Camelia, the Perl 6 bug

IRC log for #parrot, 2009-06-19

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:00 Tene r
00:00 Infinoid WYSIWYG--
00:00 Infinoid WYGIWYG++
00:01 Tene rakudo: say [~] gather take 'Infinoid++ ' for 1..5
00:01 polyglotbot OUTPUT[Infinoid++ Infinoid++ Infinoid++ Infinoid++ Infinoid++ ␤]
00:02 * Tene finally goes home
00:04 Infinoid take, huh
00:05 * Infinoid wishes there was a "perl6doc -f" command
00:05 Whiteknight pmichaud: ping
00:05 * Tene adds perl6doc to his task list.
00:05 dalek parrot: r39662 | whiteknight++ | branches/pdd09gc_part1:
00:05 dalek parrot: removing old branch that's woefully outdated and unusable
00:05 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39662/
00:05 dalek parrot: r39663 | whiteknight++ | branches/RELEASE_1_3_0:
00:05 dalek parrot: removing release-only branch
00:05 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39663/
00:06 Whiteknight msg pmichaud: is the isafast branch still useful and worth keeping?
00:06 purl Message for pmichaud stored.
00:07 Theory joined #parrot
00:12 cotto Whiteknight, should src/pmc/stringhandle.pmc +223 be checking the return value?
00:12 cotto It's producing a warning.
00:13 Whiteknight what is checking what return value?
00:14 Whiteknight Parrot_io_open?
00:15 cotto ./src/pmc/stringhandle.pmc:223: warning: ignoring return value of ‘Parrot_io_open’, declared with attribute warn_unused_result
00:15 Whiteknight ah, okay
00:15 Whiteknight I actually just created a branch to clean up some IO stuff, so I'll fix that
00:15 dalek parrot: r39664 | whiteknight++ | branches/io_cleanups:
00:15 dalek parrot: creating a branch to start cleaning up IO before YAPC
00:16 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39664/
00:17 Whiteknight Infinoid: Feel free to dig in on this one, I want to get sockets and pipes fixed up
00:22 dalek parrot: r39665 | whiteknight++ | branches/io_cleanups/src/io/api.c:
00:22 dalek parrot: [io_cleanups] remove WARN_UNUSED_RESULT from Parrot_io_open. the return value is not necessary for all IO PMC types. cotto++
00:22 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39665/
00:24 Infinoid Ok, you won't be getting any polyglot lolcode support today
00:24 Infinoid Can't open perl script "/usr/local/lib/parrot/1.3.0-de​vel/tools/dev/gen_makefile.pl": No such file or directory
00:24 Infinoid Just like most other HLLs.
00:35 sekimura_ joined #parrot
00:39 dalek parrot: r39666 | cotto++ | trunk/lib/Parrot/Pmc2c/PMCEmitter.pm:
00:39 dalek parrot: [pmc2c] make get_vtable etc only use functions from a PMC and its immediate parent
00:39 dalek parrot: This is another "pmcc will do it this way so make sure it works" patch.
00:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39666/
00:39 dalek parrot: r39667 | whiteknight++ | branches/io_cleanups/src/io/api.c:
00:39 dalek parrot: [io_cleanups] remove some unnecessary debugging assertions, and allow Parrot_io_open to create an HLLmapped PMC instead of just a vanilla FileHandle (not really supported currently)
00:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39667/
00:43 eternaleye joined #parrot
00:55 eternaleye joined #parrot
00:59 dalek parrot: r39668 | whiteknight++ | branches/io_cleanups (4 files):
00:59 dalek parrot: [io_cleanups] remove unused FileHandle ATTR file_size.
00:59 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39668/
01:02 nopaste "kid51" at 70.85.31.226 pasted "Why is Rakudo not building for me on Linux/i386?" (90 lines) at http://nopaste.snit.ch/16967
01:03 dalek parrot: r39669 | whiteknight++ | branches/io_cleanups/src/pmc/filehandle.pmc:
01:03 dalek parrot: [io_rewiring] remove unused process_id ATTR from FileHandle. It's not needed since FileHandle isn't Pipe anymore
01:03 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39669/
01:06 Whiteknight if we can get all non-inheritable ATTRs out of FileHandle, we can make it properly subclassable from PIR
01:06 Whiteknight and that's a major goal of mine
01:28 sekimura joined #parrot
01:29 TonyC joined #parrot
01:39 Whiteknight Infinoid: How do we do pipes now?
01:42 Whiteknight nevermind, I just TODO'd the tests
01:42 Whiteknight goodnight
01:43 dalek parrot: r39670 | whiteknight++ | branches/io_cleanups (2 files):
01:43 dalek parrot: [io_cleanups] change some things in src/io/filehandle.c to use the GETATTR and SETATTR macros for subclassability. Also, fudge some tests that were relying on FileHandle=Pipe behavior
01:43 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39670/
01:59 kurahaupo left #parrot
02:04 kurahaupo joined #parrot
02:22 Coke joined #parrot
02:23 kurahaupo joined #parrot
02:23 Coke wtf. I have an 'hll_map' getting invoked while in .HLL 'Tcl'; I verify it in the sub it gets added in, and it's there. If I then autobox something in that HLL, it's not working. (switching from "maps" in a PMC file to invoking 'hll_map' on a class)
02:23 Coke opbots, ping
02:23 clunker3 Pong!
02:24 Coke opbots, names
03:02 eternaleye joined #parrot
03:04 dukeleto joined #parrot
03:20 brbrooks joined #parrot
03:20 donaldh joined #parrot
03:44 dukeleto joined #parrot
03:50 Theory joined #parrot
04:02 pmichaud message kid51  Sent you email -- your error looks the same as RT #64930.  Try adding the --gen-parrot-option=--optimize option (see the ticket or the bottom of http://rakudo.org/how-to-get-rakudo for more details)
04:02 purl Message for kid51 stored.
04:50 Theory joined #parrot
04:59 japhb joined #parrot
05:02 mikehh pre/post config, smolder, fulltest ALL PASS at r39670 - Ubuntu 9.04 i386
05:43 mikehh joined #parrot
06:07 uniejo joined #parrot
06:14 uniejo joined #parrot
06:22 uniejo joined #parrot
06:33 whoppix joined #parrot
06:46 clunker3 joined #parrot
07:07 iblechbot joined #parrot
07:20 brbrooks joined #parrot
07:20 donaldh joined #parrot
08:06 muixirt joined #parrot
08:33 UltraDM joined #parrot
08:57 dukeleto joined #parrot
09:19 Infinoid pmichaud++ # rakudo #18
09:32 bacek_ joined #parrot
10:05 MoC joined #parrot
10:12 sparkymat joined #parrot
10:13 sparkymat left #parrot
10:55 clinton joined #parrot
11:20 brbrooks joined #parrot
11:21 donaldh joined #parrot
11:21 burmas joined #parrot
11:59 Whiteknight joined #parrot
12:01 dalek pynie: r72 | isop44++ | trunk/ (3 files):
12:01 dalek pynie: Implement is [not] operator, without chaining
12:01 dalek pynie: review: http://code.google.com/p/pynie/source/detail?r=72
12:06 dalek pynie: r73 | isop44++ | trunk/ (2 files):
12:06 dalek pynie: Implement type builtin function
12:06 dalek pynie: review: http://code.google.com/p/pynie/source/detail?r=73
12:06 dalek pynie: r74 | isop44++ | trunk/Lib/test/parrot/builtins.py:
12:06 dalek pynie: Remove xfail annotation for repr test - it passes now! :)
12:06 dalek pynie: review: http://code.google.com/p/pynie/source/detail?r=74
12:41 jhorwitz joined #parrot
12:58 Coke ... is .HLL now case sensitive?
12:59 Coke seems to be. wtf.
12:59 Coke perhaps in conjunction with .hll_map()
13:01 Coke tene, you around?
13:08 dalek pynie: r75 | isop44++ | trunk/ (8 files):
13:08 dalek pynie: Add simple str class and some tests
13:08 dalek pynie: review: http://code.google.com/p/pynie/source/detail?r=75
13:13 dalek pynie: r76 | isop44++ | trunk/Lib/test/parrot/literals.py:
13:13 dalek pynie: Mark commented-out tests with xfail
13:13 dalek pynie: review: http://code.google.com/p/pynie/source/detail?r=76
13:14 dalek pynie: r77 | isop44++ | trunk/ (4 files):
13:14 dalek pynie: Rename Pynie;list to just list
13:14 dalek pynie: review: http://code.google.com/p/pynie/source/detail?r=77
13:18 Whiteknight joined #parrot
13:22 Tene Coke: yes, it's case-sensitive, kinda
13:23 Tene Coke: it changes sensitivity in different places, which is now considered a bug, so the only sane solytion right now is to only use all lowercase
13:39 iblechbot joined #parrot
13:41 ilia joined #parrot
13:42 Coke Tene: is there a ticket for that? (If not, can you make sure one gets opened?)
13:43 Coke I would just say "let's never mangle the case."
13:44 mj41 joined #parrot
13:44 Tene Yes, that's the ideal situation.
13:45 Coke it just cost me god knows how much time fixing bugs, since it mattered to .hll_map()
13:45 Coke s/fixing/tripping over/
13:46 Coke (but not to pmclass's hll)
13:46 Coke Tene: let's get a deprecation notice in for 1.4 if we can.
13:47 Tene Me too, a while back.  I complained to allison, she agreed it should be fixed
13:48 Tene I don't remember if there's a ticket.
13:48 Coke there. partcl is now all lowercase.
13:50 Coke https://trac.parrot.org/parrot/search?q=​hll+case&noquickjump=1&ticket=on says no.
13:50 skids joined #parrot
13:50 Coke wishlist: a button on my trac search to exclude /closed/ tickets.
13:50 Tene iirc, I told you you'd need to do it for some HLL stuff to not suck.
13:51 Coke yup.
13:51 Coke Didn't realize it was for anything other than interp, though. =-)
13:51 Coke so, it's done now, anyway.
13:51 dalek partcl: r508 | coke++ | trunk/runtime/builtin/ (50 files):
13:51 dalek partcl: .HLL is /sometimes/ case sensitive; just always use lowercase.
13:51 dalek partcl: Tene++ for verifying I'm not crazy (about this).
13:51 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=508
13:59 kgilmer hello :)  I'm trying to build parrot, is this the right place to ask a question?
13:59 Coke kgilmer: absolutely.
14:00 kgilmer great, thanks Coke.  So i'm trying to build parrot for ARM via OpenEmbedded.
14:00 kgilmer i've written a recipe for OE for parrot
14:00 kgilmer it gets past configure and into compile
14:00 kgilmer let me pastebin the error log, sec
14:00 Coke nopaste?
14:00 clunker3 http://pasta.test-smoke.org/ or http://paste.husk.org/ or http://nopaste.snit.ch:8001/ or http://rafb.net/paste or http://poundperl.pastebin.com/ or http://paste.scsys.co.uk/
14:00 purl nopaste is at http://nopaste.snit.ch/ (ask TonyC for new channels) or http://poundperl.pastebin.com/ or http://paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at http://www.extpaste.com/ or http://paste.scsys.co.uk (for #catalyst, #dbix-class, #moose  and others) or http://gist.github.com/
14:02 kgilmer here is what configure says...nothing looks wrong here to me...  http://pasta.test-smoke.org/54
14:03 Coke yup, seems ok.
14:03 kgilmer and here's the end of the log with all the errors: http://pasta.test-smoke.org/55
14:04 kgilmer i can also provide the recipe
14:04 mtk joined #parrot
14:05 kgilmer here is the recipe...which simply shows what params i'm passing in from my build environment: http://pasta.test-smoke.org/56
14:05 mtk left #parrot
14:06 kgilmer hmm
14:06 kgilmer actually, why did i comment out the configure?
14:06 kgilmer ha
14:06 kgilmer sec let me try it again with the configure uncommented.  i stop/started this project a few times.
14:07 * kgilmer is rebuilding
14:08 kgilmer hmm yeah same error
14:09 kgilmer often i find linker errors result from something using my native compiler rather than my cross compiler
14:09 kgilmer any easy way i can check for that or other suggestions?
14:11 kgilmer from the log it does look as if it's calling my host perl rather than the one provided in OE.  im not sure though if this matters...
14:11 kgilmer | /usr/bin/perl5.10.0 -MExtUtils::Command -e mkpath blib/lib
14:11 MoC joined #parrot
14:11 Coke I am not sure anyone has done any cross compilation.
14:11 gryphon joined #parrot
14:12 Coke do you have ICU installed?
14:12 Coke perhaps on your original platform, but not on your target platform?
14:12 kgilmer probably not...as I don't know what ICU is.
14:13 Coke unicode library.
14:13 kgilmer my host is gentoo amd64
14:13 Coke your config says you have it.
14:13 Coke but if you are cross compiling...
14:13 kgilmer right...
14:13 purl ...
14:13 Coke try running perl Configure.pl --without-icu
14:13 kgilmer checking now
14:13 kgilmer ok
14:14 * kgilmer is trying that now after checking for ICU on host
14:14 Coke bah. I'm trying to convert a PMC to a have a PIR class wrapper. getting a segfault!
14:14 Coke on   $I0 = isa value, 'Undef'
14:14 * kgilmer confirms ICU 4.0.1 installed on host
14:16 skv joined #parrot
14:31 clinton joined #parrot
14:32 * Coke accidentally notices that alt-left is the same as ^P on irssi.
14:46 * kgilmer gets same error with --without-icu passed to configure
14:46 kgilmer Coke, would it be better to send my issue to ml since cross compiling is rare or unknown?
14:47 Coke yah; if that didn't help, ml is a better bet.
14:47 Coke also, try to realclean before you re-configure.
14:47 Coke 'make realclean'
14:47 purl well, 'make realclean' is defined inside of the Makefile.  Previously it was implemented with $(CONFIGURE), which translated to Configure.pl.  Now it is implemented with $(RECONFIGURE), which translates to 'perl tools/dev/reconfigure.pl'.  If you did 'svn up' first, you would have wiped out the Makefile and thereby wiped out my ability to say 'make realclean'.
14:47 Coke purl, forget 'make realclean'
14:47 purl Coke: I forgot 'make realclean'
14:48 Coke my segfault is occurring on line 1556 of src/string/api.c
14:48 kgilmer ok thanks Coke
14:52 Coke how do I invoke a vtable from inside gdb?
14:52 Coke (since the normal macro access doesn't work, IIRC)
14:52 ruoso joined #parrot
14:53 ruoso joined #parrot
14:55 ruoso joined #parrot
14:59 kgilmer after looking some more at my build script Coke I found my configure params we not actually being passed
14:59 Coke heh.
14:59 kgilmer now they are and Configure is choking on my compiler
14:59 kgilmer | inter::progs -        Determine what C compiler and linker to use...Compilation failed with 'ccache arm-angstrom-linux-gnueabi-gcc -march=armv5te -mtune=xscale -mthumb-interwork -mno-thumb'
15:00 Coke now that sounds like a cross compilation issue. =-)
15:00 kgilmer haha
15:00 Coke I'd hit the list now. =-)
15:00 kgilmer ok sounds good.  Thanks again Coke
15:04 dalek TT #776 created by coke++: segfault in Parrot_str_equal...
15:04 Theory joined #parrot
15:20 brbrooks joined #parrot
15:20 donaldh joined #parrot
15:28 Whiteknight Coke: pmc->vtable->add()
15:35 Andy joined #parrot
15:36 Coke Whiteknight: Danke. I will file that away.
15:36 Coke (in the meantime, I opened a ticket)
15:41 Tene Coke: did you still want me to file a ticket about HLL case munging?
15:41 Tene Or were you going to do it?
15:42 darbelo joined #parrot
16:17 Psyche^ joined #parrot
16:37 davidfetter joined #parrot
17:10 Debolaz joined #parrot
17:12 bacek_ joined #parrot
17:13 Coke adoughera++
17:13 Coke Tene: I figured you had more info than I did. =-)
17:13 Tene Oh, okay.
17:13 Coke I can certainly file the ticket, though.
17:13 Coke wammeto?
17:13 Tene Sure.
17:14 PacoLinux joined #parrot
17:17 Coke done
17:17 * Coke changes the default bug-version to 1.3.0
17:18 dalek TT #777 created by coke++: .HLL should not case-mangle its arg
17:23 dalek parrot: r39671 | coke++ | trunk/docs/project/release_manager_guide.pod:
17:23 dalek parrot: Just say you need admin rights all the way through for www.parrot.org
17:23 dalek parrot: (Yes, you can do some bits with less, but it'll just bite you later if you don't have them.)
17:23 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39671/
17:29 Coke anyone know what the "language" component is for?
17:29 Coke (on trac)
17:49 pmichaud hello all (from airport)
17:49 pmichaud message Whiteknight  isafast branch can be dropped.
17:49 purl Message for whiteknight stored.
17:52 Coke hello, airport.
17:52 Whiteknight thanks pmichaud
17:53 Coke Whiteknight: feel free to update BranchDescriptions on the wiki. =-)
18:08 mikehh joined #parrot
18:25 bobke joined #parrot
18:26 Theory joined #parrot
18:28 Coke Whiteknight: do we need l1bc?
18:28 Coke any reason not to compile l1 to pbc?
18:29 Whiteknight i'm envisioning compilation from PBC to L1BC
18:29 Whiteknight PBC will be table lookup for pre-compiled L1BC macros that get executed
18:30 dukeleto joined #parrot
18:30 Coke I boggle in confusion.
18:30 * Coke will read the next post.
18:38 athomason joined #parrot
18:40 muixirt Whiteknight, what is the difference introducing this L1 layer and directly targeting vms like, let's say, the jvm?
18:40 Whiteknight muixirt: the difference is that in this case, Parrot is the VM
18:43 muixirt a Parrot that somewhat shrinks to something like jvm bytecode
18:43 muixirt or to put it another way
18:44 muixirt what is the point of Parrot afterall?
18:44 Coke muixirt: I'd point you at the faq or at perhaps allison's recent article.
18:44 muixirt if the abstracton level of pbc is to high for real purposes
18:45 cotto joined #parrot
18:46 muixirt with that L1 thingy Parrot is someting like a runtime (ops and pmc) targeting something a simpler vm like the jvm
18:46 muixirt Coke, I read that and I do not comment it (don't want to be kicked :-)
18:50 muixirt (allisons article)
18:52 * muixirt probably misunderstood that L1 concept
18:54 MoC joined #parrot
19:01 Coke I think chromatic and Whiteknight both have different ideas about it. I'm waiting to see chromatic's alternative writeup.
19:01 Coke (but I'd rather see him make a profiler first)
19:02 Whiteknight Yeah, I really hope to hear a good version of chromatic's ideas too
19:02 Whiteknight think about PIR as "words" and L1 as the "Letters"
19:02 Whiteknight right now, Parrot has a complicated engine for executing all sorts of arbitrary words
19:02 Whiteknight we want to replace it with a smaller, faster system to execute the individual letters
19:03 Whiteknight input a word (PIR/PBC), get a stream of letters (L1) and execute those
19:04 Whiteknight The user writes code in easy-to-read words, and Parrot gets to use easy-to-execute letters
19:05 muixirt and these easy-to-execute letters differ from, again let's say, jvm bytecode?
19:06 Whiteknight yes, in the same way jvm bytecode differs from .net bytecode
19:06 Whiteknight different virtual machines operate on different bytecodes
19:06 Whiteknight but in a lot of cases, the fundamental operations will be different
19:07 muixirt so the Parrot vm is a Potemkin village with a much simpler L1 vm in the background?
19:07 Whiteknight Parrot is still focused on dynamically-typed languages, and it's fundamental operations will reflect that
19:07 muixirt different in what way
19:07 muixirt ?
19:08 Whiteknight different in the way that Perl is different from Jva
19:08 Whiteknight Java*
19:09 muixirt an evasive answer ;-)
19:09 JC1 joined #parrot
19:09 Whiteknight you're asking evasive questions
19:11 muixirt what can Parrot do better than the jvm which shows feasibilty of supporting dyn. languages?
19:12 Coke That's an excellent question that I've never seen a solid answer on.
19:13 Coke I think that allison might have a more technical answer for it;
19:20 donaldh joined #parrot
19:23 Coke I would recommend careful phrasing; that can be kind of a "poison" question if you're not careful. I think it's a perfectly valid question, but I don't think that anyone working on parrot is going to, for example, drop the project and go over the jvm.
19:23 Coke and I'd rather our technical people spent time improving parrot rather than defending the existence of the project. =-)
19:26 muixirt Coke, I don't question the existence of the project nor do I coax anyone to use the jvm instead of Parrot.
19:29 muixirt I question the assumptions
19:29 Tene muixirt: some of the significant reasons for Parrot are historic.  When Parrot was started, the jvm wasn't appropriate at all for a dynamic language, and was extremely closed.
19:30 Tene The dynamic stuff added to the JVM has been fairly recent.
19:30 Tene Where Parrot has been dynamic from the very beginning, because that was one of the goals.
19:30 Tene I don't know enough about the JVM to really speak about the current state, though.
19:35 Theory joined #parrot
19:36 Coke (historic) those were all handwavy at the time, to my recollection.
19:36 Coke They may have been backed by fact, but it's not like we had sample code or a use case or...
19:36 Tene I wasn't around at the time, so it's even more handwavy to me.
19:37 Tene For me, the value equation is "Parrot is fun".
19:38 Tene with some measure of "Perl 6 is fun, and rewarding to help with"
19:42 Whiteknight I'm not entirely current on what all features JVM offers, so I can't give a point-by-point comparison of it with Parrot
19:43 Coke <nod> It would certainly be nice to have a well-reasoned technical explanation about "why not mono" and "why not jvm".
19:44 Coke would make a good research project if anyone cared to do it.
19:45 Coke http://www.perlmonks.org/?node_id=142797 is an interesting read from back in the day.
19:45 Whiteknight Personally, my favorite aspect of parrot is that it's register-based, not stack-based
19:45 Whiteknight although that really doesn't have a bearing on it's usefulness for dynamic languages
19:45 Coke http://code.google.com/p/parrot-jvm/ is also interesting, but empty.
19:46 Coke Whiteknight: I think some claim it does.
19:46 Whiteknight register-based VMs have a numbe of benefits over stack-based ones, certainly, but I don't think either approach is *better* for dynamic languages
19:46 Whiteknight I
19:46 muixirt Whiteknight, the jvm  might simply nowadays be fast enough for these dynamically-typed languages with their chunky runtimes
19:46 Whiteknight I'd love to hear an alternate opinion
19:47 * Coke finds http://video.google.com/videop​lay?docid=2900106024749292774
19:47 * Coke is watching it now for the first time.
19:48 muixirt Coke I say that some weeks ago and wasn't quite convinced
19:49 Coke one of the issues is that the JVM today isn't the same as the JVM of 2001.
19:49 Coke (neither is the PVM)
19:52 Coke I really wish I could see the slides on this. =-)
19:52 Coke so far it's not P vs J, it's just "why P"
19:54 * Coke didn't realize it was 35 minutes long.
19:54 Coke Whiteknight: you should link to that video from www.parrot.org somewhere.
19:59 Whiteknight Coke: I'l throw a link up later, that's a good idea
19:59 Whiteknight I haven't seen that video in it's entirety myself, so I need to watch that.
20:00 Whiteknight Parrot has some pretty impressive built-in support for dynamic class introspection and reflection, which I don't know that a statically-typed VM will support as well
20:00 Whiteknight I don't know that it doesn't, but it would suprise me
20:01 Whiteknight Of course we could run a dynamic language on the JVM, but it won't be tuned to it in the same way Parrot is
20:02 Coke (tuned) I don't think we can  make that argument any time soon. =-)
20:02 Whiteknight in theory
20:02 purl hmmm... in theory is my pants or it's too dark to read or somewhat intrusive to David Wheeler
20:02 Whiteknight we obviously have a lot of development still to do, and a lot of optimization
20:04 Whiteknight but the current state of an actively-developed project isn't so important as it's long-term potential
20:04 Whiteknight Parrot also has a strong focus on interoperability, only .NET really comes close to it in terms of similar capability
20:04 Whiteknight JVM was never intended for things that weren't Java, even if some things have been targetd to it
20:12 Whiteknight at this very moment JVM is faster then PVM, no question
20:13 Whiteknight But there is a lot of potential for PVM in the future, and it will become quite a force to be reckoned with in the VM world
20:13 Whiteknight on top of that it's a great development project, I would work on it with this team even if it didn't have the same merit as a product
20:13 dukeleto joined #parrot
20:20 Coke she keeps saying parrot is significatntly faster. I've never seen that. =-)
20:20 Coke ah. raw PIR, not an HLL compiled to PIR.
20:26 * Coke wishes the segfault fairy would come and fix his latest blocker.
20:27 mj41 Jonathan, chromatic and Allison should write Parrot vs. JVM vs. .NET FAQ ... dynamic method dispatch, first class functions, first-class continuations, paramemeters (optiona, named, ..), register based
20:27 mj41 "Parrot is innovative and not just a .NET or JVM clone.", Jonathan Worthington, 2005
20:27 Coke devil's advocate: "Yes but how."
20:28 mj41 •.NET constrains high level semantics of languages to achieve interoperability. Parrot has interoperability provided at an assembly level – more later.  http://www.jnthn.net/papers​/2005-lpw-parrot-slides.pdf
20:28 mj41 range of platforms
20:29 mj41 faster than JVM because of Continuation Passing Scheme
20:29 mj41 for dynamic languages
20:30 mj41 I am only copy paste what I read and don't understand a lot :-).
20:32 mj41 Perl 5 has two big features that make using the JVM or .NET problematic--closures and polymorphic scalars. Perl 6 adds a third (which Ruby shares) in continuations, and a fourth (which Ruby doesn't) of co-routines." Dan Sugalski, March 25, 2003
20:33 Coke perhaps mj41 should collect these snippets and put them on a wiki page.
20:33 Coke (so we can forward people there.)
20:33 mj41 but now there are http://www.artima.com/lejava/a​rticles/dynamic_languages.html and http://en.wikipedia.org/wik​i/Dynamic_Language_Runtime
20:33 muixirt Coke, http://wiki.jvmlangsummit.c​om/pdf/38_Randal_parrot.pdf
20:33 mj41 so probably some of these arguments are history
20:41 mj41 http://www.perlmonks.org/?node_id=614832 ... MMD is another one
20:41 mj41 my English sucks. I can't create Wiki page :-(.
20:42 mj41 I read many articles http://perl6.cz/wiki/Perl_6_and_Parrot_links . I was thinking that all you folks know that JVM or .NET are not good enought for us :-).
20:43 Coke wrap all those up and send them to whiteknight, then.
20:45 mikehh I thought one of the main concepts of parrot vs other vm's was it is register based rather than stack based
20:45 mj41 I can probably quote our gurus, but we really need something fresh. ... 2007 Larry obviously if Ruby's running on .Net, they support continuations somehow
20:54 mj41 mikehh: IMHO, there are years of engineering work on Parrot to support dynamic languages. So, it is not only about register based vs. stack based. I hope :-).
20:54 mikehh there is a difference between "can do" ans "supports"
20:56 mj41 yes. In the end these are all engineering problems, no academic discussions.
20:56 mikehh in terms of .net and jvm - wheras parrot has been designed to support dynamic languages
20:58 mikehh of course one of parrots main problems is hat it is a volunteer project - it does not have paid teams developing it as .net and jvm did
20:58 mikehh although that is also one of its major advantages - it is just taking longer
20:58 purl okay, mikehh.
20:59 mj41 I don't believe there will be energy to support all what Perl 6 needs in JVM.
21:00 mj41 and comparable to Parrot. But as time go, we will see :-).
21:01 mikehh of course it could be done - but then who would want to
21:01 mikehh the parallel development of parrot and perl6 is exiting (I think anyway)
21:02 mikehh even the proof of concept work on pugs was good
21:03 muixirt exiting? :-)
21:03 mikehh exciting :-}
21:04 mj41 Yes. Volunteers. Dan Sugalski ... http://www.sidhe.org/~dan/​blog/archives/000151.html March 2003 ... the big reason we're going with Parrot rather than using the JVM or .NET is a cultural choice, rather than a technical one. :-)
21:06 mikehh perhaps - but I do think there were a lot of good technical reasond for the choices
21:07 mikehh dame this jeyboard - it just don't fit my fingers like the old one
21:08 mikehh as you can see keyboard
21:08 mikehh abd damn
21:08 mikehh and
21:08 mj41 "They're fully turing complete, so it's not an issue of capability." :-)
21:09 mj41 ... the issue is one of speed.
21:10 mikehh I read a paper once about a turing complete language that only had a subtraction and branch conditional operator
21:12 sekimura_ joined #parrot
21:14 mikehh parrot is in need of a lot of refactorings and optimizations - its getting better and better
21:18 mikehh about a year ago it took me about an hour to build and smoke parrot and a couple of hours to run the full test suite
21:18 sekimura joined #parrot
21:18 mikehh now I can do a build smoke and fulltest in about an hour
21:20 dukeleto joined #parrot
21:37 mj41 mikehh: HW upgrade? :-)
21:39 Tene Allison refers to contexts as GCable in that video. :)
21:46 mikehh no - I have been using a dual-core Pentium D system in that time
21:46 mikehh I am getting an upgrade begining July
21:56 Whiteknight joined #parrot
21:59 jdv79_ joined #parrot
21:59 jdv79_ make: *** Recursive variable `DYLD_LIBRARY_PATH' references itself (eventually).  Stop.
21:59 jdv79_ look familiar?
21:59 jdv79_ i was just following the steps at rakudo.org to get everything setup.
22:00 darbelo jdv79_: on what platform?
22:00 jdv79_ mactel
22:01 jdv79_ darwin is it called?
22:02 darbelo I call 'em "Intel Macs", but I'm not a mac user anyways :). I'm guessing you are on some version of OS X, right?
22:03 Zak joined #parrot
22:09 Infinoid hi jdv79_
22:10 Infinoid Is this error from rakudo or from parrot?
22:11 Infinoid Either way, can you stick your Makefile up on nopaste.snit.ch?
22:11 Infinoid I have a hunch it might be related to the ":=" vs "=" Makefile-debacle
22:12 Infinoid In answer to your question, no, it doesn't look familiar.  But I think Coke and kid51 are the real OSX experts, not me
22:33 rg1 joined #parrot
23:13 skids joined #parrot
23:17 Coke I'm using intel darwin on OS X 10.4, no problems.
23:20 donaldh joined #parrot
23:23 bacek_ joined #parrot
23:23 Infinoid Coke: Is there more than one available make program that could be confusing parrot?
23:23 Coke it's all gcc.
23:23 Coke "gnu"
23:24 Infinoid ok
23:24 Infinoid he would get an error like that if his Makefile was full of "=", rather than ":="
23:24 Infinoid which I think parrot does for non-gnu Makes
23:27 Coke I'd need more information about his setup.
23:28 Coke (args to configure, perl version, result of config...)
23:33 japhb joined #parrot
23:39 cotto Who else has made it to Pittsburgh so far?
23:44 dalek parrot: r39672 | cotto++ | branches/pmc_pct/compilers​/pmcc/src/emitter/pmc.pm:
23:44 dalek parrot: [pmcc] flesh out some supporting PMC initialization functions and add more to class_init's first pass
23:44 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39672/
23:58 jsut joined #parrot

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

Parrot | source cross referenced