Camelia, the Perl 6 bug

IRC log for #parrot, 2010-10-27

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:01 kid51 joined #parrot
00:02 lucian_ left #parrot
00:03 chromatic joined #parrot
00:05 dmalcolm left #parrot
00:05 chromatic There is some argument for the expedience of merging soon and cleaning up later, as long as that later is not a far-off later.
00:06 nwellnhof Break things early...
00:08 plobsing this component has suffered enough in the name of expedience. parameter types added because they were convenient, runtime framebuilder removed because it was blocking the new pcc, etc...
00:09 plobsing it's time to do it *right*
00:10 plobsing and after a lot of banging my head on it, I think I see the way to do it.
00:11 plobsing not to mention, the only thing this buys us is exploratory programming. You can get any arbitrary signature already for extensions.
00:14 chromatic How much work is it to do right?
00:16 plobsing unknown, but also unknown how much to do it wrong.
00:16 plobsing tracking the bug has proved hard for me. splitting things up bit by bit until it is obvious will find the bug.
00:20 chromatic Fair enough.
00:20 * kid51 ->laundry
00:26 kid51 left #parrot
00:28 chromatic left #parrot
00:28 bacek left #parrot
01:37 whiteknight left #parrot
01:37 nwellnhof left #parrot
01:48 dngor joined #parrot
02:21 kid51 joined #parrot
02:25 kid51 left #parrot
05:14 tcurtis left #parrot
05:18 dalek parrot: r49692 | plobsing++ | branches/gsoc_nci/lib/Parrot/Co​nfigure/Options/Conf/Shared.pm:
05:18 dalek parrot: add without-libffi as legal option to configure
05:18 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49692/
05:19 dukeleto plobsing++ # hacking on gsoc_nci
05:20 dukeleto plobsing: let me know if you need that branch smoked
05:21 plobsing not even close. I'm making progress though.
05:24 AzureStone left #parrot
06:25 fperrad joined #parrot
06:26 uniejo joined #parrot
06:33 theory left #parrot
06:41 plobsing left #parrot
06:41 plobsing joined #parrot
06:43 chromatic joined #parrot
07:33 bacek joined #parrot
07:39 chromatic left #parrot
07:49 bacek aloha, humans
07:57 cotto aloha, bacek
07:59 bacek cotto, hi
08:00 wagle left #parrot
08:00 wagle joined #parrot
08:01 particle1 left #parrot
08:07 dalek parrot: r49693 | plobsing++ | branches/gsoc_nci (53 files):
08:07 dalek parrot: split NCI signature parser out from nci.pmc
08:07 dalek parrot: this will allow multiple NCI backends (static compiler, static runtime, libffi) to share the same parser
08:07 dalek parrot: also miscelaneous fixes to get fulltest mostly working (svn props mostly)
08:07 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49693/
08:40 cotto plobsing, do the packfile PMCs' get_pointer and set_pointer VTABLE functions get exercise anywhere?
08:44 cotto or more precisely, what kind of code would exercise that?  Would it only be C code that uses the packfile PMCs?
08:49 uniejo left #parrot
08:58 * dukeleto recently learned about the Vala language, which seems interesting: https://secure.wikimedia.org/wikipedia/​en/wiki/Vala_%28programming_language%29
08:59 bacek cotto, "C code" doesn't use Packfile PMCs (yet?)
08:59 cotto Right.  I'm wondering if that's the only way those VTABLE functions would be used.
09:00 bacek ah. Then yes.
09:00 bacek set_pointer used during unpack
09:01 dukeleto http://people.gnome.org/~aruiz​/videos/gtk-kickstart-vala.ogg <-- Writing a gtk app with vala
09:01 bacek get_pointer for pack
09:02 bacek afk # dinner
09:06 AzureStone joined #parrot
09:09 dalek parrot: r49694 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
09:09 dalek parrot: seal root_objects back after processing and moving into particular
09:09 dalek parrot: generation.
09:09 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49694/
09:27 cotto I finally get it.  The Packfile PMC calls get_pointer on its PackfileDirectory PMC, which recursively calls get_pointer on all its segments, and when it's done a valid PackFile struct has been created.
09:28 cotto so get_pointer is always used during pack and the pack/unpack METHODs seem to be unused
09:29 bacek cotto, .pack/.unpack used from PIR
09:29 bacek just because .get_pointer/.set_pointer isn't available :)
09:29 cotto in make_hello_pbc, get_string is used
09:29 cotto either way, it's essentially a different spelling
09:30 cotto get_string and pack use get_pointer internally
09:30 bacek yes
09:31 cotto It's so nice to finally have a coherent understanding of how that whole mess works.
09:41 particle joined #parrot
09:41 cotto hio particle
09:51 bacek cotto, it's not so messy! I've implemented it two years ago :)
10:01 mikehh dalek is not picking up on trac wiki changes any more
10:03 mikehh anyway, I emailed the list re GCI and have worked on the wiki page, but need more ideas
10:08 * cotto checks svn log
10:11 mikehh http://trac.parrot.org/parro​t/wiki/GoogleCodeIn2010Tasks
10:13 tadzik joined #parrot
10:13 tadzik left #parrot
10:14 tadzik joined #parrot
10:14 Tene I'll try to contribute to that tomorrow.
11:11 dngor left #parrot
11:12 dngor joined #parrot
11:17 dngor_ joined #parrot
11:20 dngor left #parrot
11:38 nopaste left #parrot
11:40 tadzik left #parrot
11:40 contingencyplan left #parrot
11:45 nopaste joined #parrot
12:36 ascent left #parrot
12:43 whiteknight joined #parrot
12:57 ascent joined #parrot
13:00 whiteknight good morning, #parrot
13:04 donaldh joined #parrot
13:04 whiteknight Question: how big/complex should a Google Code-In project be? We don't want them to be too huge, but not too small either
13:08 whiteknight One great project I can think about is for students who have had linear algebra to write up some tests for some under-tested operations
13:08 whiteknight but how many tests to write that would constitute a complete "task" is a mystery
13:16 particle whiteknight: iirc there's a "getting things done" definition of a task that says if it takes you more than four hours to do something your "task" isn't granular enough, so split it up into smaller tasks.
13:17 particle i wonder if there exists a more complete set of rules for defining tasks
13:18 particle and if we can find it, and if it'll apply.
13:23 whiteknight That's not a bad definition. I just want to make sure the  "tasks" that I define are suitable for the Google Code-In kids
13:24 particle how are the tasks defined, and marked as completed?
13:25 particle i'm not too knowledgeable about the code-in specifics.
13:25 particle but taskwarrior is a floss tool
13:26 particle http://taskwarrior.org/projects/show/taskwarrior
13:26 whiteknight There is pretty sparse information right now about how Google Code-In is structured and what the scope of a task should be
13:26 whiteknight Students are expected to complete at least one, and at most 15, tasks during the summer
13:27 particle are they paid, per task?
13:27 whiteknight Tshirt for the first task, 100$ for every 3 tasks, up to 500$ total
13:27 particle i understand there's a competition to complete as many tasks as possible.
13:27 particle gotcha.
13:28 particle let me see if i can scare up a high-school cs instructor to ask about task sizing
13:29 particle when are task lists due?
13:37 whiteknight Friday
13:37 whiteknight Actually, that's when our application is due. I don't know that all tasks need to be in by then
13:40 whiteknight + Tasks should be suitable for 13-18yr olds! They will not have the
13:40 whiteknight attention span or speed of a frequent contributor.
13:40 whiteknight + Tasks should have time limits, and ideally, should not take more
13:40 whiteknight than 2 ~ 3 weeks
13:40 whiteknight + Tasks should be classified as easy / medium / hard
13:40 whiteknight + Tasks should be very specific, rather than open-ended, so that
13:40 whiteknight (that's from an FAQ I found)
13:40 particle ok then
13:41 particle i'd call them projects, at that length, but whatever.
13:42 particle for a linear algebra task, it should detail the types of success, failure, and error conditions to test for, and give an example for one function.
13:43 particle perhaps it should detail every condition to test for.
13:48 bluescreen joined #parrot
13:49 whiteknight Well, I'm going to put together a list of tests and information about test coverage that I am interested in for different classes and methods
13:49 whiteknight so it's a matter of asking how many tests, or how much coverage gets completed before we declare the particular task a success, and give the kid his prizeee
13:56 cosimo left #parrot
13:57 cosimo joined #parrot
14:08 tadzik joined #parrot
14:18 davidfetter joined #parrot
14:20 Andy joined #parrot
14:33 tadzik left #parrot
14:39 cotto left #parrot
14:47 dmalcolm joined #parrot
14:48 tadzik joined #parrot
14:54 davidfetter left #parrot
15:13 silug left #parrot
15:18 dukeleto oooh, people are talking about tasks
15:18 dukeleto supposedly, we should have tasks in each category
15:19 whiteknight okay, that's fine. What I can't figure out is how large a task should be
15:19 whiteknight I would love to assign some PLA testing tasks (write tests, verify tasks, etc), but don't know how many tests constitutes an acceptable task
15:19 dukeleto whiteknight: a task can be as small as you want
15:19 dukeleto whiteknight: just make something up
15:19 dukeleto whiteknight: err on the side of smaller
15:20 dukeleto whiteknight: students can complete as many tasks as they want
15:20 dukeleto whiteknight: so it is in our interest to make them slightly too small, than too big
15:20 dukeleto whiteknight: make sense?
15:20 dukeleto whiteknight: assume that a "task" is about 10% of a GSoC proposal, or less
15:22 dukeleto i think some translation tasks would be good, like "translate our website to spanish/german/chinese/whatever"
15:23 dukeleto Task #2 on our wiki is a good start, but is WAY to ambiguous. It needs to list specific tests in specific files to convert
15:23 dukeleto too ambiguous, even
15:27 dukeleto mikehh++ # email to the list about GCI
15:27 dukeleto Just so everybody knows, we are going to use that page for Parrot, Perl 6 and Perl 5 tasks, please spread the word!
15:27 dukeleto I will try to in a bit.
15:28 patspam joined #parrot
15:44 mikehh dukeleto: I am looking to break it down into multiple tasks, checking which of out tests can be converted, 1 - easy, 2 - medium, 3 - hard etc
15:45 mikehh dukeleto: not that easy a task :-}
15:55 theory joined #parrot
16:08 dukeleto hola
16:10 tadzik left #parrot
16:11 dngor_ left #parrot
16:21 contingencyplan joined #parrot
16:35 whiteknight hello duke
16:57 davidfetter joined #parrot
16:58 dukeleto what's new?
16:59 davidfetter hc/lambda
16:59 davidfetter oh, wait. that's nu. nm.
17:04 chromatic joined #parrot
17:05 nwellnhof joined #parrot
17:07 dngor joined #parrot
17:08 donaldh left #parrot
17:14 dukeleto What's Mu pussycat, whoa a whoa a whoa...
17:14 * dukeleto is delirious from 8 days of conferencing
17:16 davidfetter heh
17:18 silug joined #parrot
17:30 nwellnhof Anyone knows how ext/nqp-rx/src/stage0/HLL-s0.pir gets built?
17:38 dukeleto nwellnhof: it is generated by nqp-rx
17:38 dukeleto nwellnhof: is your question about how it is built or how it is generated?
17:39 whiteknight dukeleto: you can complain all day, but you're still a lucky sack. I can't go to any conferences because nobody wants to go to philadelphia
17:39 whiteknight plus, I don't have enough vacation
17:42 dukeleto whiteknight: you are thinking backwards
17:42 tadzik joined #parrot
17:42 dukeleto whiteknight: and i am only complaing that my body can't take 8 days straight of conferencing, not the conferencing itself.
17:43 nwellnhof dukeleto: How can I regenerate it?
17:49 particle regenerate your body?
17:49 particle good question.... :P
17:51 particle nwellnhof: check the nqp-rx repo... http://github.com/perl6/nqp-rx/blob/master/README
17:54 radu_ joined #parrot
18:02 radu_ left #parrot
18:08 cotto joined #parrot
18:10 cotto ~~
18:19 Tene whiteknight: I went to philadelphia for a SF con twice, about six years back.
18:29 nopaste "nwellnhof" at 192.168.1.3 pasted "Remove deprecated charset ops in nqp-rx" (26 lines) at http://nopaste.snit.ch/24966
18:29 chromatic Six years ago?  whiteknight was what, 11 years old?
18:29 whiteknight you think I'm 17?
18:29 whiteknight my fake ID says I'm at least....about 34ish
18:30 atrodo if it's on the fake ID, it's got to be true
18:30 whiteknight and I'm from Hawaii, and my name is "McLovin"
18:30 nwellnhof Who can apply the patch above to nqp-rx?
18:31 dukeleto nwellnhof: pmichaud should review it first
18:32 dukeleto nwellnhof: i can apply it, but I would want his OK first
18:33 dukeleto nwellnhof: is there an easy-to-write test to verify that your removal of that code doesn't break things?
18:33 dukeleto nwellnhof: he would probably be more open to the patch with a test case
18:34 nwellnhof dukeleto: I can have a look.
18:44 dukeleto nwellnhof: ask pmichaud first if he likes it. If he doesn't, a test may change his mind.
18:44 dukeleto pmichaud: ping?
18:45 particle nwellnhof: nqp-rx is an external tool that parrot uses.  submit your patch there, and the changes will be incorporated into parrot.
18:53 pmichaud dukeleto: pong
18:53 pmichaud ouch, that patch looks.... bad.
18:54 pmichaud Is there no longer a mechanism to transcode a string?
18:55 nwellnhof pmichaud: of course there is. have a look at trans_encoding in the lines below.
18:55 pmichaud oh
18:55 pmichaud (those lines don't show up in the nopaste/patch)
18:57 pmichaud okay, patch looks fine to me.
18:59 lucian joined #parrot
19:00 pmichaud nwellnhof: want a commit bit so you can apply it directly? or should I just apply it?
19:02 nwellnhof pmichaud: please just apply it if you think it's ok.
19:03 pmichaud applying + testing
19:05 Andy left #parrot
19:09 dalek nqp-rx/master: d2d1682 | pmichaud++ | src/cheats/hll-compiler.pir:
19:09 dalek nqp-rx/master: From Nick Wellnhofer <wellnhofer@aevum.de>: Remove deprecated charset ops
19:09 dalek nqp-rx/master: review: http://github.com/perl6/nqp-rx/commit/d​2d1682d49e9091adfe98dba3cecd3491263d793
19:16 nwellnhof pmichaud++
19:16 nwellnhof pmichaud: Can I simply copy the stage0 files from nqp-rx to parrot now?
19:18 nwellnhof pmichaud: I think the bootstrap files have to be updated in nqp-rx first.
19:24 whiteknight something something something the test suite. Something something something complete
19:30 whiteknight_ joined #parrot
19:34 whiteknight left #parrot
19:34 whiteknight_ is now known as whiteknight
19:52 pmichaud I didn't update the bootstrap yet, no.
19:53 pmichaud I can do that.
19:53 nwellnhof pmichaud: No need to rush it.
19:57 dalek nqp-rx/master: 56b142f | pmichaud++ | src/stage0/ (3 files):
19:57 dalek nqp-rx/master: Update bootstrap files.
19:57 dalek nqp-rx/master: review: http://github.com/perl6/nqp-rx/commit/5​6b142f59e5a39e715569781b2571fd8078633e4
20:00 dukeleto http://leto.net/dukeleto.pl/2010/10/gsoc-2010​-mentor-summit-and-git-together-wrap-up.html
20:05 theory left #parrot
20:13 lucian left #parrot
20:21 NotFound Uh... someone has ever used String exists_keyed vtable?
20:21 NotFound Looks wrong.
20:21 NotFound return (INTVAL)((k >= 0 && k <= n) || (k < 0 && -k <= n));
20:21 NotFound <= ???
20:27 whiteknight left #parrot
20:28 nwellnhof Yes, the first <= looks wrong.
20:29 NotFound You want a task for students? Here is one: full coverage of the String PMC.
20:30 lucian joined #parrot
20:31 cotto or any PMC with <x% coverage
20:32 NotFound Some may be too difficult for beginners. String should be easy, but maybe boring.
20:38 cotto It shouldn't be too much effort to figure out a list of recommended PMCs.
20:39 bacek Good morning, humans
20:39 cotto hio bacek
20:40 bacek hello cotto
20:40 * cotto realizes that he's not living too far from Australian time
20:41 davidfetter cotto, well, depends what you mean by "not too far"
20:41 davidfetter ...and which part of .au
20:42 NotFound Less than 24 hours?
20:44 davidfetter SELECT now() - now() at time zone 'Australia/Sydney' AS "Sydney", now() - now() AT TIME ZONE 'Australia/Perth' AS "Perth";
20:44 davidfetter Sydney   |   Perth
20:44 davidfetter -----------+-----------
20:44 davidfetter -18:00:00 | -15:00:00
20:48 dalek parrot: r49695 | NotFound++ | trunk/src/pmc/string.pmc:
20:48 dalek parrot: fix String exists_keyed for positive index
20:48 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49695/
20:56 cotto NotFound, no test?
20:56 NotFound cotto: working on it, I prefer one commit for each thing.
20:57 cotto gotcha
20:58 cotto NotFound++
21:05 dalek parrot: r49696 | nwellnhof++ | trunk/ext/nqp-rx/src/stage0 (3 files):
21:05 dalek parrot: [ext] Update nqp-rx bootstrap files
21:05 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49696/
21:05 dalek parrot: r49697 | nwellnhof++ | trunk/compilers/pct/src/PCT/HLLCompiler.pir:
21:05 dalek parrot: [pct] Remove deprecated charset ops in PCT::HLLCompiler
21:05 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49697/
21:05 dalek parrot: r49698 | NotFound++ | trunk/t/pmc/string.t:
21:05 dalek parrot: tests for String exists_keyed vtable
21:05 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49698/
21:10 theory joined #parrot
21:11 theory left #parrot
21:11 theory joined #parrot
21:19 donaldh joined #parrot
21:20 Eclesia joined #parrot
21:21 Eclesia hello
21:22 cotto hi Eclesia
21:22 Eclesia I just discover parrot, and have a few questions if I may ?
21:22 tadzik hello. Sure, go on
21:22 cotto you may have questions.  You might even get answers. ;)
21:23 Eclesia about the license first.
21:23 cotto Artistic 2.0
21:23 Eclesia yes, but I don't know much about this one. what is special about it ?
21:24 Eclesia why not usual gpl / lgpl or apache ?
21:24 cotto http://en.wikipedia.org/wiki/Comp​arison_of_free_software_licenses
21:25 fperrad left #parrot
21:25 cotto It's probably an artifact of Parrot's history with the Perl community, where Artistic 2.0 is popular
21:25 Eclesia looks like a lgpl
21:26 Eclesia I also seen that it handles several langages, including java and .net
21:26 Eclesia how far does it handle it ?
21:26 Eclesia (for java)
21:27 Eclesia (sorry if my english is not correct)
21:27 * davidfetter generally uses BSDL or increasingly MITL
21:27 cotto there are a number of language implementations at varying stages of maturity
21:27 * davidfetter finds the restrictive licenses full of BS
21:27 cotto http://trac.parrot.org/parrot/wiki/Languages
21:27 cotto davidfetter, there's a rich diversity of views on that topic
21:27 nwellnhof Eclesia: .net is not a language
21:28 davidfetter yep. i've got mine ;)
21:28 Eclesia C#, asp.net ...
21:28 davidfetter i'm not trying to evangelize them
21:29 Eclesia nice langage list :)
21:29 cotto We "just" need more stable implementations.
21:31 Eclesia I usualy work with Java and netbeans ide. you know if there is a way to integrate the parot compiler ?
21:31 tadzik I'm missing Pynie progress. It's supposed to be a bridge to PyPy these days, is it?
21:32 chromatic The other way around, I believe: Pynie can be a backend to PyPy.
21:33 chromatic Eclesia, we intend for people to embed Parrot in IDEs.  Padre does it, but the embedding system is a work in progress.
21:33 tadzik hmm. So PyPy using Parrot, and Pynie being a full-featured Python3 compiler, yep?
21:33 cotto Eclesia, I don't know of any Java implementations on Parrot that I'd want to use other than to work on the implementations themselves.
21:33 chromatic tadzik, correct as per my understanding
21:36 NotFound I think there is no java compiler, but a project to interpret JVM bytecode with parrot, isn't it?
21:36 GeJ Bonjour everyone.
21:37 tadzik Bonjour
21:37 nwellnhof I don't think there's any bytecode translator for Parrot that's beyond the "Hello world" stage.
21:38 dngor_ joined #parrot
21:38 dngor left #parrot
21:39 dalek parrot: r49699 | nwellnhof++ | trunk/compilers/opsc/src/Ops/File.pm:
21:39 dalek parrot: [opsc] Fix typo in Ops::File
21:39 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49699/
21:39 tadzik left #parrot
21:40 cotto nwellnhof, nice fix.  It's kinda funny that that one survived so long, though I can see why it did.
21:41 nwellnhof cotto: I wonder why it wasn't noticed during the 2.9.1 release.
21:43 chromatic Oh, don't ask about 2.9.1 and why global destruction didn't work.
21:44 Andy joined #parrot
21:45 Eclesia left #parrot
21:55 dalek parrot: r49700 | nwellnhof++ | branches/remove_charset_ops:
21:55 dalek parrot: Remove charset ops
21:55 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49700/
21:55 dalek parrot: r49701 | nwellnhof++ | branches/remove_charset_ops (8 files):
21:55 dalek parrot: [ops] Remove charset ops
21:55 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49701/
22:04 bluescreen left #parrot
22:06 bluescreen joined #parrot
22:07 theory left #parrot
22:08 theory joined #parrot
22:17 dngor_ is now known as dngor
22:20 bacek_at_work ~~
22:20 cotto ~~
22:29 dalek parrot: r49702 | plobsing++ | branches/gsoc_nci (2 files):
22:29 dalek parrot: remove private2 (raw) flag from nci-ffi.pmc
22:29 dalek parrot: (removed in trunk after branch start)
22:29 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49702/
22:37 cognominal left #parrot
22:39 dukeleto plobsing++ # giving gsoc_nci some love
22:39 chromatic ping atrodo
22:46 donaldh left #parrot
22:55 dngor left #parrot
22:55 dngor joined #parrot
23:00 dalek parrot: r49703 | cotto++ | branches/opmap_aware_pmcs/src/pmc/packfile.pmc:
23:00 dalek parrot: [pmc] alignment cleanup; no functional changes
23:00 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49703/
23:00 dalek parrot: r49704 | cotto++ | branches/opmap_aware_pmcs/​src/pmc/packfileopmap.pmc:
23:00 dalek parrot: [pmc] implement size calculation in get_pointer; filling in the data is nyi
23:00 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49704/
23:14 whiteknight joined #parrot
23:16 nwellnhof left #parrot
23:19 simon_ left #parrot
23:22 dngor left #parrot
23:27 simon_ joined #parrot
23:39 whiteknight left #parrot
23:48 simon_ left #parrot
23:54 dngor joined #parrot

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

Parrot | source cross referenced