Camelia, the Perl 6 bug

IRC log for #parrot, 2010-05-21

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 darbelo Use the PackFile PMCs to write a pbc_dumper replacement.
00:01 darbelo ...or a pbc_merge replacement.
00:01 darbelo ...or a pbc_dissasemble replacement.
00:02 Whiteknight "small" project
00:02 NotFound darbelo: I have both things in examples/packfile.winxed
00:03 NotFound Dump and disassembly.
00:03 bacek_at_work Use PackFile PMC to generate PBC from POST!
00:03 NotFound Whiteknight: pbc_checker is not so big.
00:05 darbelo NotFound++
00:05 Whiteknight I don't think I ever even heard of pbc_checker
00:05 darbelo NotFound: Have you considered getting a copy of that into the parrot repo?
00:05 NotFound examples/tools/pbc:checker.cpp
00:06 NotFound examples/tools/pbc_checker.cpp
00:06 ruoso joined #parrot
00:06 darbelo I meant the dissasembler/dumper. The C ones are a mess, I've wanted to deprecate them for a long time.
00:07 NotFound darbelo: we can put the generated pir, but I don't think is a good idea until winxed gets ready for plumage installing.
00:10 darbelo What would it take to get winxed plumage-ready?
00:13 NotFound darbelo: mainly to make my mind about file names and installation paths, to achieve simple usage both from command line and from load_language
00:14 NotFound Also, a decision about creating or not one more stage.
00:17 Whiteknight pbc_checker is too big to tackle tonight
00:18 NotFound Whiteknight: oh, you mean really small ;)
00:18 Whiteknight really, really small
00:18 Whiteknight I've got a kid vomitting on me, I don't have a lot of time to just focus
00:20 NotFound A text mode pacman using the curses module.
00:20 darbelo I'm told vomiting kids go well with compiler warning cleanups.
00:20 darbelo Last I checked clang was bitching about us reaching the end of non-void functions all over.
00:22 NotFound darbelo: probaly PARROT_DOES_NOT_RETURN must be adapted to it.
00:23 darbelo An great task to tackle with kid vomit on you!
00:23 darbelo I'd do it myself if I had any.
00:23 darbelo ;)
00:23 NotFound Even better if you know nothing of clang ;)
00:24 NotFound darbelo: are you already testing pir literals with ucs4 encoding?
00:25 darbelo Wait. Kid vomit is not an euphemism for compiler documentation?
00:25 NotFound I'll take a look at the branch tomorrow if I have time.
00:26 NotFound But tomorrow I'll go to http://imperialstormtroopers.com/ so maybe I don't have much time.
00:26 nopaste "darbelo" at 192.168.1.3 pasted "This all seems to work, so far." (163 lines) at http://nopaste.snit.ch/20608
00:27 NotFound darbelo: good!
00:28 darbelo NotFound: Any ideas on PIR literals tests?
00:30 darbelo I mean should I test anything else?
00:31 NotFound darbelo: I think we need a pir_output_is variant that compiles to pbc and runs the pbc.
00:31 darbelo Isn't that what testr does?
00:32 NotFound darbelo: Someone uses testr other than before a release?
00:33 darbelo Good point.
00:35 darbelo The nopaste works just fine when compiled to pbc or run with parrot -r, though.
00:35 NotFound I'm out of ideas, going to bed :O
00:35 darbelo So, I'm guessing your patch works :)
00:41 Whiteknight darbelo: there were some encoding/charset-related tests in t/pmc/namespace.t
00:41 Whiteknight mostly proving that namespaces with exotic literal names don't essplode
00:46 tcurtis joined #parrot
01:26 s1n joined #parrot
01:40 gbacon joined #parrot
01:46 dalek parrot: r46834 | tcurtis++ | branches/gsoc_past_optimization (7 files):
01:46 dalek parrot: Add small example and initial(seemingly working) prototype of PAST::Walker.
01:46 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46834/
02:11 jimk joined #parrot
02:26 JimmyZ joined #parrot
02:34 abqar joined #parrot
02:41 dukeleto tcurtis++
03:02 Andy Holy cow are the GCC 4.6 warnings sweeeeet
03:04 dukeleto src/thread.c:1398: failed assertion '!interpreter_array'
03:13 dukeleto why can't i make two interpreters?
03:15 dukeleto http://gist.github.com/408411 is very simple code that shows this issue
03:16 dukeleto The first interpreter in any process should be
03:16 dukeleto created with a NULL parent, and all subsequent interpreters in the same
03:16 dukeleto process should use the first interpreter as their parent.  Failure to do so
03:16 dukeleto may result in unpredictable errors.
03:16 dukeleto I would say it is pretty predictable. It fails.
03:19 cotto_work Should interp1 and interp2 be Parrot_Interp*?
03:25 * cotto_work goes home to check
03:26 dalek parrot: r46835 | tcurtis++ | branches/gsoc_past_optimization/runt​ime/parrot/library/PAST/Walker.pir:
03:26 dalek parrot: Added POD docs to PAST::Walker.
03:26 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46835/
03:26 Coke past::Walker is <reply>Texas Ranger.
03:27 dalek plparrot: a1dd103 | dukeleto++ |  (6 files):
03:27 dalek plparrot: Revert "Rename plparrot to plparrotu"
03:27 dalek plparrot: This reverts commit 4bc41c54467b7ea1613ec287b12ef4542c9f3f7a.
03:27 dalek plparrot: review: http://github.com/leto/plparrot/commit/a​1dd10331f4d2ca8ef9cf993fde0d1de03bbbfbb
03:27 dalek plparrot: 715aa8a | dukeleto++ | plparrot.c:
03:27 dalek plparrot: Create a trusted and untrusted interpreter object in _PG_init
03:27 dalek plparrot: review: http://github.com/leto/plparrot/commit/7​15aa8a2f5df4a349bfb524af56fc5a1ba650513
03:27 dalek plparrot: cd79f29 | dukeleto++ |  (2 files):
03:27 dalek plparrot: Make the language 'plpir' an alias for 'plparrot'
03:27 dalek plparrot: review: http://github.com/leto/plparrot/commit/c​d79f290b16f4e219ef499d62a4a14fb268672ab
03:31 Coke plarrotu, barata, nick*coughcough*
03:36 dukeleto Coke: what now?
03:36 purl The chalice contains human blood cursed in Death's name. The hooded figure kneeling beside you passes it to you and you have to drink.  LOSE 3 HIT POINTS. or the catchphrase of Jack Bauer.
03:40 Coke dukeleto: are you familiar with the day the earth stood still?
03:40 dukeleto Coke: i guess not
03:42 LoganLK joined #parrot
03:48 sorear Andy: GCC 4. ... 6?
03:48 Andy yessir
03:48 cotto he's crazy like that
03:48 sorear but - gcc 4.5 hasn't even been out a month
03:49 sorear man release cycles are short now
03:49 sorear ...and it seems 4.5 isn't even available in debian yet
03:56 cotto dukeleto, I can't reproduce the backtrace from your nopaste
03:57 janus joined #parrot
03:57 cotto let me see if it explodes with --optimize
04:01 Andy sorear: It's a development snapshot
04:01 Andy ftp://ftp.nluug.nl/mirror/languages/gcc/snapshots/
04:02 Andy http://nickclifton.livejournal​.com/5630.html?view=6142#t6142
04:06 Andy Join the party
04:08 cotto ok.  fails with 2.4.0-something build
04:08 cotto sounds like a bug
04:08 purl bzzzzzzzzzzzzzz...
04:09 cotto as does purl
04:09 * sorear wonders if there's a good way to install a sandboxed GCC that can only be used explicitly
04:12 dukeleto cotto: should I make a tt? the docs say "unpredictable results will happen", so i just gave the 2nd interp a parent
04:14 tcurtis msg chromatic PAST::Walker turned out to be really simple to implement(~32 lines of PIR, not counting empty lines, POD, and comments). Now on to test-writing.
04:14 purl Message for chromatic stored.
04:15 cotto dukeleto, good question.  I'm not sure if something like that should just count as "don't do that" or if we should make it work.
04:16 dalek parrot: r46836 | tcurtis++ | branches/gsoc_past_optimization/​examples/library/pastwalker.pir:
04:16 dalek parrot: Add an example of PAST modification in PAST::Walker.
04:16 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46836/
04:19 cotto a failing assert is better than a mysterious segfault
04:21 dukeleto cotto: yep, i agree. i think, from the docs, it is "don't do that"
04:21 cotto wfm
04:23 tcurtis Why does pir-basic-indent in editor/pir-mode.el default to 8? The required PIR coda seems to specify a 4-space indent for vim, and most of the pir in runtime/parrot/library seems to be indented to 4 spaces.
04:25 cotto because nobody is in the intersection of the set of people who use emacs and the set of people who'd notice and mention it
04:30 dukeleto tcurtis: please fix it :)
04:33 dalek parrot: r46837 | tcurtis++ | branches/gsoc_past_optimization (2 files):
04:33 dalek parrot: Change to 4-space indent for consistency.
04:33 purl dalek: that doesn't look right
04:33 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46837/
04:34 cotto purl, go jump in a lake
04:34 purl cotto: huh?
04:34 cotto tcurtis, I'd recommend doing that in trunk
04:36 tcurtis cotto: that commit was fixing the indent in runtime/parrot/library/PAST/Walker.pir and examples/library/pastwalker.pir. I just updated pir-mode.el in trunk.
04:37 cotto you're one step more awesome than me
04:37 cotto as you were ;)
04:43 Andy joined #parrot
04:49 dalek parrot: r46838 | tcurtis++ | trunk/editor/pir-mode.el:
04:49 dalek parrot: Fix pir-basic-indent default to 4.
04:49 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46838/
05:01 parthm joined #parrot
05:07 plobsing joined #parrot
05:08 dalek plparrot: 4b4f9fc | dukeleto++ |  (3 files):
05:08 dalek plparrot: Start the PL/PIR, PL/PIRU bifurcation
05:08 dalek plparrot: Some odd test failures still exist for returning values from PIRU.
05:08 dalek plparrot: review: http://github.com/leto/plparrot/commit/4​b4f9fc02daf6d0cad0b1857dff18b8aa30ab551
05:14 hercynium joined #parrot
05:18 LoganLK joined #parrot
06:03 kjeldahl joined #parrot
06:04 parthm left #parrot
06:12 dalek parrot: r46839 | petdance++ | trunk/config/auto/warnings.pm:
06:12 dalek parrot: new warnings available in GCC 4.6
06:12 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46839/
06:13 uniejo joined #parrot
06:18 aukjan joined #parrot
06:28 dalek parrot: r46840 | petdance++ | trunk (2 files):
06:29 dalek parrot: Flagging functions as pure.  Thanks, GCC 4.6!
06:29 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46840/
06:29 dalek parrot: r46841 | tcurtis++ | failed to fetch changeset:
06:29 dalek parrot: Sync with trunk.
06:29 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46841/
06:32 cotto that sync doesn't look right
06:32 iblechbot joined #parrot
06:43 tcurtis cotto: No? How so?
06:45 cotto tcurtis, it looks like it's only a bunch of svn prop changes
06:46 JimmyZ_ joined #parrot
06:50 cotto ops.num--
06:53 plobsing ops.num++ # lets me know how many I've offed
06:56 tcurtis Oops. Thanks, cotto. Reverting and trying to figure out how I messed it up now.
06:57 cotto tcurtis, which svn version are you using?
06:57 tcurtis 1.6.11
06:57 cotto ok.  merges seem to be the least painful with 1.6.x, so that's probably not the problem.
06:57 cotto (and syncs)
06:58 cotto It should just be:
06:58 cotto svn merge https://svn.parrot.org/parrot/trunk .
06:58 cotto sync?
06:58 purl somebody said sync was defined by SVID at least, to schedule the writes of all buffered data to disk, and an implementation can choose wither sync() returns before the writes are done, or after
06:59 cotto svn sync?
06:59 purl svn sync is *so much faster*
06:59 cotto no, svn sync is svn merge https://svn.parrot.org/parrot/trunk .
06:59 purl okay, cotto.
06:59 cotto svn sync?
06:59 purl rumour has it svn sync is svn merge https://svn.parrot.org/parrot/trunk
06:59 cotto no, svn sync is svn merge https://svn.parrot.org/parrot/trunk . .
06:59 purl okay, cotto.
06:59 cotto svn sync?
06:59 purl svn sync is svn merge https://svn.parrot.org/parrot/trunk . .
06:59 cotto no, svn sync is svn merge https://svn.parrot.org/parrot/trunk .
06:59 purl okay, cotto.
06:59 cotto svn sync?
06:59 purl svn sync is svn merge https://svn.parrot.org/parrot/trunk .
07:03 dalek parrot: r46842 | tcurtis++ | branches/gsoc_past_optimization (78 files):
07:03 dalek parrot: Revert failed attempt to sync with trunk.
07:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46842/
07:03 dalek parrot: r46843 | tcurtis++ | branches/gsoc_past_optimization (27 files):
07:03 dalek parrot: Really sync with trunk.
07:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46843/
07:04 tcurtis Possible source of problem: I was using 1.4, then noticed that, reverted the initial merge, installed 1.6, and used the same command line as 1.4 was demanding. which was something like "svn merge https://svn.parrot.org/parrot/trunk https://svn.parrot.org/parrot/b​ranches/gsoc_past_optimization". Although that might have not been the right thing to do even with 1.4. Anyway, this time, svn diff seems to be showing actual changes to files, so it should have wor
07:04 cotto yeah.  Using different versions of svn-- can lead to stupiness.
07:05 cotto svn gives me that syncing feeling
07:05 tcurtis karma svn
07:05 purl svn has karma of -50
07:06 cotto good
07:06 cotto karma git
07:06 purl git has karma of 288
07:07 cotto sounds proper
07:07 fperrad joined #parrot
07:08 cotto hio fperrad
07:12 fperrad_ joined #parrot
07:19 dalek parrot: r46844 | cotto++ | branches/ops_pct/compilers/opsc (7 files):
07:19 dalek parrot: [opsc] add Ops::Renumberer, make it complain when ops.num needs renumbering
07:19 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46844/
07:19 dalek parrot: r46845 | cotto++ | branches/ops_pct/compilers/opsc/ops2c.nqp:
07:19 dalek parrot: [opsc] add --debug to ops2c to disable file output
07:19 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46845/
08:03 mikehh cotto: ops_pct branch - make corevm/make coretest PASS, make -> make: *** No rule to make target `compilers/opsc/src/Ops/Renumberer.pm', needed by `compilers/opsc/gen/Ops/Renumberer.pir'. Stop.
08:08 JimmyZ_ joined #parrot
08:09 dalek parrot: r46846 | fperrad++ | trunk (2 files):
08:09 dalek parrot: [URI] add userinfo()
08:09 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46846/
08:14 cotto That shouldn't be the case.  I updated the relevant makefile fragments.
08:14 cotto nm
08:14 cotto forgot to svn add apparently
08:15 khairul joined #parrot
08:15 cotto all better
08:25 dalek parrot: r46847 | cotto++ | branches/ops_pct/compilers/​opsc/src/Ops/Renumberer.pm:
08:25 dalek parrot: [opsc] add Ops::Renumberer file, mikehh++ for noticing
08:25 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46847/
08:36 mikehh cotto: my next trick is -> make: *** No rule to make target `ext/nqp-rx/src/stage0/nqp-setting.nqp', needed by `runtime/parrot/library/nqp-setting.pbc'. Stop.
08:40 cotto that's after a reconfig?
08:40 mikehh cotto: again make corevm/make coretest PASS, but make fails after making the previous fail, I did make realclean, config, make corevm/make coretest, make
08:40 cotto so it is
08:41 cotto lemme dig a bit, but I'll need to sleep rsn
08:41 mikehh ok - I'll have a look as well
08:41 cotto speedy typing, btw ;)
08:41 dalek parrot: r46848 | fperrad++ | trunk (3 files):
08:41 dalek parrot: [LWP] proxy (step 1)
08:41 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46848/
08:43 cotto stupid bacekbot is already into his weekend and I just barely started friday
08:43 * cotto shakes his fist in vain
08:43 cotto but come Sunday, the tables will be turnede
08:45 cotto mikehh, is that a serial or a parallel build?  It looks like the Makefile dependencies are not quite complete.
08:45 cotto nm.  They're just plain broken.
08:46 mikehh I re-gened MANIFEST but haven't commited yet, may need another
08:47 mikehh once I get make sorted I will try and fix codetest failures
08:47 cotto fix't
08:47 cotto and I'm off to find some sleeps
08:47 cotto g'night
08:47 moritz sleep well
08:48 mikehh ok, let's see what happens :-}
08:53 TiMBuS joined #parrot
08:58 dalek parrot: r46849 | cotto++ | branches/ops_pct/ext/nqp-rx/​src/stage0/nqp-setting.nqp:
08:58 dalek parrot: [nqp] move nqp-setting.pm to nqp-setting.nqp
08:58 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46849/
08:59 darbelo That's going to hurt when merge time comes.
09:04 mikehh darbelo: what is?
09:05 mikehh the nqp-settings.nqp?
09:08 darbelo Renaming files in general.
09:08 darbelo And that particular file got renamed in both branch and trunk at separate times.
09:09 moritz merge with git! merge with git!
09:09 darbelo Every time that's happened to me svn reports that the missing file on one side conflicts with the missing file on the other.
09:10 mikehh we'll let bacek sort it out :-}
09:12 darbelo Yeah, bacek can handle it. He's some sort of magical coding robot, after all.
09:13 mikehh for sure
09:13 purl like totally!
09:14 dalek parrot: r46850 | mikehh++ | branches/ops_pct/MANIFEST:
09:14 dalek parrot: re-generate MANIFEST
09:14 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46850/
09:16 mikehh ops_pct branch - make corevm/make coretest PASS, make test - t/compilers/opsc/05-oplib.t - Method 'optable' not found for invocant of class ''
09:18 moritz I wish we could get rid of such less-than-awesome error messages
09:18 moritz maybe with a flag that marks a class as anonymous
09:19 moritz and then the error message could read "for invocant of an anonymous class" or so
09:20 mikehh moritz: I haven't looked at the test yet - it's new in ops_pct branch
09:21 moritz mikehh: it's not a matter of the test
09:21 moritz mikehh: it's an error message we see in rakudo too, leaking through from parrot
09:22 moritz it's more of a general parrot method calling error
09:27 mikehh morritz: yeah - I see where you are coming from
09:30 mikehh bad typing again - s/morritz/moritz/
09:31 dalek parrot: r46851 | fperrad++ | trunk/runtime/parrot/library/LWP.pir:
09:31 dalek parrot: [LWP] proxy (step 2)
09:31 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46851/
09:31 moritz mikehh: you're aware of nick tab completion?
09:32 moritz (I don't want to make you feel bad about the typo, just want to save you some key strokes in future :-)
09:32 darbelo moritz: How do you know his client has it?
09:32 moritz darbelo: most sane clients have it
09:32 moritz darbelo: and if not, it's worth searching for a better client, if possible
09:32 mikehh nah it works - I generally forget to do it :-{
09:34 darbelo If it makes you feel any better I'd rather have my nick misspelled than be called a 'dalek' ;)
09:34 moritz lol
09:34 darbelo Both have happened. The second one was way more confusing.
09:35 moritz in #perl6 there are often 6 nicks starting with 'ma'
09:35 moritz or rather 5, but one starts with [ma, and the tab completion in irssi also picks that one up on ma<tab>
09:39 moritz Coke: FWIW I couldn't reproduce the 'make install' oddity with parallel build
10:09 darbelo Hmm. Our encoding of literal UTF16 surrogate pairs seems slightly off...
10:20 moritz you don't need surrogae pairs in UTF-16 I thought?
10:20 moritz just in UCS-2
10:21 nopaste "darbelo" at 192.168.1.3 pasted "Here's a testcase." (118 lines) at http://nopaste.snit.ch/20609
10:21 darbelo moritz: You can ignore the ucs4 part in trunk.
10:21 mmcleric joined #parrot
10:30 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33962), fulltest) at r46851 - Ubuntu 10.04 amd64 (g++)
10:33 clinton joined #parrot
11:16 * bacek arrives in flames and smoke
11:17 bacek WHO CALLED MY NAME?!
11:20 viklund joined #parrot
11:30 moritz we were discussing merges :-)
11:34 bacek sigh...
11:34 bacek svn again?
11:35 moritz we discussed renaming a file in a branch, and the hilarity that will cause than on merging
11:37 moritz or renamed both in trunk and branch, or so
11:37 bacek let me try to sync branch
12:01 chinchin joined #parrot
12:01 chinchin hello, i have a problem with the mk_language_shell.pl script
12:02 chinchin can somebody help me?
12:02 bacek chinchin, just tell your problem. We'll try to help
12:03 chinchin great
12:03 chinchin when i run the perl tools/dev/mk_language_shell.pl Squaak languages/squaak command, the folder is created, but not the make file
12:04 bacek chinchin, looks like documentation is outdated. There is "setup.pir" which is pure-parrot replacement for make
12:05 chinchin k, so i run it with parrot setup.dir?
12:05 bacek yes
12:06 chinchin ok great, i'll try it
12:06 darbelo Yep. Or "parrot setup.pir help" to get the list of available targets.
12:06 bacek you can run something like "parrot setup.pir help" to get more info
12:06 chinchin ahh ok, great
12:06 chinchin thanks for your help
12:06 chinchin parrot is great btw ;)
12:07 darbelo We try :)
12:12 darbelo fperrad: ping
12:14 dalek parrot: r46852 | bacek++ | branches/ops_pct (55 files):
12:14 dalek parrot: Merge branch 'master' into ops_pct_local
12:14 dalek parrot: Conflicts:
12:14 dalek parrot: MANIFEST
12:14 purl MANIFEST is useful for lots of stuff and don't forget to regenerate the manifest with perl tools/dev/mk_manifest_and_skip.pl, unless i am using git-svn
12:14 dalek parrot: runtime/parrot/library/HTTP/Message.pir
12:14 dalek parrot: runtime/parrot/library/LWP.pir
12:14 dalek parrot: t/library/lwp.t
12:14 dalek parrot: t/library/uri.t
12:14 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46852/
12:14 dalek parrot: r46853 | bacek++ | branches/ops_pct/src/dynoplibs/Rules.in:
12:14 dalek parrot: Fix build of deprecated.ops introduced in trunk.
12:14 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46853/
12:15 parthm joined #parrot
12:19 darbelo purl: msg fperrad "Chandon" is the GSoC student that will be working on parrot's threads, letting him know of any features you think would benefit distutils is probably a good idea.
12:19 purl Message for fperrad stored.
12:23 fperrad darbelo, see http://lists.parrot.org/pipermail​/parrot-dev/2010-May/004238.html
12:26 whiteknight joined #parrot
12:28 whiteknight good morning, #parrot
12:29 bacek Morning??? It's Friday's night already!
12:30 Coke joined #parrot
12:30 dalek parrot: r46854 | bacek++ | branches/ops_pct/t/compilers/opsc/05-oplib.t:
12:30 dalek parrot: Update tests
12:31 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46854/
12:31 dalek parrot: r46855 | bacek++ | branches/ops_pct/t/compilers/opsc/06-opsfile.t:
12:31 dalek parrot: Update test to refect ops removed from core.
12:31 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46855/
12:31 chinchin @bacek i still have a problem:
12:32 bacek Go ahead
12:32 chinchin after running perl tools/dev/mk_language_shell.pl Squaak languages/squaak && cd languages/squaak && parrot setup.pir build
12:33 chinchin i get the following output: Null PMC access in find_method('command_line')
12:33 chinchin current instr.: 'main' pc 12 (squaak.pir:28)
12:33 JimmyZ joined #parrot
12:35 bacek chinchin, do you have installed parrot? E.g. from package or "make install"
12:35 chinchin yes
12:37 bacek chinchin, I suspect conflict between "installed parrot" and "svn checkout".
12:38 bacek Can you try to "make installable" in svn checkout
12:38 chinchin i can try, but i never checked out
12:38 chinchin i just downloaded 2.3.0
12:39 bacek hang on. Where is "mk_language_shell.pl" came from?
12:40 chinchin from the download package
12:41 bacek ok.
12:41 chinchin i just see, i installed both 2.3.0 and 2.4.0
12:41 bacek ouch...
12:42 parthm hello. i am trying to build cardinal (latest) with parrot 2.4.0 but the build is failing. it seems nqp.pbc is missing in the "make install" dir.
12:42 bacek It's can be a main reason
12:42 nopaste "parthm" at 192.168.1.3 pasted "cardinal build failure" (23 lines) at http://nopaste.snit.ch/20610
12:42 chinchin hmm, but when i type parrot --version, i get 2.3.0
12:42 bacek chinchin, try "../../parrot setup.pir" in languages/squaak directory
12:43 chinchin k, mom
12:44 bacek parthm, old "nqp" was deprecated and removed in current development release. You need last supported (2.3.0) to build cardinal.
12:44 bacek cardinal?
12:44 purl cardinal is probably http://mail.freesoftware.fsf​.org/pipermail/cardinal-dev/ or the Ruby-on-Parrot project. or http://xrl.us/uyz3
12:45 parthm bacek: thanks. i will try it with 2.3.0.
12:47 bacek seen treed
12:47 purl treed was last seen on #parrot 10 days, 18 hours, 25 minutes and 30 seconds ago, saying: o/  [May 10 18:22:03 2010]
12:48 dukeleto darbelo: is there a pic of you on the interwebs?
12:48 bacek msg treed Aloha! What about updating Cardinal to nqp-rx? (I can probably offer some help with it)
12:48 purl Message for treed stored.
12:49 dukeleto darbelo: we are putting pics of PL/Parrot devs on a slide, and you have commits :)
12:52 chinchin @bacek i get : Error reading source file setup.dir. :(
12:53 bacek chinchin, erm... Did you invoke it in "languages/squaak" directory?
12:53 chinchin yes
12:54 chinchin gosh, was a typo
12:54 bacek no worries :)
12:54 chinchin ;)
12:54 chinchin but now i get: Parrot revision r44549 required (currently r0)
12:55 bacek yak...
12:55 chinchin sorry ;)
12:56 bacek msg fperrad there is some shenanigans with mk_language_shell.pl and distutils. Can you take a look, please?
12:56 purl Message for fperrad stored.
12:58 bacek chinchin, what is in PARROT_REVISION in languages/squaak directory?
12:58 chinchin its 44549
12:58 fperrad chinchin, put 0 in the file PARROT_REVISION
12:59 bacek try replace it with 0
12:59 parthm left #parrot
12:59 bacek Ah! fperrad is here!
12:59 chinchin k, it worked ;)
12:59 fperrad bacek, this problem is now fixed with Parrot 2.4.0
12:59 chinchin is this a bug, or is my installation broken?
13:00 chinchin ahh, i see
13:01 chinchin ok guys, its working now, thanks for your help!
13:01 bacek fperrad, hooray!
13:01 bacek chinchin, looks like is _was_ a bug. But feel free to ask fperrad for details.
13:01 bacek (He is awesome, btw)
13:02 chinchin hehe
13:03 chinchin should i write a email about the outdated docs?
13:05 bacek Just create ticket in trac
13:05 Coke if the docs are still outdated, yes, please open a trac ticket.
13:05 bacek trac?
13:05 purl trac is a web-based software project management and bug/issue tracking system emphasizing ease of use and low ceremony. It provides an interface to the Subversion revision control systems, integrated Wiki and convenient report facilities.  http://projects.edgewall.com/trac/ or Python, SQLite and ClearSilver or killing killtrac or a bug-tracking tool or at https://trac.parrot.org/parrot/ or slow or REALLY slow
13:05 Coke parrot trac?
13:05 purl i think parrot trac is https://trac.parrot.org/parrot/
13:05 * Coke likes mine better.
13:05 bacek Yeah, this one :)
13:05 Coke -> elementary school
13:06 chinchin k, i will do that...again, thanks and see ya!
13:06 particle coke: i hope you graduate this time!
13:06 * bacek passing some scotch to Coke
13:07 dalek parrot: r46856 | fperrad++ | trunk (2 files):
13:07 dalek parrot: [URI] fix host when userinfo
13:07 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46856/
13:11 JimmyZ joined #parrot
13:11 ruoso joined #parrot
13:13 dalek rakudo: c950dec | pmichaud++ | docs/spectest-progress.csv:
13:13 dalek rakudo: spectest-progress.csv update: 448 files, 32347 (82.2% of 39356) pass, 0 fail
13:13 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​950decb88e0f8a0a82a136d90697dcf6a7291a7
13:21 dalek plparrot: 7ea3c7d | dukeleto++ | t/sql/test.sql:
13:21 dalek plparrot: Add a test that concatenates strings
13:21 dalek plparrot: review: http://github.com/leto/plparrot/commit/7​ea3c7d959fc02f4b84be6430c1d9c4c91ddc9fc
13:21 atrodo joined #parrot
13:23 dalek parrot: r46857 | fperrad++ | trunk (4 files):
13:23 dalek parrot: [LWP] call env_proxy()
13:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46857/
13:23 dalek parrot: r46858 | bacek++ | branches/ops_pct/runtime/parrot/library (3 files):
13:23 dalek parrot: Bring runtime/parrot/library from trunk.
13:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46858/
13:25 fperrad who is behind a HTTP proxy ? I need a tester.
13:28 whiteknight sorry, not me
13:31 JimmyZ fperrad: I can
13:32 JimmyZ I use SSH as proxy
13:34 fperrad JimmyZ, you need Parrot r46857 or better, a http_proxy environment variable
13:34 fperrad $ ./parrot examples/io/get.pir http://www.parrot.org/
13:34 darbelo dukeleto: No pics of me in the interwebs, that I know of.
13:35 JimmyZ fperrad:
13:35 JimmyZ ok
13:36 fperrad $ ./parrot examples/io/get.pir http://www.parrot.org/ > home.html
13:36 fperrad only stderr is useful
13:38 JimmyZ ok, I am buiding parrot ;)
13:41 fperrad JimmyZ, what are your http_proxy like http://user:passwd@proxy.net:8000 ?
13:47 JimmyZ fperrad: http://www.samair.ru/proxy/ip-address-01.htm here are proxys
13:47 JimmyZ *proxies
13:49 plobsing joined #parrot
13:49 JimmyZ parrot examples/io/get.pir http://www.parrot.org/  > home.html
13:50 JimmyZ ** GET http://www.parrot.org/ ==> 200 OK (6s)
13:52 fperrad JimmyZ, seems ok,
13:52 fperrad now try to run the test suite and send the report to smolder
13:52 fperrad $ ./parrot t/harness.pir --archive --send-to-smolder
13:54 JimmyZ fperrad: I use http_proxy=http://112.65.135.54:8080
14:00 JimmyZ Class 'GzipHandle' not found
14:00 JimmyZ current instr.: 'parrot;TAP;Harness;Archive;runtests' pc 920 (runtime/parrot/library/TAP/Harness.pir:326)
14:00 JimmyZ called from Sub 'do' pc 246 (t/harness.pir:82)
14:01 JimmyZ fperrad: I ran parrot t/harness.pir --archive --send-to-smolder and get the error
14:01 tetragon joined #parrot
14:03 fperrad JimmyZ, I think you have a Parrot without zlib
14:03 JimmyZ ah, yes
14:09 gbacon joined #parrot
14:24 theory joined #parrot
14:29 fperrad JimmyZ, I succeed with your http_proxy setting
14:29 fperrad see http://smolder.plusthree.com/ap​p/projects/report_details/33975
14:30 fperrad great
14:30 JimmyZ Good
14:31 bubaflub joined #parrot
14:32 Andy joined #parrot
14:37 Andy I love the new GCC.
14:39 bubaflub Andy: how do you manage multiple GCC's on one system?
14:39 Andy /var/gcc4[456]
14:40 bubaflub oh, that makes sense.  then a symlink to your default gcc?
14:40 nopaste Someone at 192.168.1.3 pasted "my ~/bin/parrotconfig" (46 lines) at http://nopaste.snit.ch/20611
14:41 Andy I never touch the default gcc
14:41 Andy note that I'm also using g++ as my CC
14:41 Andy because it's more stringent
14:42 bubaflub very slick.  i wonder if a perl-brew / rvm type solution would be useful for gcc
14:42 Andy ?
14:42 Andy I don't know what that is.
14:42 bubaflub rvm is a small utility that allows you to manage multiple copies and versions of ruby from the commandline
14:42 Andy ah
14:43 bubaflub from installing and upgrading and switching system ruby
14:43 Andy I don't care about building perl on multiple GCCs
14:43 Andy just the latest
14:43 bubaflub it'll handle setting the paths and making sure not to clobber
14:43 Andy my builds have no problems. :-)
14:46 smash joined #parrot
14:46 smash hello everyone
14:46 Andy hi wat's up
14:59 mmcleric joined #parrot
15:03 dalek parrot: r46859 | petdance++ | trunk/include/parrot/compiler.h:
15:03 dalek parrot: put in docs about the details of const and pure functions
15:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46859/
15:09 dalek rakudo: f9c987e | moritz++ | t/spectest.data:
15:09 dalek rakudo: run three more integration test files
15:09 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/f​9c987ec173ee1d135211ef7626aa72dff0bc180
15:10 theory joined #parrot
15:15 solarion joined #parrot
15:18 cotto yay for bacek++
15:19 dalek parrot: r46860 | petdance++ | trunk (2 files):
15:19 dalek parrot: GCC 4.6 found more ufnctions to const
15:19 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46860/
15:23 tetragon joined #parrot
15:27 Mokurai1 joined #parrot
15:29 jsut_ joined #parrot
15:36 dukeleto the PL/Parrot talk at PGCon went well!
15:36 whiteknight Andy: ping
15:36 Andy yessir
15:37 whiteknight Andy: would you know off hand if Ack is bundled with Strawberry perl?
15:40 Coke whiteknight: pretty sure it isn't.
15:40 Coke easy install, though.
15:40 Coke C:> cpan App::Ack
15:42 Coke joined #parrot
15:43 Coke dukeleto:güt on the talk.
15:43 whiteknight Coke: yeah. I got a buddy who was looking for a "grep, but on windows", and I'm trying to sell him on Ack
15:43 Andy No need to even go the CPAN route.  Just download the single file.
15:45 * cotto_work hopes that the next Ubuntu will be Naughty Newt.
15:45 Coke Andy: yes, but cpan will install it as something you can run.
15:45 Coke as opposed to a .pl file that you then have to write a .bat wrapper for.
15:46 Andy I know.
15:47 solarion joined #parrot
15:48 whiteknight I suspect it will take me less time to upgrade ubuntu to 10.04 and download the LLVM2.7 package than it will to svn up and build it myself
15:48 whiteknight building LLVM+Clang takes a hell of a long time on this machine
15:52 dalek parrot: r46861 | petdance++ | trunk/src/string/charset.c:
15:52 dalek parrot: Marking more functions as pure
15:52 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46861/
15:52 cotto_work petdance++
15:53 whiteknight agreed. petdance++
15:54 Andy thanks
15:54 Andy I love new toys
16:00 jan joined #parrot
16:09 dalek parrot: r46862 | mikehh++ | branches/ops_pct/config/auto/extra_nci_thunks.pm:
16:09 dalek parrot: fix perlcritic failure - hard tabs
16:09 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46862/
16:13 cotto_work mikehh: why are you fixing those in the branch?
16:18 dukeleto http://openhatch.org/ is really cool. Parrot needs to be on there.
16:18 * dukeleto gets to work
16:19 * moritz has an account there
16:19 atrodo That's a cool idea
16:20 cotto_work yes
16:20 GodFather joined #parrot
16:20 Coke is that from LOST or something? ;)
16:20 * cotto_work expects an offer he can't refuse.
16:21 Coke cotto_work: is the branch dead?
16:22 cotto_work Coke: ops_pct is very much not dead.  It's just that the code mikehh was changing has nothing to do with the aim of the branch.
16:22 particle http://google.com # awesome!
16:22 dukeleto moritz: what are your first impressions of openhatch?
16:23 mikehh cotto_work: I was fixin' them to get codetest to pass
16:24 cotto_work mikehh: sure but they should be fixed in trunk and they those changes synced with the branch
16:25 mikehh cotto_work:  I thought that was what bacek did
16:25 moritz dukeleto: nice, a bit slow at times
16:26 cotto_work me too.  If those tabs don't show up in trunk, I'm not sure why they'd be in the branch.
16:26 moritz dukeleto: somehow I missed a big "publish a task list" or so
16:26 dalek parrot: r46863 | mikehh++ | branches/ops_pct/t/steps/a​uto/extra_nci_thunks-01.t:
16:26 dalek parrot: fix perlcritic failure - hard tabs
16:26 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46863/
16:26 dalek parrot: r46864 | mikehh++ | branches/ops_pct (13 files):
16:26 dalek parrot: add svn propereties
16:26 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46864/
16:27 dukeleto moritz: do you have a link for where you are on openhatch?
16:27 moritz http://openhatch.org/people/moritz/
16:27 clinton joined #parrot
16:28 mikehh cotto_work: I got one file failing codetest now - two tests - include/parrot/oplib/ops.h which AFAICS is a generated file which should not be tested by codetest
16:30 cotto_work It is indeed generated.
16:32 cotto_work It shouldn't be hard to fix anyway, depending on which codingstd fails.
16:35 mikehh cotto_work: c_code_coda.t and copyright.t (the coda has an extra line -   *   buffer-read-only: t , and no copyright
16:36 cotto_work probably better to add an exception then.
16:41 Coke not if other generated files are doing this right.
16:41 Coke where right is in air quotes.
16:41 mikehh cotto_work: if I can work out why include/parrot/oplib/ops.h fails and include/parrot/oplib/core_ops.h (passes or is not tested) I'll be there
16:59 mikehh cotto_work: both files are in MANIFEST, MANIFEST.SKIP and MANIFEST.generated, both lack a copyright line and both have the same coda
17:04 * mikehh I am confuze-ed
17:06 iblechbot joined #parrot
17:08 joeri joined #parrot
17:11 Coke mikehh: perhaps it's skipped.
17:12 mikehh Coke: both files are but one is tested the other not
17:14 Coke ... if they're tested then they aren't skipped?
17:15 dalek rakudo: 6a7ded6 | moritz++ | src/Perl6/Module/Loader.pm:
17:15 dalek rakudo: Not finding a module throws an exception again.
17:15 dalek rakudo: The intention of not having an unnecessary stack trace was noble, but it made
17:15 dalek rakudo: it very hard (if not impossible) to catch the error with a try { } block, or
17:15 dalek rakudo: form within an eval.
17:15 dalek rakudo: Also awesomeified the error message to include report contents of @*INC.
17:15 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​a7ded6234296b8588761429664bcfb765ec52eb
17:19 mikehh Coke: there are two files in the directory include/parrot/oplib - core_ops.h and ops.h, both are in MANIFEST, MANIFEST.SKIP and MANIFEST.generated
17:20 mikehh both are skipped one as ops.h and the other as *_ops.h in the directory
17:21 mikehh both files fail both tests if invoked directly but only one ops.h fails codetest
17:25 Coke mikehh: are you sure the /codetest/ is skipping them both?
17:26 theory joined #parrot
17:27 Chandon joined #parrot
17:27 mikehh Coke: no it only skips one core_ops.h - it tests ops.h and fails two tests - c_code_coda.t and copyright.t - if I test core_ops.h against thes tests it fails and seeing it is not failing codetest I assume it is being skipped as required
17:29 * NotFound is back from http://imperialstormtroopers.com/
17:29 mikehh NotFound: how was it?
17:30 NotFound mikehh: very hot ':)
17:30 Coke mikehh: right. so make it skip both!
17:30 mikehh Coke: yes but how?
17:30 purl but how is it you know.. red?
17:31 tetragon joined #parrot
17:31 Coke ... read the code to see how it's skipping the other one?
17:31 NotFound Lots of people, plain sun, too much walking... I'm tired.
17:31 * Coke checks c_code...
17:31 * Coke assumes it's in lib/Parrot/Distribution's get_c_language_files
17:32 mikehh Coke: right
17:33 Coke ... but that's a mess.
17:33 Coke (I can't see where c_header_files is set.)
17:34 Coke Ah, coding standards tests. I again wonder if you're worth the bits you're printed on. =-)
17:35 mikehh Coke: ok got it
17:36 mikehh one is in exemptions, the failinfg one aint
17:36 mikehh failing
17:37 Coke (and this is why we no longer run those tests on 'make test'. =-)
17:38 * cotto_work wonders what it'd take for a pure pir program to extract op info from a dynop library
17:41 darbelo cotto_work: Best case: dlload() trickery, I'd expect.
17:42 cotto_work probably premature to think about that just now
17:45 mikehh phew - codetest now passe4s in ops_pct branch
17:45 mikehh passes
17:52 dalek parrot: r46865 | mikehh++ | branches/ops_pct/lib/Parrot/Distribution.pm:
17:52 dalek parrot: add include/parrot/oplib/ops.h to @exemptions
17:52 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46865/
18:00 Coke plparrot folks - I can help you write a date/time dynpmc for your HLL if you want.
18:01 Coke (or a PIR class.)
18:02 tcurtis joined #parrot
18:05 LoganLK joined #parrot
18:14 mmcleric left #parrot
18:22 davidfetter joined #parrot
18:39 dalek nqp-rx: def10fd | sorear++ | src/HLL/Compiler.pm:
18:39 dalek nqp-rx: Write NQP replacements for eval and interactive
18:39 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/d​ef10fd60c43741c13d6de45b7386568922c412f
18:39 dalek nqp-rx: 528fb67 | sorear++ | src/HLL/Compiler.pm:
18:39 dalek nqp-rx: Add the autoprint hook
18:39 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/5​28fb676886eaa380469cc11ae87ac3f9e0f7ddd
18:39 dalek nqp-rx: c4649e1 | pmichaud++ | src/HLL/Compiler.pm:
18:39 dalek nqp-rx: Refactor outer_ctx handling a bit in HLL::Compiler.
18:39 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/c​4649e11887874182d9bd2a56be6a02c00499834
18:39 dalek nqp-rx: fca4ee1 | pmichaud++ | src/HLL/Compiler.pm:
18:39 dalek nqp-rx: HLL::Compiler.eval() needs to pass its slurpy args to the code it evals.
18:39 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/f​ca4ee177643413f80ac426e9b565f0a62c578c1
18:42 ash_ joined #parrot
18:42 cotto_work what's autoprint?  It sounds shiny.
18:42 cotto_work autoprint?
18:44 pmichaud given an interactive line
18:44 pmichaud like
18:44 pmichaud > 3 +4
18:44 pmichaud autoprint automatically preints the return value from the interactive code entered
18:44 pmichaud *prints
18:44 pmichaud (without requiring an explicit 'say')
18:44 Coke pmichaud: does it have to be an actual RETURN ?
18:45 pmichaud how do you mean "actual RETURN" ?
18:45 Coke an exception of type... <digging>
18:45 pmichaud no
18:45 pmichaud it's whatever value is returned from the called code
18:45 pmichaud it's not an exception handler.
18:45 Coke CONTROL_RETURN
18:46 Coke so, it's the PIR return value?
18:46 pmichaud Yes.
18:46 Coke ok. is it overridable by HLL?
18:46 pmichaud Yes, it's a method.
18:46 Coke woot.
18:46 pmichaud (on the compiler object)
18:46 Coke I'll investigate adding to partcl-nqp. danke.
18:46 pmichaud the base compiler object just does pir::say(...)  on whatever comes back.
18:47 pmichaud but it can be easily overridden to provide other behavior.
18:47 pmichaud one tricky part I haven't quite figured out what to do with....:
18:47 pmichaud pmichaud@orange:~/nqp-rx$ ./nqp
18:47 pmichaud > say(3+4)
18:47 pmichaud 7
18:47 pmichaud 1
18:47 pmichaud >
18:48 pmichaud not sure if/how to eliminate that extra "1" that is returned from say()
18:48 pmichaud for now, autoprint just punts and always outputs the result
18:48 Coke I think tcl's interactive shell (the real one) checks to see if you output anything before invoking autoprint.
18:48 moritz pmichaud: better than no autoprinting at all, but LTA
18:48 Coke so you have to capture STDOUT
18:48 pmichaud right... I'm thinking it needs to be per-compiler to say "did I output anything", rather than HLL::Compiler imposing a solution on everyone.
18:49 * Coke tests.
18:49 Coke yup. %puts hi
18:49 Coke hi
18:49 purl hey, Coke.
18:49 Coke %set a 3
18:49 Coke 3
18:50 pmichaud if there's an easy way to flag stdout to say "was anything printed since the flag was last unset", we could use that.
18:50 pmichaud Or some sort of query on stdout to ask "how much has been printed?"
18:50 pmichaud I dunno if ftell has any magic for us there.
18:51 * pmichaud looks at FileHandle PMC a bit.
18:51 whiteknight that sounds like something that could be added easily enough
18:51 whiteknight a quick boolean flag that sets on write and resets on read
18:52 whiteknight er, resets when the flag value is read
18:52 pmichaud I'd prefer an explicit reset to an automatic one.
18:52 pmichaud is there a way to query file_pos on a FileHandle PMC?
18:52 whiteknight that would be fine too. They would likely be methodcalls, so if you can handle the performance it, I won't complain
18:54 pmichaud looks like the 'tell' opcode.
18:54 pmichaud testing.
18:54 purl testing is the best thing ever ZOMG
18:54 Coke pmichaud: we could add it to get_int_keyed_int.
18:55 Coke ... d'oh, but tell is better.
18:55 pmichaud woot!
18:55 pmichaud nopaste coming!
18:55 parthm joined #parrot
18:56 nopaste "coke" at 192.168.1.3 pasted "for pmichaud" (10 lines) at http://nopaste.snit.ch/20612
18:56 nopaste "pmichaud" at 192.168.1.3 pasted "detecting output to stdout" (15 lines) at http://nopaste.snit.ch/20614
18:56 pmichaud looks like both coke and I ended up with similar results there :)
18:56 pmichaud okay, I'll put that into autoprint
18:56 Coke ayup. now just need to keep track of $I0 after prompt and before autorpitn.
18:56 pmichaud easy-to-do
18:56 Coke pmichaud++
18:56 Coke pmichaud++
19:01 parthm left #parrot
19:06 Coke *bounce*
19:06 pmichaud pmichaud@orange:~/nqp-rx$ ./nqp
19:06 pmichaud > 3+4
19:06 pmichaud 7
19:06 pmichaud > say(3+4)
19:06 pmichaud 7
19:06 Coke *bounce*
19:06 pmichaud > pmichaud@orange:~/nqp-rx$
19:07 pmichaud Coke++
19:07 pmichaud awesome.
19:07 mikehh ops_pct branch - All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33979), fulltest) at r46866 - Ubuntu 10.04 amd64 (g++)
19:07 Coke hell, you did all the work. =-)
19:07 Coke (it's way too easy to get karma here. =-)
19:07 pmichaud I hadn't thought of monitoring stdout, though.
19:07 pmichaud (Actually, you mentioned it long ago, also, but I had forgotten between then and now)
19:07 Coke what, you been busy? =-)
19:07 pmichaud A bit. :-)
19:08 Coke going to yaps::na 2010?
19:08 Coke going to yapc::na 2010?
19:08 purl hmmm... going to yapc::na 2010 is qw(jhannah rbuels cfedde apeiron dha nacmac dhoss mst chargrill kyriel triddle DrForr coke packy dukeleto kolibrie colomon jasonmay arcanez dylan doy ology waltman autarch)
19:08 pmichaud haven't decided yet
19:08 * Coke was trolling the bot.
19:08 pmichaud at this point, probably no, because travel expenses would be out of my pocket and I promised Paula I wouldn't do that this year.
19:08 pmichaud oh.
19:08 Coke yah. I'm trying to coordinate $$ with the wife also.
19:09 Coke she's all for me going, but I'm trying to add a cheapo laptop in there. =-)
19:09 pmichaud heh
19:14 dalek parrot: r46866 | mikehh++ | branches/ops_pct/MANIFEST.generated:
19:14 dalek parrot: rebuild MANIFEST.generated - not sorted correctly and duplicates
19:14 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46866/
19:14 dalek parrot: r46867 | pmichaud++ | trunk/docs/pdds/draft/pdd31_hll.pod:
19:14 dalek parrot: [docs] Note in pdd31 that eval() also passes along its slurpy args
19:14 dalek parrot: to the evaluated code.
19:14 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46867/
19:29 ash_ with (un)managedstruct's why do you have to push multiple 0's into them?
19:30 pmichaud .oO( because otherwise stuff leaks out into the Gulf of Mexico )
19:30 ash_ snap, those must by leaky
19:33 ash_ ah, i found some documentation for them, just not in as obvious of a place as i would of guessed, i guess
19:36 moritz then add a pointer to the non-obvious place in the obvious palce
19:37 ash_ sure, i'll try to update the documentation for the struct's (and the tests... the tests for unmanagedstruct is blank)
19:38 ash_ okay, its not blank, it has 1 test, it tries to: new $P0, ['UnManagedStruct'] thats all... i imagine UnManagedStruct has a bit more functionality than that
19:40 dalek nqp-rx: 3ad3114 | pmichaud++ | src/HLL/Compiler.pm:
19:40 dalek nqp-rx: Improve autoprint so that it only outputs the result value if the
19:40 dalek nqp-rx: executed code didn't itself send anything to stdout.  Coke++
19:40 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/3​ad311437df65191e7a63f4434409d47b2f3a8dd
20:00 bubaflub joined #parrot
20:06 davidfetter joined #parrot
20:19 dukeleto i don't think i will be able to make it to yapc::na 2010 (backlogging)
20:19 dukeleto sad, but true
20:21 dalek parrot: r46868 | NotFound++ | trunk/docs/pdds/pdd13_bytecode.pod:
20:21 dalek parrot: reality check
20:21 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46868/
20:24 tetragon joined #parrot
20:54 ash_ can you override assignment operations? so if i have my own pmc on the left hand side of the operation it does something different? (ie. $P0 = new 'myclass' \n $P0 = 123    does a special operation and not override the value)
20:56 ash_ ah, nevermind, i found out how, its in the documentation, i should read that more throughly
20:56 dalek parrot: r46869 | tcurtis++ | branches/gsoc_past_optimization (2 files):
20:56 dalek parrot: Add some tests for non-modifying traversal with PAST::Walker.
20:56 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46869/
21:12 dalek TT #1107 closed by NotFound++: Opcode get_class has failed assertion with garbage input
21:12 dalek TT #1107: http://trac.parrot.org/parrot/ticket/1107
21:26 Mokurai joined #parrot
21:29 NotFound Coke: ping
21:33 theory joined #parrot
21:35 wagle joined #parrot
22:01 kiwichris joined #parrot
22:18 Coke NotFound: yes?
22:20 NotFound Coke: please take a look at TT #1034
22:24 tcurtis joined #parrot
22:29 Whiteknight joined #parrot
22:33 Coke I don't think another way to avoid the problem necessarily helps, does it?
22:34 Coke or is this supposedly the right way?
22:35 ash_ can parrot objects freely inherit from core PMC's?
22:35 Whiteknight sortof
22:35 NotFound Coke: with our current architecture, this is the only way I can imagine.
22:35 Whiteknight in theory they can, but the mechanism has bugs
22:36 ash_ hmm, okay
22:37 ash_ http://gist.github.com/409521 anyone have any comments on that? if that was potentially how NCI worked? I still am shying away from the whole a string defines the signature of a function, that just seems odd to me personally. I mean what if you wanted to abstract away a data type, the string mechanism wouldn't let you.
22:38 tcurtis msg chromatic I only just realized that PAST::Walker's current design is suboptimal for changing the tree. $walker->walk($node) can't replace $node with a different type of node. I'll work on figuring out a good API for modifying traversal tomorrow(busy with graduation tonight). Probably similar to python's ast.NodeTransformer, with walk() returning the new node to replace the original(or null to delete it). PAST::Walker as is could be nice to keep for thin
22:38 tcurtis that don't need to modify the tree, though.
22:38 purl Message for chromatic stored.
22:42 Whiteknight ash_: looks reasonable to me, on first glance
22:42 tcurtis ash_, maybe not tie C struct names to Structure subclass names.
22:42 tcurtis Nevermind. That was a silly thing to say. C struct names wouldn't actually matter.
22:42 NotFound ash_: the string mechanism has the advantage of shortness.
22:43 * PerlJam likes the string mechanism.
22:43 PerlJam reminds me of pack/unpack
22:43 PerlJam (but maybe I'm just crazy too  :)
22:43 NotFound ash_: also, creating a PMC for all arguments of nci call will kill performance.
22:45 ash_ I am sure I could work out some mechanism that tries to convert given types to their C equivalents, so if its supposed to take an c_int32 it could convert to an int first, and see if its still a valid value, so you don't have to make throw away objects
22:46 tcurtis The string mechanism would be nice to keep for simple things where it is sufficient, but the non-string-based mechanism would be helpful when one needs it.
22:46 Mokurai2 joined #parrot
22:47 ash_ I am just thinking when you get to a point where you need to call a library with a struct, the current system seems, IMO, a bit lacking there
22:48 ash_ you could say "P" in the signature and pass it a ManagedStruct but even the ManagedStruct is a bit cumbersome and won't work with a return type (granted, you don't normally see struct's as return types)
22:49 NotFound ash_: yes, but IMO the cost of complicating and slowing down the simpler cases is too high.
22:49 ash_ I don't plan on breaking old functionality, yet anyway
22:51 PerlJam ash_: I like the hash interface to the library though
22:52 ash_ so, if you had a function that was:  int foo(float, double);   and you wanted to invoke that, you could do: $P1 = new 'c_int'; $P1 = 123; $P2 = new 'c_float'; $P2 = 123.123; $P3 = new 'c_double'; $P3 = 123.123456;  $P4 = 'dlfunc'(lib, 'foo', $P1, $P2, $P3); $P4($P1, $P2, $P3);
22:53 PerlJam ash_: ETOOMANYPMCS
22:54 NotFound ETOOMANYCODE also
22:56 tcurtis I'd rather do $P0 = 'dlfunc', lib, 'foo', 'ifd"); $I0 = $P0(123.123, 123.123456); for that.
22:58 bacek Good morning, humans
22:58 ash_ I don't have to break old functionality, i just think the added bits would be useful when you get to structs or unions
22:59 Psyche^ joined #parrot
23:00 tcurtis ash_: I agree. Your new way of doing things sounds great for things that the old way can't do.
23:02 ash_ also, you can expand on the basic types pretty easy, for instance you could add a 'pointer' method that takes  type and returns a pointer to the type, or a 'byref' method for functions that take parameters by reference instead of by value
23:02 PerlJam ash_: $P0 = 'build_struct'('ifd');  $P1 = 'fill_struct'($P0, 123, 3.1415, 2.718281828459045);  # or some such
23:02 PerlJam just trying on an alternate reality
23:04 ash_ i could expend the syntax of the strings too, to try to include some of the other cases I am talking about
23:05 ash_ Coke made a good suggestion before too, i could try 'i(fd)' where ( ) makes the fd into a structure
23:05 PerlJam Coke++
23:05 PerlJam that too reminds me of pack/unpack
23:05 PerlJam steal the good ideas where you can find them!
23:06 ash_ maybe also add * for pointers, (in addition to the generic void*, so you get more type safety if you want it)
23:07 ash_ 'i*(fd)'   could be a int foo(struct { float; double; } *);
23:12 tcurtis ash_, also great idea. Possibly use u as a way of turning a type into unsigned. Anyway, good night folks.
23:12 * tcurtis is about to graduate.
23:13 ash_ tcurtis congrats
23:13 plobsing_work joined #parrot
23:14 tcurtis thanks, ash_.
23:14 bubaflub joined #parrot
23:15 plobsing_work hello
23:15 Whiteknight graduating highschool is no small feat. I barely did it
23:15 Whiteknight hello plobsing
23:15 ash_ hi
23:16 plobsing_work ash_: we should schedule a weekly meeting time. I had hoped it would happen naturally, but that doesn't seem to be happening
23:16 ash_ plobsing_work: yeah, so far it seems we keep missing eachother
23:18 plobsing_work I work 8h-17h Pacific Time. I prefer to work on Parrot outside of those hours, but its not a hard requirement. What times work for you?
23:18 ash_ I am in the central US timezone, and obviously free monday-friday anytime starting next week, although I do have a robotics competition on june 2-4, but I am planning on working around then and on the weekends to make up for the lost time
23:20 ash_ starting next monday, i am free just about anytime, my plan from monday till the last day of the GSoC is to hold up in my computer room from 8 to around 5 monday-friday but i am flexible
23:21 plobsing_work cool. how about ~10h Central (noon Pacific) on one of those days?
23:22 plobsing_work oops wrong way (I fail at timezones)
23:22 ash_ (me too, i just had to google the difference between pacific and central)
23:22 plobsing_work so ~14h Central (noon Pacific)
23:22 ash_ umm if your done at 5pm pacific, thats... 7pm central,  if you want to meet after your done with $work
23:24 ash_ noon your time is fine by me, or when your done with work, both are good
23:24 plobsing_work I can try to make that work, but I bike home (and frequently take "detours"), so there's somewhat of a gap in my availability
23:25 ash_ we could push it back a bit further so you have time to get home, i am a night owl anyway
23:26 plobsing_work Sure. say 18h30 my time (20h30 yours) to give me sufficient buffer? how are wednesdays for you?
23:26 ash_ wednesdays are fine by me
23:28 plobsing_work sounds like a deal
23:31 NotFound http://notfound.posterous.com/the-unroll-opcode
23:33 plobsing_work ash_: have you had a look at http://github.com/plobsing/parrot-libjit-fb ?
23:33 NotFound Coke: here is a explanation.
23:33 ash_ no, but i will
23:33 ash_ looks interesting
23:35 ash_ plobsing_work: is there a time before next wednesday we can talk? I was trying to get started on the NCI improvements, and am kinda running into some design consideration issues
23:35 plobsing_work ash_: yes, I saw in the logs
23:36 plobsing_work I'll dig up some stuff I said and also comment on your suggestions tomorrow (sorry, tonight is taken)
23:36 ash_ thats fine
23:36 ash_ should I start a mailing list email about it? or should i wait until we have talked?
23:37 ruoso joined #parrot
23:37 plobsing_work ash_: your ideas (about changing signatures) are fairly similar to the ideas I put out on the list a month or two ago
23:38 plobsing_work I've significant portions of the dlfunc and (Un)?ManagedStruct interfaces in anticipation of such changes (see DEPRECATED.pod)
23:40 plobsing_work s/I've/I've deprecated/
23:40 ash_ alright, i'll read up on those changes (I wasn't aware of them)
23:41 ash_ was also surprised by the (lack of) tests for UnManagedStruct (it has 1, it sees if you can call "new 'UnManagedStruct'")
23:42 ash_ i suppose part of that is that managedstruct inherits from unmanaged struct, and managedstruct has more tests, but still
23:42 plobsing_work there's more somewhere IIRC
23:44 NotFound The funny part is that it works better than parts with many more tests X-)
23:46 plobsing_work NotFound: would you consider adding a way to access arbitrary pirops in winxed? perhaps "using pirop gcd;"?
23:47 plobsing_work I hate having to poke you every time I need another op.
23:47 NotFound plobsing_work: yes, but it has been easier until now to add predef for the urgent needs.
23:48 plobsing_work I'll try working on a patch to add that to stage2 sometime soon.
23:49 NotFound plobsing_work: the problem is that I think in a function-like syntax, and I don't have yet a syntax to specify return type.
23:50 plobsing_work NotFound: I was thinking I'd just have it STFU about types. winxed is parrot-ly typed anyways.
23:51 NotFound plobsing_work: I'm open to ideas.
23:52 plobsing_work this is just a vague idea ATM. I'll submit a more refined idea in the form of a patch (when I get the tuits)
23:53 NotFound plobsing_work: BTW I lost your patch for named and don't remember what was the problem. Do you still have it?
23:53 plobsing_work the problem was winxed can have variables legally named "$P0", which would cause significant problems
23:53 plobsing_work or maybe it was "P0"
23:55 NotFound plobsing_work: yes, but what was the problem you were trying to solve?
23:55 plobsing_work named slurpy params IIRC
23:55 theory joined #parrot
23:56 NotFound This is part of PCC that I don't know much about. Can you provide a short use case?
23:59 NotFound I think I've lost part of my mind with the encoding and pbc issues %-)
23:59 plobsing_work function sayall(var x[named,slurpy]) { for (var y in x) say(y + "=>" x[y]); }
23:59 NotFound Ah, yes, thanks.

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

Parrot | source cross referenced