak: thanks. could i have found this information somewhere in the source tarball?
17:16 masak namenlos: yes, I think so.
17:16 masak namenlos: there are README files both in parrot/ and perl6/
17:17 masak yup, the latter one says how.
17:17 namenlos masak: ... sorry, my fault
17:17 masak namenlos: no worries.
17:17 masak namenlos: let us know how that perl6 scripting goes!
17:19 namenlos masak: sure - i love perl 5 very much, and in some blog chromatic (or so) said, there should be more testers...
17:19 namenlos masak: so i am trying to get those things done.
17:19 masak namenlos: that's the spirit.
17:19 masak namenlos: might I also recommend #perl6 on irc.freenode.net?
17:19 namenlos masak: already there ;)
17:20 * moritz sees the Nameless One, but he has a name. Pretty confusing ;-)
17:20 masak moritz: I think it's sort of a zen riddle.
17:20 jonathan In my next commit, I'm going to put in some code relating to type registry stuff. However, it's going to be disabled for now. I'll do follow-up commits that bring us towards being able to enable it.
17:20 namenlos masak: one thing bothers me: there seems to be no man pages for perl6 (like perl5) only those extremely long synopsis, which are hard to read
17:21 jonathan namenlos could also be a reference to http://herzeleid.com/en/lyrics/reise_reise/los
17:21 namenlos masak: i mostly learned perl out of the man pages
17:21 masak namenlos: that's why you need #perl6
17:21 ruoso joined #parrot
17:21 masak namenlos: for now.
17:21 purl For now, it works but once I figure out how to get the lube to stick to the mouse I'll be able to market it.
17:21 masak purl: forget for now
17:21 purl masak: I forgot for now
17:22 namenlos masak: the name comes only becaue all other names were already reseved on freenode (and i am german speaking...)
17:23 masak namenlos: I guessed that. it's a nice (non-)name
17:23 namenlos yes
17:24 donaldh Whiteknight: .p6 +1
17:26 tomyan std: say [*] < 1 2 3 >;
17:26 dalek r35444 | jonathan++ | trunk/languages/perl6/src/parser (2 files):
17:26 dalek : [rakudo] Add more type registry implementation. For now it's not enabled, as it causes us to regress two sanity tests and 20ish spectests; follow up commits will deal with this.
17:26 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35444
17:26 jonathan tomyan: std is on #perl6, not #parrot
17:26 tomyan yeah, just realised
17:26 jonathan :-)
17:27 moritz rakudo: say [*] < 1 2 3 >;
17:27 polyglotbot OUTPUT[6␤]
17:32 Whiteknight I've got Padre asking me what to name my little Perl6 test file, and it's causing like a little crisis
17:34 Lorn_ joined #parrot
17:35 moritz OMG
17:39 PerlJam Whiteknight: foo.p6
17:39 PerlJam :-)
17:41 moritz test files end in .t, please ;-)
17:49 dalek r35445 | jonathan++ | trunk/languages/perl6/src/classes (2 files):
17:49 dalek : [rakudo] Mark Bool and Order as enums, so they are seen as types during the parse.
17:49 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35445
17:50 dalek r35446 | jonathan++ | trunk/languages/perl6/src/parser:
17:50 dalek : [rakudo] Fix up type variable declaration handling in is_type.
17:50 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35446
17:51 Whiteknight yeah, i'm definitely not writing any test files
17:51 Whiteknight my perl6-foo isn't strong enough yet
17:52 kj isnt' that the way to learn it? ;-)
17:53 ask_ joined #parrot
17:57 particle Whiteknight: you rang?
17:57 Whiteknight oh yeah, I wanted to know if we were still ripping STM out and, if so, when?
17:57 kj isn't STM working at all?
17:57 kj or maybe it's a nice start, but not finished?
17:58 lathos I started prototyping the new Parrot string stuff... in Perl 6.
17:58 kj Maybe it's an idea to store these "it's a good idea, but never finished" code somewhere
17:59 Whiteknight yeah, that's what I was thinking
17:59 Whiteknight I'm happy to rip it out if we can replace it with something better later
17:59 moritz lathos: anything you want to share yet?
18:00 lathos Not yet, I really have only just started. Hopelly by the end of tomorrow I'll get most of the functions working with a test suite.
18:02 pmichaud "good idea, never finished"  -- I wouldn't be opposed to having an "ideas" branch where such things could go.
18:02 moritz lathos: when you have something, tell me. I'm interesting in string handling, but my C-fu is *very* bad, so this could be an opportunity for me to help nonetheless
18:03 particle of course, all those ideas will get stale if there aren't frequent merges
18:03 pmichaud particle: they're already stale, I think.
18:03 pmichaud and I was about to say that anything that moved to that branch is likewise a candidate for simply being deleted.  After all, svn already stores the "good idea, never finished" stuff.
18:03 particle so store them in a ticket
18:03 kj is it possible to mark some directory as "don't check this out by default"? So if you do a co, you don't get it, unless you request it.
18:04 particle kj: that's what branches are for :)
18:04 pmichaud or even tags
18:04 particle LNG_STM
18:04 particle "last known good"
18:04 kj mmm. ok. I thought branches are more for developments that are too big to do in trunk, and not to bother other developers
18:04 particle or whatever
18:05 jonathan particle: known begins with a K ;-)
18:05 pmichaud bah, I don't even need that.  I can just look in tags/RELEASE_0_8_2  and find the version of stm that we had as of 0.8.2
18:05 pmichaud We _really_ don't need a separate place for this.  Subversion already provides that.
18:06 particle yep
18:06 pmichaud okay, more errands -- bbl.
18:09 jonathan Damm. Stack overflows are...hard to debug...
18:09 Whiteknight so particle, we're ripping out STM?
18:09 Whiteknight cause I'm just crazy enough to do it
18:10 particle Whiteknight: yes, start a branch for it
18:10 Whiteknight will do
18:16 particle are there updates to the parrot roadmap? i saw a bunch of commits and branch merges in the past few days
18:16 particle i see annotations landed. yay!
18:16 dalek r35447 | jonathan++ | trunk/languages/perl6/src/parser:
18:16 dalek : [rakudo] The term action was emitting sub calls incorrectly (not doing namespace lookups), however we didn't notice in a couple of tests because of the typename hack. The type registry picks this up. This patch is from rakudoreg branch, with updates to follow other changes to this action method.
18:16 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35447
18:16 jonathan particle: Yeah, but I don't know what the "bytecdoe test suite" or similar that was assigned to me is.
18:17 particle we need a way to test that the annotations work as expected
18:17 jonathan particle: t/op/annotate.t has nearly 30 tests ;-)
18:18 particle damn, parrot is failing to build on windows 7, due to some manifest problem (not parrot manifest)
18:18 particle i don't understand mt.exe
18:19 dalek r35448 | jonathan++ | trunk/languages/perl6/src/parser:
18:19 dalek : [rakudo] Make enums declare the types they introduce in the block symbol table.
18:19 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35448
18:19 * jonathan is down to three spectests failures to investigate before he can turn on type registry.
18:25 barney Is there an up-to-date Git mirror of Parrot ?
18:26 dalek r35449 | jonathan++ | trunk/languages/perl6/src/builtins:
18:26 dalek : [rakudo] Refinement types need to do the Abstraction role.
18:26 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35449
18:29 jonathan OK, I've got two more failures. Both look...curious.
18:30 jonathan pmichaud: ping me when you're back later
18:30 purl I can't find me in the DNS.
18:30 jonathan OMG I CAN'T FIND YOU IN THE DNS EITHER PURL!
18:31 particle1 joined #parrot
18:34 Infinoid purl: therefore you don't exist.  so ssssh.
18:34 purl Infinoid: excuse me?
18:35 slavorg joined #parrot
18:37 * jonathan tries removing the "uppercase first" hack to make sure there's no more failures by doing that
18:38 DietCoke particle: your name is on the deprecation todo for 0.9.0.
18:38 DietCoke do you have any plans for items remaining in Deprecated.pod that don't require one more release?
18:39 particle i didn't parse that
18:39 DietCoke "are you going to be working on any of those tickets."
18:39 particle no :(
18:40 DietCoke i'll keep slugging along, then.
18:40 particle DietCoke++
18:41 DietCoke I won't /finish/, mind you.
18:41 DietCoke I will probably write up something to hit the list tonight with the remaining items and ask for volunteers.
18:41 jonathan particle: So if we have tests for annotations, can I mark that other task as landed?
18:42 particle jonathan: you're not testing the bytecode, there. that is, that it's written to a file, loaded back in, and working as expected
18:42 jonathan particle: Ah.
18:42 jonathan That's a good point.
18:43 jonathan The awesome. I can rip out the "starts with uppercase" type hack and it's just the same two spectest fails that are there with it.
18:44 particle inter_call.obj : error LNK2019: unresolved external symbol _CONST_STRING referenced in function _start_flatten
18:44 particle libparrot.dll : fatal error LNK1120: 1 unresolved externals
18:44 particle wonder why i'd get that only on windows 7?
18:46 Whiteknight because windows 7 sucks
18:46 Whiteknight actually, I dont know that. I haven't tested windows 7 yet
18:46 particle i've got no time for haters.
18:46 Whiteknight it is still just a beta at this point, you can't expect it will get everything 100% right
18:47 jonathan Linking, OTOH, is an odd thing to get wrong... :-S
18:48 jonathan Oh, it's not link time...
18:48 jonathan Oh, it is.
18:48 jonathan particle: Check that the CONST_STRING ain't over multiple lines, but I don't know why that would trigger on just that platform.
18:50 particle i'm installing ack to look for the #define
18:50 particle it's faster than installing ctags
18:53 particle where is CONST_STRING defined?
18:53 particle i only see it in src/dynpmc
18:56 allison joined #parrot
18:56 chromatic joined #parrot
19:01 mberends joined #parrot
19:02 jonathan particle: I'm not sure, chromatic probably knows. ;-)
19:03 jonathan (where CONST_STRING is defined, that is)
19:03 particle jonathan: i was going to wait until type registry went in to rebuild parrot/rakudo, should i bother?
19:04 particle if it'll be within the next hour, i'll wait
19:04 chromatic .str files
19:04 purl .str files are just doing that
19:04 particle inter_call.obj : error LNK2019: unresolved external symbol _CONST_STRING referenced in function _start_flatten
19:04 particle chromatic: i get this on windows 7 only
19:04 chromatic Don't tell me Windows 7 is too dumb to understand #include "oo.str"
19:05 chromatic Though that's actually src/inter_call.str
19:05 jonathan particle: Depends on whether I catch pmichaud by then or have a flash of inspiration on the Right Solution... :-)
19:05 particle chromatic: that file is empty
19:06 chromatic touch src/inter_call.c and make src/inter_call.o
19:06 chromatic That file then should not be empty.
19:06 particle ok, so no idea why it was empty, but rebuilding now
19:07 chromatic Timestamps?
19:07 purl Timestamps are seconds since epoch, plus about 6 more digits on the end...
19:07 particle oh, and maybe make tags-vi should be modified to search .str files as well
19:07 particle or, ctags should search .str files, or whatever
19:08 chromatic It'll get a lot of duplicate hits.
19:08 particle yeah, well, right now it only gets hits in the src/dynpmc/*.c and languages/perl6/src/classes/*.c files
19:08 particle that's less than helpful, since i should be looking in *.str
19:09 particle and why don't dynpmc files use .str files?
19:10 chromatic dynpmc's CONST_STRINGs can't get compiled into libparrot, because dynpmcs don't get compiled into libparrot.
19:10 dalek r35450 | Whiteknight++ | trunk/docs/book:
19:10 dalek : [Book] Update the section about methods, include some more details about syntax, the "self" keyword and the ":invocant" flag.
19:10 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35450
19:11 particle ok, clearly i forgot what .str files were for. of course.
19:11 chromatic Next question: "Why do dynpmcs use CONST_STRING then?"
19:11 chromatic So that people can copy code from PMCs into dynpmcs and not get things wrong.
19:12 szabgab chromatic: can you please take a look at https://trac.parrot.org/parrot/ticket/77 ?
19:12 Whiteknight maybe it's worth renaming them to PARROT_INTERNAL_CONST_STRING instead, so people in external modules don't try to use them?
19:13 szabgab chromatic: and/or tell me whom else should I nag with the Parrot::Embed issues?
19:14 chromatic Ooh, and make our code lines even longer.
19:15 chromatic szabgab, it's on my list.  I'm in the office today but I might be able to get to them tonight.
19:16 szabgab chromatic: thanks, I would also like to talk to you about how the API should be for running Perl 6 code via Parrot::Embed should look
19:16 * PerlJam recalls that this thing has been on chromatic's list for a while
19:16 PerlJam I know a similar conversation happened back in Nov or Dec.
19:17 szabgab so I can write tests
19:17 chromatic I haven't had anything to use it for, so it's stalled a little bit.
19:17 chromatic szabgab, if you can show me a decent API you'd like to use, that'll help me figure it out.
19:17 szabgab PerlJam: yeah, that was when I was writing my grant proposal
19:17 Whiteknight chromatic: I'm just saying, if you don't want people to use it outside of core, then you should name it something that people shouldn't use outside of core
19:17 szabgab now I got the grant so I should write it
19:18 chromatic I do want people to use it out of core!
19:18 chromatic It's just a name for a concept.  The implementation of that concept varies slightly out of core
19:18 Whiteknight ...oh. EIGNOREWHITEKNIGHT
19:18 chromatic There's no reason that name can't encapsulate those different behaviors such that no one else has to know the implementation difference.
19:19 chromatic ... not that the name CONST_STRING is really descriptive (though not that I have anything substantially better).
19:21 ask_ joined #parrot
19:22 DietCoke we could use const_string. oh wait, we do.
19:26 chromatic That isn't descriptive either.
19:27 chromatic Plus do you want to explain to dynpmc writers why core PMCs use CONST_STRING and they have to use const_string for compile-time constant strings?
19:28 ask- joined #parrot
19:28 riffraff joined #parrot
19:29 particle #define const_string CONST_STRING ;)
19:30 chromatic The dynpmc tool defines CONST_STRING so dynpmc writers can use it.
19:35 DietCoke chromatic: I like being able to write the same code in dynpmcs.
19:35 estrabd joined #parrot
19:36 chromatic I figured.
19:39 dalek r35451 | chromatic++ | trunk (3 files):
19:39 dalek : [Parrot::Embed] Added a test file for using multiple languages from
19:39 dalek : Parrot::Embed (Gabor Szabo, TT #77).
19:39 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35451
19:40 szabgab chromatic: added some ideas about API to #77  and thanks for applying my test
19:40 chromatic I need to think about the API some, but that gives me some ideas.
19:42 dalek r35452 | chromatic++ | trunk/examples/nci:
19:42 dalek : [examples] Fixed NCI ls example not to rely on the POSIXly-correct behavior of
19:42 dalek : dlopen NULL to get a handle to the current process image.  Only sane operating
19:42 dalek : systems support this, which is a problem because two of our target platforms
19:42 dalek : are not sane with respect to this standard POSIX behavior.  Try to guess which
19:42 dalek : ones!
19:42 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35452
19:43 PerlJam chromatic++ indeed.
19:44 chromatic Oh, dogging on Windows and Mac OS X is easy.
19:44 dalek r35453 | chromatic++ | trunk/src:
19:44 dalek : [src] Removed some PMC DOD registration from invoke code.  If PObjs are getting
19:44 dalek : collected while they're only available from the stack, we have a problem.  If
19:44 dalek : weird errors occur, this is the first commit to consider reverting.
19:45 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35453
19:45 chromatic Fun fact: Smalltalk 80 could only support 32768 objects in an image.
19:49 particle who could ask for anything more?
19:50 chromatic 15 bit integers
19:50 chromatic Object references in Smalltalk 80 weren't pointers but instead indices into tables, somewhat like Unix file descriptors.
19:50 chromatic Thus the GC could compact heap segments almost transparently.
19:50 chromatic Stop the world during compaction, of course....
19:54 riffraff chromatic, where SmallInt excluded ?
19:56 chromatic I don't know exactly where SmallInt fell into Smalltalk 80, unless they were the 15-bit integers that didn't participate in GC.
19:58 riffraff seems reasonable
20:00 szabgab crap, it seems there are some paths compiled into parrot
20:00 szabgab that's probably part of the problem that stops the parrot installation
20:01 riffraff is a patch to introduce HLLRepl  taking care of  HLLCompiler#interactive interesting ?
20:01 Zaba joined #parrot
20:01 PerlJam szabgab: so ... are you vounteering to fix the "parrot install problem"?  :-)
20:01 szabgab after building parrotand rakudo  in a directory I renamed the directory and tried to run parrot.exe languages\perl6\perl6.pbc -e "42.say"
20:03 dalek r35454 | infinoid++ | trunk/languages/perl6/src/parser:
20:03 dalek : [cage] Fix a t/codingstd/cuddled_else.t failure.
20:03 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35454
20:03 dalek r35455 | infinoid++ | trunk/src/gc (4 files):
20:03 dalek : [cage] Tag some GC functions with ASSERT_ARGS().
20:03 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35455
20:03 dalek r35456 | infinoid++ | trunk/docs/project:
20:03 dalek : [docs] reformat the quoted text in cage_cleaners_guide.pod slightly,
20:03 dalek : so perldoc's output looks better on an 80 column terminal.
20:03 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35456
20:03 szabgab "load_bytecode" couldn't find file 'PCT.pbc'
20:03 Whiteknight yeah, I was getting that error too
20:04 szabgab PerlJam: I am just complaining, someone might listen to
20:04 szabgab so it appears that the path to PCT.pbc (and maybe other files) is hard code in either parrot or rakudo
20:05 szabgab hard coded = I mean after compilation
20:05 particle riffraff: submit the patch and find out ;)
20:05 Whiteknight I think Parrot expects them all to be in PARROT_ROOT/runtime/*
20:05 szabgab and it is not a relative path, nor does it look for ...
20:05 szabgab the real PARROT_ROOT environment variable?
20:06 PerlJam well ... remember how long it took us to get a relocatable perl?
20:06 particle szabgab: did you build an installable parrot?
20:07 szabgab particle: no, is there such a thing?
20:07 szabgab particle: how do I build that one?
20:07 particle make installable
20:07 particle seems it's not documented in make help :(
20:07 Whiteknight I've never heard of make installable
20:08 particle ah, it *is* there
20:08 szabgab I never read the help :-(
20:08 Whiteknight ENODOCUMENTATION
20:08 particle @echo "  installable:       same as 'world', but targets for installation"
20:08 particle make help
20:08 purl make help is my friend
20:08 szabgab ok, trying it again
20:08 Whiteknight yeah right, purl has no friends
20:08 PerlJam particle: when did that get added?  this is the first time I've heard of it too
20:08 PerlJam (and I've run make help before and not noticed it either)
20:09 PerlJam maybe I was just myopic
20:09 particle it's been around as long as i can remember
20:09 particle 10883       rafl INSTALLABLEPARROT   = $(CUR_DIR)/installable_parrot$(EXE)
20:09 particle at least since r10883
20:10 PerlJam wow. My eyes must've glazed over it each time.
20:14 geof joined #parrot
20:15 silug joined #parrot
20:22 riffraff problem: in my code I need to generate labels and branch statements. I'm relying on Op.new(:inline("LABEL:")  inline('goto LABEL') but this seems fragile
20:23 riffraff is there another way?
20:23 szabgab particle: when building with make installable, will that mean that I can move the files to other locations and it should work ?
20:23 DietCoke is that NQP?
20:24 DietCoke szabgab: I think that's what's /implied/, anyway.
20:24 Whiteknight urg, I don't think "make reallyinstall" does what it is supposed to do on Win32
20:24 szabgab I am building it now on windows, maybe I should first try it on linux, it is both faster for me and more handy
20:25 Whiteknight it installed everything int C:\usr\local\* instead of C:\Program Files\Parrot\*
20:25 DietCoke did you /tell/ it to install to c:\pf\p ?
20:25 szabgab Whiteknight: I don't want to install it, just to make it movable
20:25 szabgab (or maybe it is a bigger problem ?)
20:26 DietCoke szabgab: rurban is a good person to talk to about what currently works outside of "running from the build directory"
20:26 szabgab but rurban is not here now
20:27 szabgab and I am falling asleep soon
20:27 Whiteknight DietCoke: I didn't tell it anywhere to install it, but it should still follow platform conventions
20:28 particle tuits, Whiteknight, tuits.
20:29 Whiteknight particle: I think it's spelled "tickets", and I'll open one :)
20:32 szabgab particle: rakudo does not have an installable target
20:34 DietCoke Whiteknight: I don't thing c:\pf\ is expected for non-GUI apps.
20:34 DietCoke "think"
20:34 DietCoke and I'd rather just see people use the prebuilt windows version if they're not hacking.
20:35 Whiteknight DietCoke, I would be fine with C:\Parrot\* instead. That would be acceptable
20:36 szabgab particle: so I built parrot using make installable and rakudo using make
20:36 Whiteknight but C:\usr\runtime\* is so far away from conventions as to be considered "just plain wrong"
20:36 szabgab but if I move the build tree to another place, I still get the same error about not finding PCT.pbc
20:36 Whiteknight I'm pretty sure that the gnuutils install to C:\Program Files\GNUUtils\* or something like that
20:37 Whiteknight and so long as the folder name ends up in your path, that's all that matters
20:40 Casan best practice should be to follow platform conventions and that is a suggested installation in \Program Files\ on windows, yet smart perl programmers should be able to add the extra finish and let it be installable anywhere suitable to the user.
20:41 Casan and moveable.
20:42 Casan with icing on the top and a cherry to pop even.
20:42 DietCoke casan; that's fine; my point is to make that part of the nice gui installer for windows, not "make install".
20:42 DietCoke (just to limit the amount of effort.)
20:44 Casan DietCoke: ahh I understand. just needed to throw in my �.02 as it should still be possible to compile a movable parrot.exe for example which can be deployed with environments such as the padreonportablestrawberryperlandparrotandblabla project :)
20:44 DietCoke (i'd gladly apply a patch to config patch for that.)
20:50 szabgab particle: https://trac.parrot.org/parrot/ticket/167
20:52 dalek r35457 | Whiteknight++ | trunk/docs/book:
20:52 dalek : [Book] Small reconfigure of the headings in chapter 4. VTABLEs are defined on a class, so I'm putting them in the section with namespaces and methods.
20:52 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35457
20:52 particle1 joined #parrot
20:56 lathos How do you do bitwise negation in Perl 6? (And should it be implemented in Rakudo)
20:56 lathos I mean the equivalent of Perl 5 ~$a
20:59 * barney created http://github.com/bschmalhofer/hq9plus/tree/master
20:59 particle1 +^$a
20:59 TimToady or ~^$a, if you want stringwise
20:59 particle right-o
21:05 nopaste "particle" at 76.121.106.245 pasted "reference grammar for S19" (94 lines) at http://nopaste.snit.ch/15290
21:06 particle whoops, old habits... that should have been to #perl6
21:06 particle gotta move some smartlinks based on today's refactoring, too
21:09 lathos Ah, cool.
21:09 mberends Hi everyone, you're welcome to browse a new Rakudo HTTP::Daemon running at http://autoexec.demon.nl:8888/ , the source code is within http://github.com/eric256/​perl6-examples/tree/master
21:10 lathos Oh urgh, this is so ugly but it works.
21:10 Infinoid aaw, Firefox can't establish a connection to the server at autoexec.demon.nl:8888.
21:10 lathos $src.flags +&= +^ StringCOW;
21:12 Infinoid mberends++
21:13 DietCoke mberends++
21:14 particle lathos: ugly indeed, but short :)
21:15 lathos Right, because $foo &= ~ StringCOW is so long winded.
21:16 DietCoke does svn (1.4) track ancestry of where merges came from? or only that a commit occurred in the new location?
21:16 mberends Infinoid: try again, the traffic has subsided now.
21:16 DietCoke (it doesn't, right?)
21:16 allison joined #parrot
21:16 DietCoke (trying to make sure I'm not shooting myself in the foot doing a merge a certain way here.)
21:17 Infinoid mberends: yeah, seems like it drops connections at times, but if I wait a bit and hit refresh it works eventually.
21:17 desertmax joined #parrot
21:17 Infinoid DietCoke: I don't think it tracks either, you have to specify revision ranges manually
21:18 Infinoid or are you talking about history/log context?
21:18 dalek r35458 | bernhard++ | trunk (11 files):
21:18 dalek : Remove languages/hq9plus. It's now on http://github.com/bschmalhofer/hq9plus/tree/master
21:18 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35458
21:19 DietCoke barney: ping.
21:19 particle DietCoke: you need svn 1.5 for that
21:20 DietCoke msg barney thanks for updating LeaveTheNest. =-)
21:20 purl Message for barney stored.
21:22 TimToady bit manipulations *ought* to be ugly  :)
21:22 mberends joined #parrot
21:23 DietCoke I sense a YourMom joke.
21:23 barney DietCoke: Moving to git, github, is very easy if you don't need to keep history.
21:23 DietCoke barney: I was just about to ask you to update the page after the ping, then saw you already had.
21:23 DietCoke more choices are fine. =-)
21:23 DietCoke do you have any notes you could add about the cleanup you did in the parrot repo?
21:24 DietCoke and thanks for updating the languages wiki page. you are on top of this. =-)
21:24 barney just 'ack-grep hq9plus' and remove without thinking
21:25 particle is there a languages page on parrot.org that gives info on where to find language project pages, download/package info, etc?
21:25 DietCoke particle: yes, it's called "Languages". (but it's on trac)
21:25 particle heh, nice
21:25 barney https://trac.parrot.org/parrot/wiki/Languages
21:27 DietCoke barney+=
21:27 DietCoke barney++
21:27 DietCoke karma barney?
21:27 purl barney has karma of 19
21:28 DietCoke barney+=5
21:28 DietCoke karma barney
21:28 purl barney has karma of 19
21:28 DietCoke ah well. =-)
21:28 moritz barney++
21:33 TimToady lathos: if there's something in the design of P6 that prevents a simple $src.cow = False, then I'd like to know about it :)
21:33 particle rakudo: class foo { method bar { 1 } }; class foo { method bar { 1 } }; class foo { method bar { 1 } };
21:33 polyglotbot OUTPUT[Class foo already registered!␤␤current instr.: '!meta_create' pc 17206 (src/builtins/guts.pir:391)␤called from Sub 'parrot;foo;_block31' pc 203 (EVAL_17:99)␤called from Sub 'parrot;PCT;HLLCompiler;evalpmc' pc 888 (src/PCT/HLLCompiler.pir:494)␤called from Sub 'parrot;PCT;HLLCompiler;compile'
21:33 polyglotbot ..pc 428 (src/PCT/HLLCompiler.pir:301)␤called from...
21:33 lathos TimToady: I'm using P6 to prototype bits of Parrot; the C code will probably use a flags bitfield.
21:34 TimToady just seems like a premature lack of abstraction  :)
21:35 lathos The closer I keep it looking like reasonable C, the less chance I have of making a complete hash of it when I come to implementation... (Particularly when the encoding handling kicks in.)
21:35 TimToady *nod*, not really trying to get in your way...
21:36 TimToady but you can theoretically do named bitfields in C too...
21:36 lathos Yes, but then all the other Parrot committers break my fingers.
21:36 particle where has theory ever gotten us....
21:37 cognominal ils oublient l'olpc qui est en train  de se microsoftiiser à  fond la caisse.
21:37 TimToady .oO(the sins of the fathers are visited unto the tenth generation)
21:37 DietCoke lathos: we're still made at you for r3, is it really going to get worse? =-)
21:37 DietCoke *mad
21:38 cognominal oops
21:38 cognominal wrong channel again
21:47 Tene lots of trailing spaces in r3!
21:49 jonathan Yeah, but then we didn't have a *!&#@~% test for them!
21:49 jonathan Them were the days... ;-)
21:52 * jonathan points at r9 ;-)
21:52 particle timtoady: have you met robert gentleman?
21:53 particle (co-creator of R)
21:53 DietCoke jonathan: ... that's cold.
22:02 * jonathan had never read the earliest commits before
22:03 DietCoke lathos++ for all that early work.
22:04 jonathan Indeed.
22:04 * jonathan hadn't realized lathos had written the initial chunk of stuff until this point.
22:04 lathos Seven years and I'm *still* trying to get the string API correct.
22:05 jonathan persistence++ :-)
22:05 lathos Dan and I made the initial pot of stone soup between us.
22:05 * jonathan is trying to find the first patch where PMCs appeared, just out of curiosity.
22:05 DietCoke jonathan: no, freeze/thaw still don't work.
22:06 DietCoke oh, wait.
22:08 jonathan I didn't discover and dig into Parrot until 2003.
22:08 jonathan And then took 2 years to earn my commit bit. :-)
22:08 lathos jonathan: r356.
22:09 * jonathan waits for parrotvm.org...
22:10 lathos We had PMC structures before that but that's the first time they start behaving in a PMC-like way.
22:10 * barney asked http://isitchristmas.com/
22:15 jonathan Interesting...
22:16 DietCoke so that has nothing to do with perl6?
22:16 jonathan Not sure.
22:16 DietCoke wish github would fix their JS errors. :P
22:16 TimToady particle: no, I don't know that I have.  I know of Mr Gentleman from the Read or Die anime though....
22:17 particle probably not the same guy :)
22:17 Whiteknight joined #parrot
22:17 particle he's at the hutch here in seattle, thought i might pick his brain (or that of a nearby lackie) wrt R on parrot
22:18 chromatic Read or Die!
22:18 purl read or die is not hentai!
22:19 donaldh joined #parrot
22:19 jonathan purl++
22:20 * jonathan afk for a bit...got concentration span of about 10 minutes today. :-|
22:27 TiMBuS joined #parrot
22:31 alvar joined #parrot
22:33 cotto alphabet?
22:33 purl somebody said alphabet was overrated anway or http://www.underconsideration.c​om/speakup/archives/002383.html
22:33 cotto alphabet is also http://www.splashhall.org/poetry​_forums/index.php?topic=14076.0
22:33 purl okay, cotto.
22:33 shorten cotto's url is at http://xrl.us/beccgs
22:52 Whiteknight cotto's url?
22:52 Whiteknight purl cotto's url?
22:52 purl whiteknight: bugger all, i dunno
22:57 Theory joined #parrot
23:02 dalek r35459 | moritz++ | trunk/languages/perl6 (2 files):
23:02 dalek : [rakudo] implement Str.trim, patch courtesy by Ovid++
23:02 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35459
23:10 kj joined #parrot
23:11 kj joined #parrot
23:12 particle1 joined #parrot
23:14 dalek r35460 | kjs++ | trunk/compilers/pirc/src (5 files):
23:14 dalek : [pirc] commen out all line noise and add infrastructure to run the code after compilation. Use: ./pirc -x -b foo.pir. -x for 'execute after compile' and -b for 'compile to bytecode'. Multiple :main flags does not work correctly: it should run the last :main marked sub. It doesn't.
23:14 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35460
23:20 Limbic_Region joined #parrot
23:46 dalek r35461 | kjs++ | trunk (8 files):
23:46 dalek : [pirc] remove all tests, and replace them currently with a placeholder. All tests must be run with -x -b options, so that PIRC tests will actually be real PBC running.
23:46 dalek : + fix some POD errors.
23:46 dalek : [lib] change commandline options for running PIRC tests.
23:46 dalek review: http://www.parrotvm.org/svn​/parrot/revision?rev=35461

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

Parrot | source cross referenced