Camelia, the Perl 6 bug

IRC log for #parrot, 2010-08-30

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:02 Psyche^ joined #parrot
00:03 kid51 GeJ:  coming right up
00:06 dalek parrot: r48720 | jkeenan++ | trunk/t/codingstd/c_operator.t:
00:06 dalek parrot: Applying patch submitted by GeJ++.  While this does not display numbers of lines violating the standard, it does clearly display the lines themselves, thereby facilitating corrections.
00:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48720/
00:08 Patterner left #parrot
00:08 Psyche^ is now known as Patterner
00:13 GeJ kid51++
00:13 GeJ Thanks.
00:17 kid51 Paul_the_Greek: In your everyday, non-beach existence, are you able to run Parrot smoke tests on Win32?
00:17 Paul_the_Greek kid51: I have not tried it. Should I try it in the next day or two?
00:17 kid51 Well, let me explain why I'm asking ...
00:18 kid51 1. We are always short of testers on various Windows platforms.
00:18 kid51 2. Until we get our own Smolder server, we are whistling in the dark re much of our work, such as Win32.
00:19 kid51 3. I'm looking over old Trac tickets and noting that many of them are stalled because the patch needs testing on Win32.
00:19 kid51 So we need both regular smoke testing on Windows, as well as folks who can evaluate particular patches.
00:20 kid51 (or, at least, *test* particular patches)
00:20 kid51 speaking of which ...
00:20 kid51 whiteknight: How's your Win32 access these days?
00:21 whiteknight kid51: spotty. I can get on if needed. What's up?
00:21 kid51 see backscroll re need for testers on Windows
00:30 Paul_the_Greek I'd be happy to smoke test and test patches on Windows. I'm running 32-bit XP.
00:31 GeJ GeJ--
00:31 GeJ for begging commiters to apply a patch on coding standards while the patch actually violates said standards.
00:34 Paul_the_Greek kid51: Are the instructions here correct? http://trac.parrot.org/parrot/wiki/SmokingParrot
00:35 whiteknight kid51: I'm busy trying to get Paul_The_Greek commit access, and probably going to bed shortly thereafter. If you could msg me the details about what's needed I will do it tomorrow
00:39 dalek TT #684 closed by jkeenan++: pgegrep needs tests
00:39 dalek TT #684: http://trac.parrot.org/parrot/ticket/684
00:55 kid51 Paul_the_Greek:  Those instructions would be correct *if* our Smolder system were working ... which it hasn't been for many weeks now.
00:55 kid51 It was donated to us; it wasn't running on Parrot.org.
00:56 kid51 I read that we are trying to get our own Smolder set up.
00:56 kid51 but, in meantime, it's mainly a question of running 'make test' on as many platforms as possible, as regularly as possible (preferably >=1/day)
00:57 kid51 And ISTR you saying you were working on a Win32 box.
00:57 Paul_the_Greek Yes, a 32-bit XP system.
00:58 kid51 So, as a first step, could you set up a cron job (or whatever it's called on windows), doing a fresh checkout, configure, make, make test once a day
00:59 kid51 (If and when we got a Smolder setup again, you would simply change that 'make test' to 'make smoke')
00:59 Paul_the_Greek Sure can. I'll start that tomorrow or Tuesday.
00:59 kid51 Paul_the_Greek++
01:00 kid51 Sooner or later, we need to go thru the open TTs and see which have comments to the effect of "If we could get this patch tested on Windows, we could close this ticket."
01:12 Coke whatever was done with paul's svn commit, his login email looks nothing like that of the other commiters.
01:12 aloha bacek asked me to tell you aloha now delivers messages on any event.
01:12 aloha Paul_the_Greek asked me to tell you I tried to commit something today, but it appears I haven't been given commit privilege. Could you set that up?
01:13 dalek parrot: r48721 | Paul C. Anagnostopoulos++ | trunk (5 files):
01:13 dalek parrot: find_lex no longer throws exception on undefined name
01:13 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48721/
01:13 dalek parrot: r48722 | jkeenan++ | trunk/t/codingstd/c_operator.t:
01:13 Paul_the_Greek Coke: whiteknight has fixed me all up. I was a developer but not a committer. I just committed a change.
01:13 dalek parrot: [codingstd] Eliminate trailing whitespace.
01:13 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48722/
01:13 dalek parrot: r48723 | jkeenan++ | trunk/t/codingstd/c_todo.t:
01:13 dalek parrot: Have test output tell us how many TODOs, etc., remain.
01:13 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48723/
01:13 Coke *(&^#$*() ok, the dual aloha-purl is now annoying. what's holding up just using aloha?
01:14 Coke Paul_the_Greek: you're not all set yet, as you may have noticed your commit email hasn't shown up on list.
01:14 Coke (unless you're not subscribed to it, in which case case you wouldn't have noticed. )
01:14 whiteknight that isn't automatic?
01:14 Coke no, he has to be whitelisted.
01:15 Coke and his email came from "paul@osuosl.org", while everyone else's is from foo@svn.parrot.org
01:16 Coke From: Paul@osuosl.org, C.Anagnostopoulos@svn.parrot.org
01:16 Coke lemme guess. Space?
01:16 Coke *sigh*. Ah well, I'll just whitelist whatever the address came through as and hope for the best.
01:16 Coke done.
01:19 whiteknight okay. Coke++
01:19 whiteknight and now that all is right with the universe again, I'm going to sleep
01:19 Paul_the_Greek Interesting. Let's see how it goes.
01:20 Paul_the_Greek whiteknight++
01:20 whiteknight goodnight
01:20 whiteknight left #parrot
01:22 dngor_ is now known as dngor
01:24 dalek rakudo: 2ee5725 | colomon++ | src/core/Any-list.pm:
01:24 dalek rakudo: Escape { } in postcircumfix:<{ }> not defined message.
01:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​ee57258294084fbd47ea8f9fff871489d738fd6
01:26 kid51 is now known as kid51_at_dinner
01:27 edenc joined #parrot
01:29 tetragon joined #parrot
01:29 dalek TT #1207 closed by Paul_the_Greek++: Non-existent lexical throws exception
01:29 dalek TT #1207: http://trac.parrot.org/parrot/ticket/1207
01:33 mikehh_ joined #parrot
01:36 mikehh left #parrot
01:46 dalek TT #481 closed by Paul_the_Greek++: some basic math ops are missing
01:46 dalek TT #481: http://trac.parrot.org/parrot/ticket/481
01:55 edenc left #parrot
01:57 Paul_the_Greek left #parrot
02:16 kid51_at_dinner is now known as kid51
02:30 bacek_at_work Coke, aloha isn't too reliable atm. Just because it's on my laptop at home. I have to find some kind of "hosting" for her.
02:35 janus left #parrot
02:41 janus joined #parrot
02:44 kid51 left #parrot
02:52 tcurtis joined #parrot
02:53 theory joined #parrot
02:54 Topic for #parrot is now Parrot 2.7.0 "Australian King" Released! | parrot.org Log:  irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 20 tickets (-7 to go), merge outstanding branches; parrot-members: vote by 2PM PT Tues Aug 31
03:16 tetragon left #parrot
03:17 edenc joined #parrot
03:28 dafrito cotto, ping
03:47 cotto dafrito, pong
03:57 dafrito hows the plugin comin? I got back from my trip today and skimmed the commits
04:04 cotto not bad
04:05 cotto still some features needed, but I've given up on the idea of providing a stub implementation of a repository interface
04:06 cotto I don't feel like implementing 23 functions just to make the browser link show up when there's a plugin that does it already.
04:34 sorear Who was it who was working on getting rid of strstart?
04:35 cotto sounds like something darbelo would be working on
04:35 cotto I thought he got it.
04:35 sorear yes, but it didn't work, and I just realized why
04:36 cotto ?
04:36 sorear take a look at src/PAST/Compiler-Regex.pir lines 137-151
04:36 sorear in the nqp repository
05:01 Topic for #parrot is now Parrot 2.7.0 "Australian King" Released! | parrot.org Log:  irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 20 tickets (-8 to go), merge outstanding branches; parrot-members: vote by 2PM PT Tues Aug 31
05:03 theory left #parrot
05:09 dalek TT #1014 closed by jimmy++: NQP: Add defined(), exists(), undef.
05:09 dalek TT #1014: http://trac.parrot.org/parrot/ticket/1014
06:11 bacek_at_work left #parrot
06:13 uniejo joined #parrot
06:14 patspam left #parrot
06:19 baest_ is now known as baest
06:48 fperrad joined #parrot
07:05 antivigilante joined #parrot
07:05 antivigilante i downloaded the source for 2.6.0
07:06 sorear ...and?
07:06 antivigilante the Squaak tutorial says I should build parrot but I already have parrot installed
07:07 sorear then skip that step
07:07 antivigilante there is already a Squaak directory but there is no src/parsers
07:07 sorear tcurtis: ping
07:08 antivigilante so do I just clear that and start from scratch?
07:08 tcurtis sorear: pong
07:08 antivigilante or just use a diff name for my tutorial
07:12 antivigilante can parrot do modal languages?
07:15 tcurtis What do you mean by "modal"?
07:17 antivigilante for example this is my format (var func value) - if you leave any of them out interesting things happen
07:17 sorear tcurtis: you did the squaak tutorial, right?
07:17 antivigilante like (var nil value) is an assignment
07:18 tcurtis sorear: I updated it to use NQP-rx, yes.
07:18 antivigilante (var func nil) is a query
07:18 antivigilante (nil func value) is a test
07:20 antivigilante (nil nil nil) is the same as saying break in a loop
07:21 tcurtis I still don't think I fully understand what you mean, but I think it should certainly be doable in Parrot.
07:22 tcurtis Also, antivigilante, where are you reading the Squaak tutorial at? It shouldn't mention src/parsers any longer.
07:25 antivigilante i'm reading it at parrotblog
07:25 antivigilante anyway by modal I mean that different forms imply different types of statements
07:27 mikehh_ is now known as mikehh
07:27 mikehh opbots, names
07:28 tcurtis antivigilante: The version of the tutorial on parrotblog is quite outdated. http://docs.parrot.org/parrot/​latest/html/PCT_Tutorial.html is the up-to-date version.
07:29 antivigilante ok
07:29 NotFound squaak?
07:29 purl squaak is examples/languages/squaak or the language in kj's tutorial at parrotblog.org
07:30 tcurtis purl: squaak is also http://docs.parrot.org/parrot/​latest/html/PCT_Tutorial.html
07:30 purl okay, tcurtis.
07:30 tcurtis squaak?
07:30 purl squaak is examples/languages/squaak or the language in kj's tutorial at parrotblog.org or http://docs.parrot.org/parrot/​latest/html/PCT_Tutorial.html
07:31 antivigilante the thing about this little language is that AND and OR are implied by the structure of the program
07:32 antivigilante statements that go horizontally across the screen imply and
07:32 antivigilante vertically it's or
07:32 antivigilante NOT is done via (nil nil nil)
07:33 antivigilante with that if else can also be done
07:34 tcurtis That might be tricky to parse, but it could be done.
07:34 antivigilante looking to see how easy it is to do it in: parrot, lisp, javascript and assembler
07:36 sorear anything you can do in assembler you can do just as easily in javascript or parrot
07:36 antivigilante right - basically i'm gonna cheat a little by using the final bytecode as guide for the assembler version
07:37 antivigilante the point is to see if it's possible to reasonably boot into this from disc and just go nuts with the computer's resources
07:39 antivigilante should be interesting
07:39 dalek parrot: r48724 | NotFound++ | trunk/t/pmc/exception.t:
07:39 dalek parrot: some more Exception tests
07:39 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48724/
07:39 sorear antivigilante: I recommend using C, then
07:40 antivigilante thx guys - i'll bother you again if I hit a snag
07:40 sorear Parrot's parsing engine does a lot of stuff which will drive you mad if you try to copy it in assembly
07:40 sorear s/Parrot/NQP-rx/
07:40 sorear so.. don't start with NQP-rx
07:40 sorear start at a lower level
07:41 antivigilante well i am going toward something that goes past APIs - high level conceptual programming with low level muscle
07:42 antivigilante C is certainly something to use but I am going as fas even creating a new executable format
07:42 antivigilante s/fas/far as/
07:43 antivigilante anyhoo I'll be back I'm syre
07:43 antivigilante s/syre/sure
07:44 sorear seen bubaflub
07:44 purl bubaflub was last seen on #parrot 13 days, 12 hours, 25 minutes and 23 seconds ago, saying: hey GSoC-ers, pencils-down time has already passed and the final student evaluations are on Melange now  [Aug 16 19:18:51 2010]
07:44 aloha Sorry, I haven't seen bubaflub.
07:44 sorear bubaflub was supposed to have Parrot running on RTEMS earlier
07:44 sorear dunno if he succeeded
08:12 antivigilante what I ultimately want to achieve is a system where language specific bindings are no longer necessary
08:18 mikehh All tests PASS (pre/post-config, make corevm/make coretest, test, fulltest) at r48724 - Ubuntu 10.04 amd64 (g++)
08:25 tcurtis Good night, #parrot.
08:25 tcurtis left #parrot
09:19 dalek winxed: r613 | NotFound++ | trunk/winxedst0.cpp:
09:19 dalek winxed: backport new indexed to stage 0
09:19 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=613
09:51 mikehh_ joined #parrot
09:54 mikehh left #parrot
10:03 smash joined #parrot
10:04 smash hello everyone
11:09 dngor left #parrot
11:09 dngor joined #parrot
11:27 dalek rakudo: 6ccdcda | colomon++ | src/core/ (2 files):
11:27 dalek rakudo: Add Cool.Real, Cool.Int, Cool.Rat, and Cool.Num.  Eliminate Str.Num and Str.Int.
11:27 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​ccdcdae3b14ae24bec09d0980016ea0469e33d9
11:50 Paul_the_Greek joined #parrot
11:50 Paul_the_Greek Morning, folks.
11:55 antivigilante left #parrot
12:06 Paul_the_Greek left #parrot
12:16 whiteknight joined #parrot
12:34 moritz morning
12:38 szbalint good $localtime! :)
12:52 whiteknight good morning moritz
12:53 moritz good morning mr. knight
12:53 smash moritz: mornin'
12:55 masak joined #parrot
13:27 mkt joined #parrot
13:28 mkt parrot \o/
13:31 * masak squaaks
13:53 davidfetter joined #parrot
13:56 particle1 joined #parrot
13:59 particle left #parrot
14:01 smash particle1: thank you for checking mailboxes
14:01 particle1 is now known as particle
14:01 particle my pleasure :)
14:02 davidfetter left #parrot
14:03 bluescreen joined #parrot
14:07 davidfetter joined #parrot
14:21 uniejo left #parrot
14:24 lucian joined #parrot
14:27 Coke left #parrot
14:36 mikehh_ is now known as mikehh
14:37 mikehh opbots, names
15:00 davidfetter left #parrot
15:09 patspam joined #parrot
15:09 Coke joined #parrot
15:15 cotto_work ~~
15:16 theory joined #parrot
15:19 ruoso left #parrot
15:25 nwellnhof joined #parrot
15:30 dalek parrot: r48725 | nwellnhof++ | trunk (2 files):
15:30 dalek parrot: Fix old size in realloc
15:30 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48725/
15:30 dalek parrot: r48726 | nwellnhof++ | trunk/src/hash.c:
15:30 dalek parrot: Fix r48687
15:30 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48726/
15:34 patspam left #parrot
15:37 tcurtis joined #parrot
15:40 masak nwellnhof: ping
15:41 cotto_work pbc_dump gives valid output for large pbc files now.  Not knowing for certain who fixed it and being too lazy to bisect, I blame nwellnhof++
15:42 moritz it seems nwellnhof++ is to blame for a lot of good stuff recently :-)
15:42 nwellnhof pong
15:43 masak nwellnhof: oh hai. about #1746.
15:44 nwellnhof yes
15:44 masak nwellnhof: first, thanks for your excellent detective work.
15:44 masak it feels good to have the reason for the segfault.
15:44 masak second, what actually needs be done now to solve it?
15:45 masak I re-read the backlog with the discussion between nwellnhof, NotFound and pmichaud. but I didn't grasp what needs fixing, and whether it's in Rakudo or Parrot.
15:45 nwellnhof masak: the reblessing code in rakudo has to be fixed
15:45 masak ah.
15:45 nwellnhof well, ideally, parrot should provide a rebless function
15:46 masak :)
15:46 NotFound masak: I've not found any problem with attrs yet.
15:46 masak NotFound: have you been able to reproduce the segfault?
15:46 nwellnhof but the easiest solution right now would be to change rakudo's rebless code
15:47 nwellnhof it has to reallocate the PMC data when reblessing
15:47 NotFound masak: yes, following the instructions in the ticket it segfaults for me in linux i386
15:47 moritz is there a simple reallocate call?
15:47 masak so, it's rebless_subclass in src/ops/perl6.ops?
15:47 moritz or does it have to allocate, copy, change some pointers, deallocate?
15:47 NotFound nwellnhof: the Undef PMC has a sort of reblessing that seems to work
15:47 masak it says it's in-place.
15:48 nwellnhof moritz: i think not. but allocate/free should be straight-forward
15:48 nwellnhof to be more precise, the PMC attributes have to be reallocated
15:49 moritz at least to those familiar with the parrot memory API :-)
15:49 moritz nwellnhof: could we maybe coax you into writing a patch for it? :-)
15:50 NotFound The problem is that auto_attrs PMCs needs that allocation and deallocations match
15:50 lucian left #parrot
15:50 NotFound If you allocate in some other way and the PMC auto dealocates another, boom.
15:50 masak nwellnhof: there might be $beverage-of-your-choice involved at the next meatspace opportunity :)
15:50 nwellnhof I think it should be enough to call Parrot_gc_allocate_pmc_attributes, then memcpy, then Parrot_gc_free_pmc_attributes
15:51 nwellnhof And making sure the GC functions are called with the right vtable pointer in the PMC
15:51 NotFound nwellnhof: it should, yes. But that can fail with manual_attrs PMCs that does its own way.
15:52 nwellnhof NotFound: how can i tell if a PMC has auto_attrs or manual?
15:53 NotFound nwellnhof checking: if the size field in the vtable struct is not zero
15:53 nwellnhof NotFound: then rebless should reallocate only when the size field it not zero
15:54 NotFound And what to do when zero? Throw?
15:54 nwellnhof no, this case should simply work without reallocating
15:55 moritz (can the case occure that the parent class uses auto_attrs, and the child class that is reblessed to uses manual_attrs? or the other way round?)
15:56 NotFound moritz: I don't know any such usage, but sure is possible.
15:56 nwellnhof moritz: i don't know but that would make things more complicated
15:57 nwellnhof but i think a derived class must keep the attrs setting of its base class.
15:58 NotFound The intention is that all PMC get converted to auto_attrs, and use manual only for very special purposes, but I don't know if all HLL have make progress in that direction.
15:59 nwellnhof masak: i can try to whip up a patch for rakudo
15:59 masak \o/
15:59 integral left #parrot
16:00 nwellnhof and we should try to get the reblessing code into parrot
16:00 moritz NotFound: at least rakudo has only one manual_attrs left
16:00 * NotFound thinks about forbidding using manual_attrs when inheriting  from an auto_attrs PMC
16:00 integral joined #parrot
16:00 NotFound moritz: good
16:01 tcurtis NotFound: why not the reverse?
16:01 moritz that's perl6multisub, and probably the most complex of the rakudo PMCs
16:01 NotFound tcurtis: maybe both, yes
16:04 cotto_work http://github.com/resistor/BrainFTracing - trace-based JIT for bf
16:06 moritz perl6multisub is manual_attrs and inherits from multisub, which is auto_attrs
16:07 atrodo cotto_work> Awesome and scary at the same time
16:08 moritz that's the hallmark of any great piece of technology :-)
16:10 ash_ left #parrot
16:20 ruoso joined #parrot
16:24 masak left #parrot
16:28 nwellnhof masak: ping
16:30 nwellnhof pmichaud: ping
16:33 Andy joined #parrot
16:34 moritz anything rakudo related I might be able to help with?
16:34 nwellnhof moritz: I think the solution to the rebless_class problem is actually very simple
16:35 moritz that would be really great
16:35 nwellnhof we don't to need change the vtable at all
16:36 nwellnhof i'm not sure i fully understand the code, but i think the proxy PMC is also useless
16:36 BillyOregon joined #parrot
16:36 BillyOregon is now known as singalong
16:36 darbelo left #parrot
16:37 whiteknight nwellnhof: I would love to hear that proxy PMC was truly useless, but I suspect we cannot kill it yet
16:37 nwellnhof i only mean the use of a proxy PMC in Rakudo's rebless_class function
16:37 whiteknight ah, okay. more likely, less awesome
16:38 singalong I was following episode-3 of the PCT Tutorial and I got this error: http://pastie.org/1126892
16:38 nopaste "nwellnhof" at 192.168.1.3 pasted "Fix Rakudo's rebless_class" (32 lines) at http://nopaste.snit.ch/23194
16:39 whiteknight singalong: line 37. Change <statementlist> to <statement_list>
16:39 nwellnhof Here's my proposed patch
16:39 whiteknight (that damned tutorial! We cannot keep it current!)
16:39 nwellnhof Rakudo 'make test' passes
16:40 moritz I'll do a spectest
16:40 singalong whiteknight: thanks!
16:40 whiteknight Grammar.pm, by the way. I should have said
16:40 moritz chances are that 'make test' doesn't even call it once
16:40 singalong whiteknight: can the tutorial be edited in any way? there's a typo too some where.
16:40 whiteknight singalong: no problem. Sorry it didn't work for you the first time
16:40 whiteknight singalong: yes, the tutorial can be edited. If you make a patch, we will apply it
16:40 tcurtis singalong: I think that particular problem has already been updated in the repository, but I'll check.
16:41 whiteknight or if you open a bug report, somebody will get to it eventually
16:41 singalong whiteknight: ah no. it was something i saw in the generated language i think. i remember now.
16:41 singalong It was PST instead of PAST in the Actions.pm file
16:42 whiteknight oh, okay
16:43 singalong or maybe i made a mistake when i pressed keys when watching tv.
16:45 singalong I got another error. but this time I'm able to figure out what the error is about http://pastie.org/1126929
16:45 whiteknight oi
16:46 moritz might be token stat_or_def recursing
16:47 lucian joined #parrot
16:48 singalong moritz: stat_of_def recursing?
16:48 singalong *or
16:49 moritz it's a token or regex that apears twice in the backtrace
16:49 moritz seeing more of the backtrace would help
16:50 singalong moritz: what else can I paste?
16:51 moritz output of
16:51 moritz /usr/local/bin/parrot squaak/squaak.pbc t/00-sanity.t
16:52 tcurtis Odd. stat_or_def shouldn't be recursing.
16:53 tcurtis singalong: can you paste your grammar so far?
16:53 singalong tcurtis: sure.
16:54 mkt left #parrot
16:54 singalong one sec. my router seems to be kicking me off.
16:55 singalong moritz: http://pastie.org/1126947 the output for squaak.pbc
16:56 moritz yep, 989 recursive calls of stat_or_def
16:58 singalong moritz: how can I fix that?
16:58 moritz singalong: what's your definition of stat_or_def ?
16:58 singalong my code for actions and grammar http://pastie.org/1126947
16:59 moritz rule stat_or_def { <stat_or_def>
16:59 moritz }
16:59 tcurtis singalong: the body of rule stat_or_def should be <statement>.
16:59 singalong moritz: ah fund out I guess
16:59 dalek TT #1760 created by Paul_the_Greek++: Spurious newline in version string
16:59 dalek TT #1760: http://trac.parrot.org/parrot/ticket/1760
17:01 * moritz -> afk
17:02 singalong tcurtis: *found
17:03 tcurtis singalong: let me know if you have any other problems or if anything is unclear in the tutorial.
17:05 singalong tcurtis: actually i got another error.
17:05 singalong tcurtis: http://pastie.org/1126978 last one please.
17:07 singalong I hate to hijack a channel with consecutive errors
17:07 tcurtis singalong: can you paste t/00-sanity.t? I suspect that the generated test file isn't valid Squaak (and am almost certain it isn't valid Squaak near the beginning of the tutorial).
17:07 singalong would be nice if the tutorial actually had states of code after each episode for download
17:08 singalong tcurtis: http://pastie.org/1126985
17:09 tcurtis not the result of running t/00-sanity.t, but the actual contents of t/00-sanity.t.
17:10 singalong tcurtis: updated the same pastie http://pastie.org/1126985
17:11 jsut_ joined #parrot
17:13 tcurtis That's the problem, yes. That test file is auto-generated and isn't valid Squaak. You could try copying the tests from examples/languages/squaak/t in Parrot's repository, but the tests files wouldn't parse at that point in the tutorial, either.
17:14 singalong tcurtis: oh. anyway to obtain states of code after each episode? that would be cool for everyone.
17:14 nwellnhof left #parrot
17:15 singalong tcurtis: actually i got an idea :) i'll try to strip the test files from the repo to suit the current episode. maybe i'll succeed
17:15 tcurtis singalong: unfortunately, not at the moment. I'll try to remember to add that sometime soon.
17:16 singalong tcurtis: of when i finish this i'll try to add them in compressed file
17:16 singalong *or
17:16 jsut left #parrot
17:16 tcurtis singalong: good luck. :) Although, I don't think Squaak as of Episode 3 supports function calls, so I don't know how you'd manage to do output.
17:17 singalong tcurtis: oh! k now i'm going to try some stunts to pass the test then. like cheat by using a blank test file
17:18 singalong using a blank file fails too :P
17:18 davidfetter joined #parrot
17:19 singalong isn't this error because of the line "Finally, remove the rules proto token statement_control, rule statement_control:sym<say>, and rule statement_control:sym<print>." in the episode-3?
17:23 tcurtis Well, those aren't the right syntax for Squaak's output. It seems simpler to remove them and replace them with the appropriate Squaak syntax (I think that's done in Episode 4, but I'm not absolutely certain) than to tweak the parser to support them temporarily.
17:24 singalong tcurtis: what should I actually be going for? PIR or PCT?
17:25 tcurtis singalong: What do you mean by "going for"?
17:25 singalong tcurtis: do things directly via PCT or PIR.
17:26 singalong coz the PCT tutorial seems broken. and no other upto-date tutorial available. But I'm going to try tomorrow to just read thru the docs and understand PCT before I do PIR
17:27 singalong pir seems to have a few tutorials available already on the net.
17:27 tcurtis What things are you trying to do? Writing a compiler? PCT. Writing some other code? Either PIR or some HLL.
17:27 tcurtis The PCT tutorial (hopefully) isn't broken. If it is, I'll try to fix that.
17:28 singalong tcurtis: just trying to implement a language. but wanted to go thru the PCT tutorial.
17:28 singalong tcurtis: just found this pretty good tutorial here http://coolnamehere.com/geek​ery/parrot/learn/index.html
17:28 singalong but its PIR :P
17:29 Coke singalong: I would recommend NOT using PIR directly. it is helpful to know when you have to drop down to PIR from NQP, but if you're starting from scratch, use PCT/NQP/etc.
17:30 Coke and yes, the current state of document leaves much to be desired.
17:30 Coke *documentation
17:30 singalong Coke: oh! then i should probably stick to PCT.
17:30 singalong Coke: Would love to help with the documentation.
17:30 Coke If you're looking for an example (as opposed to a tutorial), partcl-nqp might be of use.
17:31 Coke partcl-nqp?
17:31 purl i heard partcl-nqp was http://github.com/partcl/partcl-nqp
17:31 singalong Coke: ya sure! even an example would do.
17:31 Coke that's written mainly in nqp-rx, uses the PCT classes from there, and has some callouts to PIR where necessary.
17:32 Coke the main grammar is in src/Partcl/Grammar.pm, the actions for that grammar in Actions.pm in the same dir.
17:32 singalong Coke: thanks!
17:32 Coke other mini-languages in tcl have similar src/*/Grammar.pm,Action.pm pairings.
17:34 Coke Have to run, $DAYJOB
17:41 singalong left #parrot
17:42 mmcleric joined #parrot
17:48 ash_ joined #parrot
17:55 ash_ left #parrot
17:57 nwellnhof joined #parrot
17:59 ash__ joined #parrot
18:05 ash__ left #parrot
18:05 mmcleric left #parrot
18:15 whiteknight left #parrot
18:22 ash_ joined #parrot
18:35 pmichaud nwellnhof: pong
18:37 triper joined #parrot
18:39 triper left #parrot
18:40 moritz pmichaud: nwellnhof came up with a patch to fix the memory corruption that we discovered in conjunction with masak's enum patch
18:40 moritz pmichaud: I've applied it in the enum_memory branch
18:41 pmichaud moritz: okay, looking
18:43 pmichaud ummmm, -1 on the patch, I think.
18:43 nwellnhof moritz, pmichaud: forget my old patch
18:44 nwellnhof see the comment i just made: http://trac.parrot.org/par​rot/ticket/1746#comment:10
18:44 nwellnhof i didn't realize what was going on with the proxy object
18:44 pmichaud line 150 is the one that switches out the vtable?
18:45 nwellnhof yes
18:45 pmichaud I'm not sure we can just eliminate it.
18:45 pmichaud Unless it's fossil code, but jnthn++ would probably need to check that.
18:45 nwellnhof it operates on the *old* proxy PMC
18:45 nwellnhof which will be never used again
18:46 pmichaud looking
18:46 nwellnhof looks like a part from some old code that was accidentally left in
18:58 pmichaud if spectests pass with that line removed, then +1 from me
19:02 Paul_the_Greek joined #parrot
19:03 Paul_the_Greek Hey folks.
19:03 Paul_the_Greek Question about the Integer PMC?
19:04 whiteknight joined #parrot
19:05 dalek tracwiki: v14 | particle++ | TracSpammers
19:05 dalek tracwiki: http://trac.parrot.org/parrot/wiki/Tr​acSpammers?version=14&amp;action=diff
19:06 particle spammers all up in our drupal :(
19:06 Paul_the_Greek Wah!
19:08 cotto_work I thought Drupal was unsinkable, like the Titanic.
19:08 particle if any admins can help me delete the users and update Mollum about the spammers listed at https://www.parrot.org/admin/content/comment i'd be grateful
19:08 whiteknight IM IN UR DRUPAL SPAMMIN UR PARROTZ
19:08 whiteknight particle: I can help a little (though I am coming late and don't know what happened)
19:10 sorear Is EvanCarroll allowed on #perl6?
19:10 sorear He's banned from all of irc.perl.org for a reason
19:10 particle apparently he's allowed
19:10 sorear well, I haven't banned him yet
19:11 particle anyway, whiteknight i'm looking at the comments listed in that report
19:11 sorear I mean "allowed" in a social sense
19:11 particle if the comment is seo spam, i'm deleting the user
19:11 particle then i'll go back to the report, where the author is now listed as (not verified), and mark the comments for Mollum to pick up on in the future
19:12 particle mark the comment by clicking on the subject in the report, then hitting the 'report to mollum' link, and reporting as spam
19:29 whiteknight particle: I don't see any obvious spam on pages 2 or 3
19:31 particle yeah, i've started at the bottom :)
19:31 whiteknight oh, that's the idea I had, since I figured you would start at the top :)
19:32 particle for some reason, there are a lot of spam comments about parrot 1.7.0 created in the last few weeks
19:33 whiteknight popular release?
19:33 ash_ left #parrot
19:34 Paul_the_Greek Question about compiling .pmc files?
19:34 ash_ joined #parrot
19:34 ash_ left #parrot
19:35 cotto_work you have one?
19:35 Paul_the_Greek Heh. Yes, I do.
19:35 cotto_work There can be no answer without a question.
19:36 Paul_the_Greek If I have changed a .pmc file, what make do I have to do in order to run tests on it?
19:36 whiteknight there can be no answer until you realize the truth: There is no question
19:36 Paul_the_Greek Then please provide the answer without my bothering to ask.
19:36 whiteknight should be part of the normal make target
19:36 Paul_the_Greek The default make target?
19:37 whiteknight just type "make"
19:37 Paul_the_Greek Will do. Thanks. Working on integer.pmc.
19:37 cotto_work Once you've run make, you can run the direct tests via prove.  it's not as complete as make test but it'll point out any glaring issuess more quickly.
19:38 Paul_the_Greek Yes, I have new tests just for these changes, so a prove will do the trick, at least initially.
19:41 ash_ joined #parrot
19:43 particle i need a report of drupal users who have never posted and have no roles
19:45 particle many spam comments here: https://www.parrot.org/news/2009/Parrot-1.7.0 ... i'm dealing with them now. sigh.
19:53 cotto_work yech
19:58 Paul_the_Greek The Integer PMC does not implement the negate method. I assume, then, that it inherits it from Scalar?
19:59 cotto_work looks like it
19:59 purl No it doesn't, shut your hole!
19:59 ruoso left #parrot
20:00 cotto_work botsnackkick
20:00 purl thanks cotto_work :)
20:00 cotto_work karma botsnack
20:00 purl botsnack has neutral karma
20:00 aloha botsnack has karma of 0.
20:00 purl thanks aloha :)
20:00 Coke left #parrot
20:03 nwellnhof karma karma
20:03 purl karma has karma of 61
20:03 aloha karma has karma of 0.
20:03 purl has karma of 0. has neutral karma
20:03 aloha has karma of 61 has karma of 0.
20:04 Paul_the_Greek Do you think we can put those two bots in an infinite loop?
20:04 Paul_the_Greek karma purl aloha karma
20:04 purl purl aloha karma has neutral karma
20:04 cotto_work close but no cigar
20:04 aloha purl aloha karma has karma of 0.
20:04 purl aloha: what?
20:04 aloha purl: Search me, bub.
20:04 purl aloha: excuse me?
20:04 aloha purl: I give up.
20:04 purl aloha: excuse me?
20:04 aloha purl: I have no idea.
20:04 purl aloha: i'm not following you...
20:04 Paul_the_Greek That was pretty good.
20:04 nwellnhof i'm impressed
20:04 cotto_work Paul_the_Greek++
20:05 Paul_the_Greek karma purl aloha karma loop
20:05 purl purl aloha karma loop has neutral karma
20:05 aloha purl aloha karma loop has karma of 0.
20:05 purl aloha: sorry...
20:05 Paul_the_Greek karma purl karm aloha purl
20:05 purl purl karm aloha purl has neutral karma
20:05 aloha purl karm aloha purl has karma of 0.
20:05 purl aloha: excuse me?
20:05 aloha purl: No clue. Sorry.
20:05 purl aloha: excuse me?
20:05 aloha purl: No clue. Sorry.
20:05 purl aloha: i'm not following you...
20:05 moritz any reason not to kick purl?
20:05 whiteknight left #parrot
20:05 cotto_work This is a dangerous game we're playing.
20:06 cotto_work moritz: loss of karma state
20:06 moritz cotto_work: I plan to extract #parrot karma from the public IRC logs
20:07 cotto_work I'd prefer not to kick purl until everything's in place
20:07 moritz understandable
20:09 cotto_work but if some approximation of karma is preserved and the messages get less spammy, I won't mind
20:09 szbalint who runs purl anyway? hachi?
20:09 ash__ joined #parrot
20:09 szbalint couldn't you get a karma dump from him?
20:09 cotto_work owner?
20:10 purl i guess owner is hachi, see also #purl
20:10 ash_ left #parrot
20:10 ash__ is now known as ash_
20:12 Coke joined #parrot
20:17 dalek parrot: r48727 | nwellnhof++ | trunk/src/string/api.c:
20:17 dalek parrot: Fix a bug in Parrot_str_join
20:17 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48727/
20:30 tcurtis left #parrot
20:33 bluescreen left #parrot
20:42 lucian left #parrot
20:43 fperrad left #parrot
20:43 bluescreen joined #parrot
20:44 dduncan joined #parrot
20:44 dduncan left #parrot
20:44 p4rr0t joined #parrot
20:50 luben joined #parrot
20:53 moritz my #parrot logs have about 51k lines that contain ++ or --
20:55 Coke moritz++++++++++++++++++++++++++​+++++++++++++++++++++++++++++++
20:58 cotto_work2 joined #parrot
20:58 ash_ left #parrot
20:58 Coke (my consulting company)--
20:58 moritz postfix:< has karma +3 :-)
20:59 cotto_work2 left #parrot
20:59 cotto_work2 joined #parrot
21:01 ash_ joined #parrot
21:01 ash_ left #parrot
21:01 cotto_work left #parrot
21:01 cotto_work2 is now known as cotto_work
21:01 GeJ Bonjour everyone.
21:03 szbalint chromatic++ # profilic blogger
21:04 * davidfetter has been antilific lately
21:06 jsut joined #parrot
21:10 Coke http://thingsrbetterwithaparrott.typepad.com/
21:11 jsut_ left #parrot
21:19 cotto_work wedding invitations are serious business
21:20 Paul_the_Greek Alright, whose haribrained idea was it that the negative of Boolean true should be true?
21:21 * cotto_work looks and prepares for lulz
21:22 cotto_work looks like an unintended side-effect of Boolean's inheritance
21:22 Paul_the_Greek Yes, because it extends Integer. Which is an absolutely ridiculous result of decades of refusal to make Booleans first-class types.
21:24 Paul_the_Greek However, the problem is a result of changes I just made. Integer no longer inherits neg from Scalar.
21:25 cotto_work Are you working to rip Scalar out?
21:25 Paul_the_Greek No, I'm fixing the bugs neg minint and abs(minint).
21:26 Paul_the_Greek Let me look at Scalar neg ...
21:27 Paul_the_Greek Weird, it behaves exactly as you would expect. Where does the Boolean behavior come from? A little investigation needed...
21:28 ash_ joined #parrot
21:29 cotto_work My guess would be that it happens because non-zero evaluates to true.
21:30 p4rr0t left #parrot
21:31 Paul_the_Greek Ah, so it's whatever is generated by is() that I have to check.
21:31 dalek TT #1245 closed by Paul_the_Greek++: src/pmc/integer.pmc:  Check overflow for -maxint in absolute()
21:31 dalek TT #1245: http://trac.parrot.org/parrot/ticket/1245
21:31 Paul_the_Greek How can I see what a .t file is compiled into?
21:31 cotto_work opbots, names
21:31 Topic for #parrot is now Parrot 2.7.0 "Australian King" Released! | parrot.org Log:  irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 20 tickets (-9 to go), merge outstanding branches; parrot-members: vote by 2PM PT Tues Aug 31
21:32 cotto_work which .t file?
21:32 cotto_work Most of them are run (somewhat) directly.
21:33 Paul_the_Greek boolean.t
21:33 Paul_the_Greek I need to see what is() generates, so I can figure out how it's comparing the two Booleans.
21:35 Paul_the_Greek The weird thing is that the 2-operand form of neg is failing, but the 1-operand form is not.
21:36 ash_ left #parrot
21:38 ash_ joined #parrot
21:38 cotto_work That PIR is run directly.  The functions are defined in runtime/parrot/library/Test/More.pir .
21:40 seatek joined #parrot
21:40 Paul_the_Greek It uses iseq.
21:41 silug left #parrot
21:42 Paul_the_Greek Which does what you'd expect. This is mysterious.
21:42 hercynium joined #parrot
21:46 Paul_the_Greek I get it. It's the behavior of set_integer_native.
21:47 Paul_the_Greek Boolean simply has to implement neg.
21:47 bluescreen left #parrot
22:43 kid51 joined #parrot
22:45 nopaste "kid51" at 192.168.1.3 pasted "gsoc_branch: 'make fulltest' failures at r48727" (2233 lines) at http://nopaste.snit.ch/23199
22:51 cotto_work2 joined #parrot
22:52 cotto_work2 kid51: there is no gsoc_branch branch
22:54 cotto_work left #parrot
22:59 kid51 Correction: gsoc_threads
23:06 dalek parrot: r48728 | jkeenan++ | branches/gsoc_threads (7 files):
23:07 dalek parrot: [codingstd] Set Subversion metadata.
23:07 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48728/
23:07 dalek parrot: r48729 | jkeenan++ | branches/gsoc_threads (7 files):
23:07 dalek parrot: [codingstd] trailing space; pod syntax.
23:07 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48729/
23:07 dalek parrot: r48730 | jkeenan++ | branches/gsoc_threads/src/threads.c:
23:07 dalek parrot: [codingstd] no cuddled else
23:07 dalek parrot: review: http://trac.parrot.org/parrot/changeset/48730/
23:07 hercynium left #parrot
23:07 kid51 Lots more still to be done in gsoc_threads branch
23:07 kid51 is now known as kid51_at_dinner
23:13 particle left #parrot
23:25 cotto_work2 dafrito: ping
23:25 cotto_work2 is now known as cotto_work
23:25 bluescreen joined #parrot
23:30 dafrito cotto_work: pong
23:32 tcurtis joined #parrot
23:32 cotto_work dafrito: can you come up with a way to get git log to spit out every commit in a repository?
23:34 cotto_work It seems to be much faster than map-svn (and doesn't make as many assumptions) to use git log, but its output only has ~39000 revisions.
23:34 dafrito git rev-list --all is probably the right direction for that
23:36 dafrito That might be faster, but I don't think the assumptions are any better: It'd be a difference between using git-svn's data, or an artifact of that data
23:37 cotto_work rev-list only returns 40792 lines of output
23:37 whiteknight joined #parrot
23:37 kid51_at_dinner left #parrot
23:38 dngor_ joined #parrot
23:39 * dafrito shrugs
23:39 dngor left #parrot
23:39 dafrito The old map-svn used "git rev-list --all"
23:42 dafrito If you're not getting a complete view of the data, then your checkout might be incomplete? Both ways are hackish in that they have to assume existing data is actually useful
23:43 dafrito doing git rev-list --all on my repo gets me 84k lines of output
23:43 cotto_work is that parrot/parrot from github?
23:44 dafrito No, I'll check that one out and see what I get
23:47 dafrito I get 40784
23:47 silug joined #parrot
23:48 cotto_work same here (mostly)
23:51 nwellnhof left #parrot
23:56 dafrito which is probably accurate, considering there's been pruning and things like that going on
23:57 cotto_work I'm tempted to go with whatever rev-list spits out.
23:58 dafrito yeah, it doesn't bother me
23:58 dafrito rev-list is nice since it works on any git repo
23:58 dafrito map-svn only works on repos with SVN crap lying around
23:59 cotto_work The more important question is whether it'll cause omissions that bother parrot hackers using the trac site once the plugin installed.

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

Parrot | source cross referenced