Parrot 1.7.0 "African Grey" is out! Feel free to nuke trunk | Improve PCC testing this week or the release gets it!
Set by moderator on 21 October 2009.
00:00 cotto_work joined 00:01 cotto_work joined 00:03 cotto_work joined
dalek rrot: r42002 | chromatic++ | trunk/src (2 files):
[PMC] Made CallSignature PMC used fixed-size allocator for its storage. This
 00:03
darbelo wiki.postgresql.org/images/5/58/11_...review.pdf 00:05
shorten darbelo's url is at xrl.us/bftsbb
darbelo We need to make one of those for parrot.
ttbot Parrot trunk/ r42002 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118077.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 00:10
00:10 confound_ joined 00:11 cotto_work left, cotto_work joined
dukeleto japhb: +50 to the "upgrade parrot" plumage feature 00:13
japhb: i have been doing that manually and will probably write a script soon if you don't beat me to it
Whiteknight okay, I give up on my runcore optimization for now
too much hassle 00:14
00:14 confound joined
darbelo is checking out pmc_pct now 00:14
00:14 cotto_work joined 00:15 Limbic_Region joined
chromatic Which runcore optimization? 00:15
kid51 Am experiencing a long wait on 'svn commit' 00:17
00:17 cotto_work2 joined
Limbic_Region chromatic++ - thanks for the Perl 6 Design Meeting Minutes 00:19
Whiteknight chromatic: passing a Parrot_Context** to each fastcore op function, instead of calling Parrot_pcc_get_context_struct in each one
probably not much of a win now that bacek has converted that to a macro
chromatic Oh, that one. I still like the idea. 00:22
cotto_work2 whiteknight, is it just too much repetitive mindless patching or what? 00:23
kid51 All my 'svn' commands are hanging from my server ... but not from my laptop!
Whiteknight cotto_work2: Just need to rethink it 00:24
00:25 dukeleto joined 00:26 confound joined
kid51 Don't know what went wrong there, but I can now perform 'svn' commands again. 00:28
00:29 jonathanturner joined
dalek rrot: r42003 | jkeenan++ | branches/auto_libjit (7 files):
Applying modified version of plobsing's alloca_optional.patch. This configures cleanly; pre- and post-configuration tests pass without warnings, but has not yet been run thru 'make'.
 00:29
Whiteknight I've got some cleanups to the fixed-size allocator coming 00:32
and I have some ideas for some small improvements to it
00:32 cotto_work2 left
Whiteknight my hope is that modest improvements to an oft-called function will provide big benefits 00:33
00:34 particle joined 00:41 preflex joined 00:42 spinclad joined 00:52 abqar joined
cotto_work chromatic, r42001 is very perlish C 00:53
00:54 darbelo left
cotto_work home for me 00:55
chromatic Perlish, how? 00:56
Whiteknight chromatic: could you benchmark r42004? 00:59
I don't have a baseline here
I don't expect much improvement if any, it was a very small change. 01:00
chromatic I'll give it a shot.
dalek rrot: r42004 | whiteknight++ | trunk/src/gc (3 files):
[gc] a few modest improvements to the fixed-size allocator. Reduce the number of unnecessary redundant checks that we were doing on each allocation by doing better initialization
 01:02
dukeleto i assume from my benchmarks that trunk is about 2x slower than 1.7.0 after the pcc_reapply merge. i can do some more benchmarks tonight. let me know if there are specific benchmarks that you want me to run. I will probably run the GC benchmarks and primes.pasm/fib.pir 01:05
chromatic oofib.pir is good too.
Are you running optimized builds, by the way?
dukeleto chromatic: yeps 01:06
chromatic: and by optimized I mean passing --optimize to Configure and nothing else
chromatic Can you verify that that defines DNDEBUG in your Makefile? If the string appears there, that's fine. 01:08
ttbot Parrot trunk/ r42004 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118151.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
dalek rrot: r42005 | whiteknight++ | trunk/src/gc/api.c:
[gc] fix a small issue that would prevent this from building if GC_USE_FIXED_SIZE_ALLOCATOR is 0
 01:12
01:12 Smushers joined
dukeleto chromatic: checking now 01:12
dalek TT #1126 closed by coke++: tools/dev/pprof2cg.pl not installed.
dukeleto chromatic: yep, on OS X, DNDEBUG gets defined in $CFLAGS in the makefile 01:13
chromatic: it does -DNDEBUG but other things are like -DFOO=1 . what is the reason for that? 01:14
chromatic: just wanted to verify that it is only defining it but not setting it
chromatic Whiteknight, a 0.084 percent improvement on my benchmark. 01:15
Whiteknight chromatic: yeah, I figured it would be small. 01:16
chromatic Nearly noise.
Whiteknight I've wanted to tighten it up for a while now though, so I did that
chromatic Nothing wrong with that.
Whiteknight remove a few if() checks for each allocation
chromatic We're up 13.232% today on this benchmark. 01:17
Whiteknight I thought Parrot_gc_mark_PObj_alive had been deleted
replaced with the PMC and STRING variants 01:18
chromatic We can't delete it because we don't always know what type of PObj we have.
Whiteknight okay, that's fair
dalek rrot: r42006 | chromatic++ | trunk/include/parrot/gc_api.h:
[GC] Optimized the core version of Parrot_gc_mark_PMC_alive() further, this
 01:19
rrot: r42007 | chromatic++ | trunk/src/gc/api.c:
[GC] Moved mark_special() into Parrot_gc_mark_PMC_alive_fun().
ttbot Parrot trunk/ r42005 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118188.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
chromatic That's another 1.206% there.
Whiteknight I was in the middle of doing that very thing! 01:20
I was just about to commit the mark_special thing
damnit
entire minutes of my life wasted
chromatic If you want to waste minutes of your life, get rid of some of the extra context getters/setters in Sub's invoke(). 01:21
Whiteknight aye aye capn' 01:24
Coke hey, fix my segfault.
chromatic The block starting on line 451 looks like a tempting target. Yank it out, make it static, and merge it with the block on 418 perhaps. 01:25
ttbot Parrot trunk/ r42007 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118227.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 01:26
Whiteknight should we just blatantly break Context encapsulation in Sub.invoke? 01:40
the two PMC types are pretty closely intertwined
chromatic That tempted me a bit too. 01:41
Let's take a two-step on this one.
First, clean it up a bit.
That won't hurt us at all.
Then let's break encapsulation and measure how much it helps.
Whiteknight oaky 01:43
I'm going to have to stash this though, my day is quickly running out
I'll hit it tomorrow if nobody beats me to it
01:52 preflex joined
dalek rrot: r42008 | jkeenan++ | branches/auto_libjit/config/gen/libjit.pm:
No hard tabs permitted. Rebreak some long lines (just for readability).
 02:01
rrot: r42009 | jkeenan++ | branches/auto_libjit/config/gen/libjit.pm:
Conform to codingstd: no cuddled 'else'.
 02:07
rrot: r42010 | jkeenan++ | branches/auto_libjit/config (3 files):
[codingstd] Set SVN properties on recently added files.
rrot: r42011 | jkeenan++ | branches/auto_libjit/config (3 files):
[codingstd] Set SVN properties on recently added files.
 02:11
rrot: r42012 | jkeenan++ | branches/auto_libjit/config (3 files):
[codingstd] No trailing whitespace.
02:12 patspam1 joined
dalek rrot: r42013 | jkeenan++ | branches/auto_libjit/config/auto/libjit.pm:
[codingstd] No hard tab characters.
 02:14
02:14 dukeleto joined
dalek rtcl: b0c45e7 | coke++ | t/cmd_lsort.t:
TODO a test failure from the pcc-merge
 02:16
shorten dalek's url is at xrl.us/bftt4z
02:42 janus joined 03:19 buildbot joined 03:22 preflex joined 03:41 ash_ joined 04:13 eternaleye joined 04:19 Austin joined
dalek kudo: 0e662a7 | pmichaud++ | docs/spectest-progress.csv:
Revised spectest numbers.

Failure summary: S02-lexical-conventions/unicode.rakudo aborted 5 test(s) S03-junctions/autothreading.rakudo aborted 78 test(s) S12-methods/parallel-dispatch.t aborted 21 test(s)		 04:28
shorten dalek's url is at xrl.us/bftukx
dalek TT #1133 created by Austin_Hastings++: MMD matching does not match 'scalar' with primitive types 04:35
p-rx: efe456a | pmichaud++ | src/ (4 files):
Some code for naive token prefixes for regexes.
 04:39
shorten dalek's url is at xrl.us/bftumu
05:33 theory joined 05:48 riffraff joined 05:50 riffraff joined 06:00 uniejo joined 06:10 eternaleye joined
dalek p-rx: d025580 | pmichaud++ | (4 files):
Some initial work on leading token prefixes in regexes.
 06:23
shorten dalek's url is at xrl.us/bftuwh
dalek p-rx: c206489 | pmichaud++ | src/ (4 files):
Updated version with simple protoregex token pruning.
shorten dalek's url is at xrl.us/bftuwj
dalek p-rx: 29e11dd | pmichaud++ | src/ (2 files):
Clean up some debugging messages a bit.
shorten dalek's url is at xrl.us/bftuwm
dalek kudo: dda5262 | pmichaud++ | docs/spectest-progress.csv:
spectest-progress.csv update: 453 files, 32614 (85.1% of 38326) pass, 83 fail

S02-lexical-conventions/unicode.rakudo aborted 5 test(s) S03-junctions/autothreading.rakudo aborted 78 test(s)		 06:39
shorten dalek's url is at xrl.us/bftux5
dalek p-rx: 2577049 | pmichaud++ | src/NQP/ (2 files):
[nqp]: Add elsif/else clauses.
 06:40
shorten dalek's url is at xrl.us/bftuyb
06:42 namenlos joined 07:17 payload joined 07:38 fperrad joined 07:44 bacek joined
bacek o hai 07:45
07:45 payload joined 08:20 athomason joined
dukeleto_ 'ello 08:26
dukeleto feather is back 08:27
so it looks like the pcc_reapply merge broke make -j, is there a TT for that? 08:30
chromatic Looks easy enough to fix. The src/call/args$(O) target is missing two dependencies on PMC headers. 08:33
dalek kudo: edb2547 | jnthn++ | build/Makefile.in:
Correction to Makefile for Win32 (regression introduced by accident during resig2 merge).
 08:45
shorten dalek's url is at xrl.us/bftvft
dukeleto whoa, i just stumbled upon rakudoport.sourceforge.net/ 08:47
moritz wonders who set up this project, and if he has interest in getting the patches upstream 08:50
dalek rrot: r42014 | fperrad++ | trunk/t/pmc/float.t:
[test] remove superfluous pop_eh

current instr.: 'negative_zero' pc 1528 (t\\pmc\\float.t:352) called from Sub 'test' pc 236 (t\\pmc\\float.t:48)		 08:54
ttbot Parrot trunk/ r42014 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118476.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 09:02
09:06 JimmyZ joined
dalek rrot: r42015 | fperrad++ | trunk/config/gen/config_h/config_h.in:
fix banner, broken since kill_jit merge :

This is Parrot version 1.7.0 built for nojit.		 09:16
kudo: 924e4f3 | moritz++ | README:
mention FreeBSD port in README
 09:19
shorten dalek's url is at xrl.us/bftvmc
ttbot Parrot trunk/ r42015 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118525.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 09:23
09:31 payload joined 09:32 payload1 joined
dalek p-rx: 045112d | pmichaud++ | src/ (3 files):
Rework building of initial tokens in terms of PAST.
 09:57
p-rx: 40d3046 | pmichaud++ | src/Regex/Cursor-protoregex-peek.pir:
Some infrastructure to allow protoregex peeks.
shorten dalek's url is at xrl.us/bftvyp
shorten dalek's url is at xrl.us/bftvyr
dalek p-rx: 503d396 | pmichaud++ | src/PAST/Regex.pir:
More peekability for enumcharclass nodes.
shorten dalek's url is at xrl.us/bftvyt
dalek rrot: r42016 | mikehh++ | trunk/src/gc/mark_sweep.c:
fix codetest failures - space between keyword and open parenthesis and missing documentation
 09:58
ttbot Parrot trunk/ r42016 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118574.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 10:06
10:08 payload joined
dalek rrot: r42017 | mikehh++ | trunk/config/auto/frames.pm:
fix codetest failure - perlcritic does not like TODO in perl modules
 10:08
10:11 donaldh joined
ttbot Parrot trunk/ r42017 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118608.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 10:14
Austin seen japhb 10:32
purl japhb was last seen on #parrot 10 hours, 50 minutes and 39 seconds ago, saying: I'm thinking of having metadata for each HLL that says where to find their current desired parrot-rev, and then allowing users to say 'upgrade to tip' or 'upgrade to version blessed by project foo'
10:45 Austin joined 10:52 masak joined 11:20 payload joined
nopaste "mikehh" at 81.149.189.7 pasted "t/pmc/threads.t is failing in testr (fulltest)" (46 lines) at nopaste.snit.ch/18411 11:25
dalek rrot: r42018 | mikehh++ | trunk/t/pmc/threads.t:
fix TODO passes in testC and testS - TODO passed: 13
 11:33
11:39 kid51 joined
ttbot Parrot trunk/ r42018 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118672.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 11:45
12:05 whiteknight joined
Austin howdy, whiteknight 12:05
whiteknight good morrow, Austin
how are you doing today?
Austin I'm doing pretty good. 12:06
I'm taking a bunch of code out of the close project and making it a general-purpose library.
Now, of course, it needs testing and docs. :( 12:07
12:07 mokurai left
masak hi, does this look strange to anyone? gist.github.com/215912 12:10
Austin I wonder what is the difference between language PIR and language parrot, compiler wise? 12:11
whiteknight masak: extremely weird
masak whiteknight: well, now you know how to reproduce it.
whiteknight Can you create a ticket?
masak just push stuff to an array in a long Perl 6 for loop.
whiteknight: I'll try.
whiteknight awesome. Thanks 12:12
dalek TT #1134 created by masak++: Extremely weird stack trace produced my Parrot when allocating things in a ... 12:16
masak er, s/my/by/ 12:17
dalek rrot: r42019 | jkeenan++ | branches/auto_libjit/config/auto/libjit.pm:
Add documentation patch contributed by plobsing++ in trac.parrot.org/parrot/ticket/1105.
 12:18
p-rx: 0abc653 | pmichaud++ | (5 files):
Convert token peeks to use separate *__PEEK methods.
 12:27
p-rx: 9197d53 | pmichaud++ | src/ (5 files):
Switch to using !PREFIX__* instead of *__PEEK.
p-rx: c82db7d | pmichaud++ | (5 files):
Initial code for subrule peeking.
p-rx: 4388311 | pmichaud++ | src/Regex/Cursor-protoregex-peek.pir:
Handle the case where !PREFIX_* returns a nested array.
shorten dalek's url is at xrl.us/bft2nk
dalek's url is at xrl.us/bft2nt
dalek's url is at xrl.us/bft2nx
dalek's url is at xrl.us/bft2nz
whiteknight I feel like maybe shorten doesn't serve much of a useful purpose here 12:28
moritz purl: shorten that
purl That URL is at xrl.us/bft2pc [xrl.us]
moritz whiteknight: agreed. Purl can do the same thing on demand
whiteknight well that's a nice feature 12:29
moritz so should we just kick shorten?
szbalint One bot shorter would be nice 12:30
:)
Austin So can I unignore shorten, now? 12:33
Or is this a temporary bootage?
moritz Austin: I do plan to keep it away unless I'm told otherwise 12:35
Austin cool
dalek rrot: r42020 | mikehh++ | trunk/t/op/debuginfo.t:
remove TODOs - test now passes on all runcores
 12:37
ttbot Parrot trunk/ r42020 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118727.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 12:44
dalek rrot: r42021 | mikehh++ | trunk/t/compilers/imcc/syn/regressions.t:
remove a TODO that passes in switch runcore
 12:50
12:52 bluescreen joined 12:54 PerlJam joined
dalek rrot: r42022 | mikehh++ | trunk/t/compilers/imcc/syn/macro.t:
remove TODOs that pass in switch runcore - the test now passes all runcores
 12:57
ttbot Parrot trunk/ r42021 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118765.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
13:01 davidfetter joined
ttbot Parrot trunk/ r42022 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118795.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 13:04
dalek rrot: r42023 | mikehh++ | trunk/t/compilers/imcc/syn/regressions.t:
remove SKIP/TODO that now passes
 13:07
13:08 particle joined
dalek l: 14487d8 | fperrad++ | handler/xmlwriter/xmlwriter.pir:
Now the class Xml::Handler::XmlWriter is generated in gen_actions.pir
 13:09
ttbot Parrot trunk/ r42023 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118830.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 13:13
dalek TT #1102 closed by mikehh++: test failures with t/compilers/imcc/syn/macro.t and regressions.t in fast, ... 13:22
Coke /win 3 13:24
dalek TT #1135 created by coke++: t/op/annotate-old.t fails using -f core 13:29
whiteknight yay! activity 13:38
dalek and the commit feed might be my single favorite part of the whole Parrot projet
it's just so fun to watch!
dalek TT #1124 closed by mikehh++: t/op/arithmetics_pmc.t is skipping bigint/bignum test even when gmp is ... 13:39
Coke hopes parrot's public irc face is a little less scary than perl's. 13:42
davidfetter it is in here
dalek TT #1136 created by coke++: interactive partcl segfaults in Parrot_str_byte_length 13:44
13:45 ash_ joined
Coke that seems like a simple fix. 13:47
(1136)
14:11 theory joined 14:16 ash__ joined 14:27 bluescreen joined
whiteknight I'm consistently very pleased with the professional demeanor that we've displayed around here 14:35
the chatroom and the mailing list are both very hospitable
14:48 allison joined 14:53 Psyche^ joined
whiteknight I've never seen how bad #perl is supposed to get though, so I can't compare 14:54
14:54 Andy joined
mikehh trunk - pre/post-config, smoke (#29303) PASS, fulltest FAIL at r42023 - Ubuntu 9.10 (beta updated) amd64 14:56
t/op/annotate-old.t - Failed test: 1 - in testf and testg (TT #1135)
t/pmc/threads.t - Failed tests: 8-9 - in testr
whiteknight join #perl 14:58
NotFound Enjoy: code.google.com/p/winxed/ 15:01
Coke finds github.com/perl6/book 15:03
15:03 rdice joined
nopaste "fperrad" at 77.206.50.26 pasted "[PATCH] Lua PMC with pcc" (625 lines) at nopaste.snit.ch/18413 15:03
"fperrad" at 77.206.50.26 pasted "I try without to convert Lua to pcc" (3 lines) at nopaste.snit.ch/18414 15:04
whiteknight NotFound: I can't believe you are doing it in C++ 15:09
NotFound whiteknight: Is my favourite tool. 15:10
whiteknight power to you!
excellent tokenizer
NotFound At least, until winxed get powerful enough to bootstrap. 15:11
whiteknight yes
NotFound Note that it uses a infinite memory gc ;) 15:12
whiteknight NotFound: there is an ECMAScript compiler somewhere, any thought on that? 15:14
NotFound whiteknight: I worked on it sometimes, but is too bitrotten. Need attention from someone used to the P6 classes used. 15:15
whiteknight okay, so a new JavaScript implemenation was necessary
Coke wishes github's fork button had a "are you sure" on it.
NotFound Winxed is not javascript, just uses a familiar syntax. 15:16
whiteknight ok
NotFound Maybe, just maybe, some day can be javascript compatible. 15:19
Coke *::YourNameIsPudge 15:20
dalek kudo: 37956bf | duff++ | (3 files):
Updates in preparation for release
 15:26
whiteknight NotFound: so winxed is just a compiler for your language to PIR? 15:35
it doesn't appear to link to Parrot directly at all
NotFound whiteknight: no, just execute parrot after compiling, unless options to just compile are used. 15:37
dukeleto 'ello 15:38
Tene hi leto
whiteknight hello duke
dukeleto Tene: good localtime()
Tene dukeleto: AFK 15:41
dukeleto msg darbelo if you do "git config branch.master.rebase true" in your plumage git repo, we won't see when you merge as a commit 15:45
purl Message for darbelo stored.
Coke away afk
ww
dalek p-rx: 076c241 | pmichaud++ | src/ (3 files):
Refactor protoregex token list generation a bit, add some dump methods to
 15:52
p-rx: adf59e2 | pmichaud++ | src/ (3 files):
More refactors, add in ability to build peek tokens from subrules.
p-rx: 8ffa9b0 | pmichaud++ | src/ (3 files):
Fix secondary sorting of protoregex candidates.
fperrad I try without success to convert Lua to pcc 15:56
16:00 payload joined
cotto_work good day 16:19
purl every day above ground is a good day
dalek kudo: 281e3b2 | duff++ | t/spectest.data:
Uncomment regressed spectests
 16:22
16:26 mikehh joined
whiteknight "Every day above ground is a good day" <- I used to have a boss who said that 16:36
and he was the most evil, rotten sonofabitch I ever met
16:38 donaldh joined
ash_ whiteknight: i know you made a blog post about someone working on a libjit branch, i was wondering if you think that person would mind if i took that branch and changed it to use llvm's jit to see if there are any major differences (speed or otherwise) 16:44
whiteknight ash_: we want a framework that supports both 16:45
so we won't want to delete the libjit stuff, but add some #ifdefs and support both with configure flags
but yes, it is a project that we want people to work on
if you wrote a ground-up patch to add llvm frame builder support to trunk, we could integrate it with the libjit work later 16:46
cotto_work whiteknight, how hard is it to write a frame builder? It seems like plobsing++ hacked one out pretty quickly. 16:47
whiteknight cotto_work: shouldn't be too hard, for a naive implementation
that is, we're not doing any kind of aggressive optimizing on these, because building NCI frames happens infrequently 16:48
japhb Austin, were you trying to reach me earlier? 16:56
17:21 jan joined
particle . o O ( whiteknight worked for purl? ) 17:30
whiteknight ?
particle good day 17:31
purl every day above ground is a good day
cotto_work maybe it was in Soviet Russia
17:39 rdice joined 17:41 bacek joined 17:47 bluescreen joined 17:51 mokurai joined 17:54 fperrad_ joined 17:57 chromatic joined
dukeleto purl, good day is also I love the smell of napalm in the morning. It's the smell of victory. 18:06
purl okay, dukeleto.
18:18 desertm4x joined
whiteknight good day 18:19
purl every day above ground is a good day or I love the smell of napalm in the morning. It's the smell of victory.
bacek Good morning (fsvo good) 18:24
whiteknight hello bacek
bacek g'day whiteknight
whiteknight dukeleto: I've never much appreciated the smell of napalm
cotto_work hi bacek 18:25
I
've never smelled napalm.
bacek hody cotto_work
whiteknight cotto_work: for the best. Nothing good comes from it
cotto_work notes he should probably avoid typing apostrophes.
whiteknight incredibly easy to make some, if you're interested
chromatic ''''''''''''''''''''''''''''''''''' 18:26
I have extra.
whiteknight I hit return far too often when I'm reaching for the apostrophy 18:27
cotto_work realizes that chromatic wasn't saying that he has extra napalm.
whiteknight something about my stupid, awkward little pinky not being accurate as I would like
now my index finger, it's like a damn sniper rifle 18:28
japhb "Bad Pinkie, No Biscuit!"
18:39 payload joined
cotto_work Is TT #1134 with a post-merge parrot? 18:42
apparently not 18:43
whiteknight I think it's pre-merge. It was generated with Rakud
and Rakudo doesn't work yet with post-mege parrot
cotto_work I just noticed. 18:44
It doesn't even build, which is at least a nice early failure.
whiteknight considering the goofyness of the backtrace, I'm not convinced it uses any real version of parrot
cotto_work gen-parrot++ 18:45
chromatic I like this analysis: dustin.github.com/2009/10/22/memcac...tcard.html 18:55
18:55 darbelo joined
cotto_work It's funny to think of 20 open bugs being a large number. 19:03
dalek p-rx: 981e5ea | pmichaud++ | src/ (3 files):
Allow prefix token peeks to traverse through initial subrules.
 19:04
p-rx: 3b82660 | pmichaud++ | src/ (3 files):
Rename some internal methods to be more consistent.
rrot-mysql: r11 | julian.notfound++ | trunk/parrot-mysql.json:
update plumage metadata
 19:06
whiteknight Parrot isn't nearly stable enough that we could guarantee all crashes and failures reported be fixed within a month 19:09
but a part of that is that Parrot has a much larger external-facing API then memcached does
cotto_work I'm not saying it should be a goal for us. 19:10
chromatic It'd be interesting to graph the mean time to fix reported bugs though.
cotto_work There are major differences in maturity and scope between Parrot and memcached.
whiteknight I suggest that the majority of bugs are still open
cotto_work you could probably implement memcached in parrot (why is another question) 19:11
darbelo We still have frekin' RT tickets to close, that doesn't make us look too good in the 'speedy resolution' category.
whiteknight many of those open bugs are far too vague to ever be closed 19:12
I tried to close a bunch of vague old tickets, was rebuked, and stopped doing it
NotFound Lots of projects have few bugs because they are speedy to close with "Not a bug" or "Wontfix"
chromatic Anything older than a couple of years that we can't reliably reproduce and can't get feedback from the reporters within a couple of weeks seems like a viable candidate.
cotto_work NotFound, have you considered generating PAST from your compiler? 19:13
NotFound cotto_work: taking into acoount that I don't have a clue about PAST, no.
pmichaud_ looks forward to writing some books containing clues about PAST. 19:17
Coke rt slooow 19:18
19:19 KatrinaTheLamia joined
chromatic Hm, gcc *does* perform tail call elimination. 19:21
nopaste "chromatic" at 72.87.39.97 pasted "-t" (30 lines) at nopaste.snit.ch/18417 19:23
chromatic Hm, that didn't work like I expected.
Likely-case first attribute pool allocation (for whiteknight): nopaste.snit.ch/18417 19:24
whiteknight Yeah, I couldn't figure out whether we should look in the freelist or the fresh arena first, so I didn't pick a default 19:25
but that patch is superior in either case
chromatic Freelist first, for sure. 19:26
whiteknight depends on lifetime of the program, I think 19:27
but in general, you're probably right
chromatic Upwards of 80% I'm right, I'm certain. 19:28
I can imagine pathological cases where running full GC *won't* put anything on the free list, but that's a very complex program almost certainly designed to build and keep a full object graph around.
whiteknight sounds like a sane default to me. Ship it
chromatic Even so, you have to do a lot of work to get away from internal-use GCables that we're likely to free.
cotto_work see also: programming satan's computer 19:29
darbelo Hm. What should Parrot_str_byte_length return with a NULL argument?
chromatic I don't mind a modest penalty for a single-subroutine PIR program that builds a highly-cyclic object graph.
whiteknight darbelo: What does it do now in trunk, or what should it do theoretically?
darbelo segfault. 19:30
purl well don't DO that, then.
darbelo Parrot_str_byte_length(SHIM_INTERP, ARGIN(const STRING *s))
return s->strlen;
whiteknight ah, so that's probably the worng thing
dalek rrot-mysql: r12 | julian.notfound++ | trunk/Makefile:
fix library dir
darbelo We could fake it and return 0
whiteknight right now, I think it should return 0 or -1 (ask allison)
in the world of STRINGNULL, no issues 19:31
darbelo A NULL pointer in *not* a string, we shouldn't return a valid length for it.
chromatic That reminds me, STRINGNULL would be nice for 1.8.
whiteknight chromatic: estimate, how much effort to do that?
cotto_work chromatic, would that just be a sentinel value? 19:32
whiteknight cotto_work: it would be a global singleton, like PMCNULL
an immutable global singleton
Tene you know what I realized we're really lacking in terms of production quality? IO.
cotto_work ok
whiteknight Tene: agreed
NotFound whiteknight: PMCNULL is able to catch misusages because it has a function table. 19:33
chromatic Adding STRINGNULL is reasonably easy.
whiteknight NotFound: A STRINGNULL would be a STRING structure with properly initialized fields
chromatic Using it everywhere we need it would be less so.
whiteknight it could be used transparently for most cases like a normal STRING 19:34
chromatic Exactly my thoughts.
NotFound whiteknight: that way error will be hard to catch. At least a segfault is viewable.
whiteknight and a if(STRING_IS_NULL()) {} for other cases
Tene I keep considering starting to tackle the IO tasklist. 19:35
whiteknight without STRINGNULL, segfaults are inevitable. With it, no so
Tene: I got stalled on pipes. I don't know a lot about pipes. So I'll be happy to work on IO if pipes can get implemented
I'll be happy to help somebody who is more knowledgable then I 19:36
dalek p-rx: b7a03fc | pmichaud++ | src/NQP/Grammar.pm:
[nqp]: Make statements smarter about detecting end of block/arglist/compunit.
 19:38
whiteknight I think that Parrot's huge bug list actually acts like a detriment. It's so large and unmanagable that people ignore it 19:40
plus there aren't clear priorities that people can focus on, so some failures never get the attention they deserve
chromatic We have 493 open TTs. That doesn't seem huge.
whiteknight not huge, but not reasonable either
dalek rrot: r42024 | chromatic++ | trunk/src/gc/mark_sweep.c:
[GC] Rearranged branches in Parrot_gc_get_attributes_from_pool() in descending
 19:41
chromatic 492
dalek TT #1062 closed by chromatic++: Type safe mark alive
whiteknight this is why I like the idea of testing and bug-fixing hackathons before a release
otherwise there's no motivation to triage through the ticket queue 19:42
chromatic Half of the GC bugs look closeable to me. 19:43
darbelo The "That will be fixed when we unfuck subsytem 'Y'" problem doesn't help.
Tene I know pretty much nothing about pipes... maybe time to start learning.
whiteknight Infinoid had done some work in the io_cleanps branch at one point. I'll try to give you his patch so you can look at it 19:44
japhb Our open ticket count is in the same range as the release-critical bug list for the entire Debian "testing" distro. Mind you, those are RC bugs. But they're also across tens of thousands of packages ....
chromatic mikehh, is TT #1110 still an issue for you? 19:46
ttbot Parrot trunk/ r42024 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118997.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 19:47
whiteknight I personally would like to see a more comprehensive roadmap, with a more standard process for putting items onto it 19:48
then a lot of "Implement feature X" tickets can just be turned into items on the roadmap
I would also like to find a way to group tickets better by subsystem, and produce a report that shows the grouping 19:49
that way we can see which systems need work, and which systems are in the most trouble (to inform large-scale refactors and fix operations)
chromatic Someone should remove files we've removed from the exception list in t/codingstd/c_function_docs.t. 19:51
We should also decide when we're going to remove src/gc/res_lea.c.
darbelo chromatic: what stands in the way of that?
whiteknight nothing 19:52
darbelo votes for 'now' then.
whiteknight do it. Seriously 19:53
19:53 soxet joined
whiteknight print the file out, burn the paper, then delete the file 19:53
dalek TT #533 closed by chromatic++: headerizer not headerizing src/gc/gc_malloc.c 19:54
19:54 hercynium joined
chromatic TT #800 needs tests. Who likes to write tests? 19:54
darbelo Wait, this is the malloc gc. Didn't I kill it already? 19:55
whiteknight chromatic: assign it to me 19:58
I need to increase my test-foo
19:58 joeri joined
chromatic No need for assignment if you hop to it now! 19:59
whiteknight can't do it now. At work and about to pack up and go home
actually...packing up and going home now. Later 20:07
mikehh chromatic: checking 20:11
chromatic: yes g++ 4.4.1 fails on building - src/string/api.c:2801: error: invalid conversion from ‘const char*’ to ‘char*’ 20:19
last time I checked g++ 4.3 worked in Ubuntu 9.04 - this is in 9.10 (beta - updated)
but I think the objection might be valid as the sub modifies the so called const string passed 20:21
it definately looks that g++ 4.4.1 catches things that slip through earlier compilers
20:22 barney joined
cotto_work hi barney 20:22
NotFound strchr is a know source of that type of problems.
barney hi
purl salut, barney.
chromatic Thanks, mikehh. I think I can work around it. 20:23
darbelo We shouldn't be using strchr at all. STRINGs can have NULL bytes in them.
bacek chromatic: ping 20:24
NotFound darbelo: strings intended to be unescaped, I hope not.
cotto_work Hmm. It sounds like we should have more tests in which STRINGs have non-terminal null bytes.
mikehh what worries me is how many other problems it is going to pick up - that's the first routine it tries to compile after the tools/build/c2str.pl
nopaste "bacek" at 114.73.173.198 pasted "Small patch for PCC to reduce GC pressure for chromatic++ for review :)" (722 lines) at nopaste.snit.ch/18418 20:25
NotFound mikehh: do a grep for strchr, will be a good estimation ;) 20:26
dukeleto small patch != 722 lines
bacek dukeleto: :)
cotto_work It's a Russian "small" apparently. 20:27
chromatic There's a push_pointer vtable?
darbelo NotFound: You are right, I hadn't looked at the function being discussed.
bacek chromatic: I just added it. 20:28
Sorry, forgot to put it into diff
mikehh ack strchr from the parrot root gives about 20 instances
chromatic How's it benchmark?
bacek chromatic: didn't tested yet, rebuilding with --optimise now 20:29
NotFound darbelo: BTW, you asked yesterday if I posted Winxed somewhere. Done, code.google.com/p/winxed/ 20:30
dukeleto bacek: you mean --optimize. those brits don't know how to spell 20:32
nopaste "bacek" at 114.73.173.198 pasted "Full version of diff" (742 lines) at nopaste.snit.ch/18419 20:33
bacek darbelo: nope, and it's colour and gaol :) 20:34
mikehh dukeleto: hey youse guys messed that up and didn't complete the project completely
japhb Plumage team: I need some design input. Rakudo (and I assume probably the other HLLs) will soon have rather tight and possibly rapidly-changing version requirements on Parrot and especially nqp-rx. I had initially been assuming we would add version requirements directly to the metadata files, but for an HLL in rapid development, the lag between the project's requirements changing and when someone gets around to updating the metadata is u 20:36
nacceptable. So now I need some way to update dependency requirements after fetch (or possibly after configure). I've got a couple very rough ideas on how to specify in the metadata that some dependencies are dynamic, and what to require of a project to state these dependencies, but I feel like I'm in a local minima in the solutions space, and I'd like suggestions if y'all have any.
bacek chromatic: 1M less pmc allocations 20:37
chromatic: on fib.pir 20:38
darbelo NotFound: Saw the link in the irclogs. Haven't looked at the code yet.
NotFound Don't look, just compile it ;) 20:39
darbelo I'll try it later. 20:42
japhb darbelo, dukeleto, Tene? Any ideas on my quandry above? 20:43
Tene lemme look
japhb: you already have a solution in the design space somewhere... have the HLL host a metadata file. 20:44
japhb If it helps to get the mental juices flowing, I can explain my ideas, but I just didn't want to poison everyone's thinking processes.
NotFound japhb: having the posibility having an updated metadata in the fetched package and use that will be nice.
Tene then just refer to that metadata file as included or whatever 20:45
japhb hmmm, Tene, NotFound, that definitely sucks less than my idea. 20:46
NotFound BTW the writable check on my system fails.
japhb NotFound, oh?
NotFound, why does it fail? 20:47
bacek chromatic: about 3% on fib.pir. On something heavier (with returning more results it will give more)
NotFound It uses sudo, but my parrot is installed under my home dir
japhb NotFound, so clearly it should not have needed the sudo ...
Hmmm. Debugging assistance greatly appreciated. 20:48
Assuming you have a signed PaFo CLA in, I can just add you to the repo as a committer. :-)
NotFound japhb: there is some way to execute the test isolated?
Uh.... I forgot again to send the CLA. 20:49
japhb NotFound, should be able to run parrot_nqp from the command line, load_bytecode() the helper libraries, and then use the writable test.
darbelo Grandfathers and their forgetfulness... 20:50
japhb There will eventually be tests for it, but I don't think dukeleto's had a chance to catch up with my recent additions.
NotFound I'm not a grandfather, I don't even have a child 20:51
davidfetter ...that you know of 20:52
darbelo NotFound: Pun on 'grandfather cluse'. 20:53
20:53 Austin joined
NotFound Will you still give me a commit bit when I'm sixty-four? 20:54
ooo oooo ooo 20:55
japhb We'll still need you. Don't know about feeding you, though.
Maybe buy you a $beverage
NotFound Good enough
chromatic Hm, 1M fewer PMCs is good, but 3% doesn't seem like a big improvement for adding one vtable entry. 20:57
cotto_work as a trade-off, we could kill several others.
chromatic ... especially if that vtable entry goes unused after we merge Context and CallSig. It may or may not, but .... 20:58
darbelo Mark it 'experimental' and kill as soon as you're done with it.
chromatic How long will it last, though? 21:01
darbelo It depends on how soon can context and Callsig be merged. 21:04
I think that blocked on the deprecation policy. So, february?
chromatic I don't think that has to be the case. 21:05
cotto_work I don't think anything prevents that from happening now. We might need a trivial PMC that extends the (yet-nonexistent) merged Context. 21:06
chromatic If we removed CallSignature entirely, we could add a fake CallSignature that does nothing more than inherit from Capture again.
bacek we still need something to replace PRA of CPointers 21:14
21:15 colomon joined
Coke chromatic: how small is very small for stack size? 21:17
cotto_work btw, I can't reproduce the crazy stack trace in TT #1134 on my x64/linux box 21:23
has anyone else tried? 21:24
bacek bah... Looks I compared non-optimised build of CallSignatureReturns with optimised build of trunk. 21:28
chromatic Coke, 35 mb, perhaps. 21:29
cotto_work bacek, looks like a ~5% improvement
(4926938130-4488471358)/4926938130 21:30
purl 0.088993764571588
cotto_work or that
21:30 Whiteknight joined
cotto_work (860015945-794708479)/860015945 21:30
purl 0.0759375060191471
cotto_work fib and oofib, respectively 21:31
bacek 8.8 and 7.5%
purl 8.875
japhb not bad
cotto_work now let's have a look at this code which I downloaded over http, built and ran without checking it first 21:32
darbelo cotto_work: was it a gift from the greeks? 21:33
cotto_work bacek, do you have any Greek blood? ;) 21:35
bacek cotto_work: I've got Greek name :)
21:35 payload joined
darbelo realizes he doen't know bacek's name. 21:35
chromatic I thought it was Sky Net. 21:36
bacek (7382453772 - 6546974232) / 7382453772 * 100
purl 11.3170981600831
bacek yay, 11% on my box for fib.pir
darbelo Oh, is that your serial number?
;)
bacek darbelo: Vasily :) 21:37
Austin darbelo: bacek.name eq 'Vasily Chelkin', modulo spelling errors.
bacek Austin: quite few of them :) Actually Chekalkin 21:38
Austin My bad. 'Vasily Chekalkin'
bacek Austin: no worries
darbelo Hm. Is it a spelling error to use the wrong character set?
Austin bacek: What? Manhattan distance was only 2. 21:39
:)
darbelo: No, it's a transcoding. There's a method for that in Parrot.
bacek Austin: Levenstein :)
japhb bacek: "edit". I hate naming concepts after people when the person's name has no intrinsic link to the concept. 21:40
Austin If people just remember the name, that won't be a problem. 21:41
Hey, I have a plumage problem. How do I report it? 21:42
japhb Austin: People are not wired that way. Computers are.
Austin, you talk to me. :-) Or do a parrot TT, and assign it to me.
bacek japhb: what do you mean about "concept"? 21:43
Austin japhb: Au contraire! People are definitely wired that way. Nobody in the world knows WTF amotropic lateral sclerosis might be, or how it differs from other lateral scleroses, but if you mention Lou Gehrig's diease you'll get a whole different response.
japhb bacek: if you did not already know, how would you remember whether "Hamming" or "Levenshtein" distance was the one you meant?
Austin bacek: He's talking about Levenshtein distance vs. "edit distance". 21:44
japhb: (plumage) I'm getting "invoke() not implemented in class 'Undef'" after fetching and trying to configure a library.
dalek rrot: r42025 | darbelo++ | trunk (9 files):
Remove unused file src/gc/res_lea.c and cleanup mentions in the documentation.
 21:45
japhb Austin, I did not mean "people can't remember names". I should have been more clear. I meant "people can remember more things, more quickly, when the name tells you something about the concept, as opposed to when the name is arbitrary."
bacek japhb: Levenshtein. Or my brainz is totally faded and I forgot first year CS stuff...
darbelo Like, how we all know how Ohm such a resistive dude that they named the unit after him.
;)
Austin bacek: Where and when did you go to school, that this is first-year stuff? Most grads now-a-days can barely spell recursion after 4 years. 21:46
pmichaud_ and Volt was known for his transportation abilities? ;-)
nopaste "bacek" at 114.73.173.198 pasted "Test summary with CallSignatureReturns" (12 lines) at nopaste.snit.ch/18420
bacek Austin: may it was second year. Kazan State University in Soviet Russia :) 21:47
darbelo Ouch, packfiles.
bacek if someone want to polish my patch - feel free to commit. I have to go to $dayjob.
japhb bacek: Of course you are correct. about which is which. I'm saying "How hard was it to remember which is which, and would it have been easier if they were referred to as 'substitution distance' and 'edit distance'?"
pmichaud_, Volt and Ohm are single nouns. I am referring to adjectives in part of a noun phrase. 21:48
bacek japhb: may be. But Levenshtein was Soviet scientist, so it's not problem for me to remember his name :)
japhb LOL
OK, I can't argue that logic.
nopaste "Austin" at 98.235.55.43 pasted "Plumage dies on 'configure kakapo'" (23 lines) at nopaste.snit.ch/18421 21:49
darbelo kakapo? 21:50
Austin a library.
purl i think a library is Crypt::OpenSSL::CA
japhb I just remember junior year math being assaulted with a whole bunch of metrics and spaces and such named after various people and going "Ack! Your Hilbert is eating my Reimann!"
Austin Kind of like glue.pir, except somewhat larger.
bacek packfile tests are failing because there is no annotation segment by default anymore.
NotFound "We think you have the Stranessfoun disease" "Is bad, doctor" "We still don't know, Mr. Stranessfoun"
Austin NotFound: :)
japhb :-)
pmichaud_ how do you feel about mythical names, then? 21:51
japhb pmichaud_, try me
pmichaud_ "Apollo" "Gemini" "Mercury"
(lunar program missions)
japhb Sure. Which was first, Gemini or Mercury? How do you know that? 21:52
21:52 patspam joined
japhb What differentiated the two mission series? 21:52
pmichaud_ Mercury were the first attempts to get off the planet
Gemini referred to two-man missions
Apollo were the missions to reach the moon
Austin And Apollo were missions to the sun!
ttbot Parrot trunk/ r42025 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/119075.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 21:53
japhb AH! Now that actually has meaning. And you remembered Gemini, I assume, because of the twin information
Austin, nice
pmichaud_ A lot of the names have meaning that way, though.
Austin And lin 4850 is a .const Sub declaration.
*line
darbelo japhb: Like plumage. "The part of the parrot you actually look at."
Austin You guys need to strip annotations, I think.
pmichaud_ I've noticed that Parrot tends to do a lot of error reporting at sub declaration lines
japhb pmichaud_, but Austin really does have a point about Apollo 21:54
darbelo, huh?
darbelo Hidden meaning in names. You re guilti of it.
pmichaud_ perhaps Apollo (missions) were meant to go to the sun but they missed. :-)
maybe they should've gone to the sun at night instead :) 21:55
darbelo s/guilti/guilty/
japhb darbelo, oh heh. That was not my intent. But again, Plumage is not an adjective describing a noun. (Nor for that matter are the names of the space missions, but I was willing to overlook that :-)
Anyway ... back to actually solving someone's problem. 21:56
darbelo Austin: can we get a look at the metadata file? 21:57
Austin Sure.
japhb Gar. The problem is that the annotations from the NQP compile to PIR are replaced with the ones from the PIR compile to PBC. That ... bites.
Austin gitorious.org/kakapo/kakapo/blobs/m...akapo.json
pmichaud_ HLLCompiler (in nqp-rx) will have better backtrace reporting (like Rakudo does) 21:58
Austin The config file parses okay, and running 'plumage configure kakapo' causes it to do a git pull and report up-to-date before it whorfs. 21:59
21:59 joeri left
japhb Austin, you want 'perl5_configure', not 'perl_configure' 21:59
But the error message is LTA
Austin :) 22:00
japhb pmichaud_, oh thank you. Yet Another Reason to jump to nqp-rx as soon as it's ready
Austin That worked. japhb++ for hella fast.
bacek hooray, I fixed extend_18
cotto_work lta?
less than awesome? 22:01
pmichaud_ I'm aiming to have it ready nlt Nov 10
Austin adequate
darbelo pmichaud++
pmichaud_ I'm thinking that nqp-rx release cycle will be second tuesday of each month (1 week before parrot)
darbelo pmichaud++
Austin (That 'adequate' was to cotto_work, not pmichaud)
japhb cotto_work, precisely. Noticed it in #perl6, think it fits well. Especially with chromatic's
Austin pmichaud: Is that one week before, or three weeks after? 22:02
japhb '1. Don't break the build. 2. Increase the awesome.' rules
NotFound japhb: pir lines are not reported by hll annotations, they are handled other way
japhb NotFound, ack. Twice today with the Mental Model Fail.
pmichaud_ Austin: intended to be one week before, mainly 22:03
japhb Sadly, my mental models usually include 'this was not implemented as a hack'
nopaste "bacek" at 114.73.173.198 pasted "Final patch for CallSignatureReturns." (791 lines) at nopaste.snit.ch/18422
pmichaud_ that way parrot can potentially include nqp_rx in its bundles
Austin Ahh. I though you were allowing time to pay the upgrade tax.
Whiteknight pmichaud_: I'm still wondering why nqp-rx wouldn't just be merged into Parrot trunk 22:04
bacek Whiteknight: "deprecation policy"
Whiteknight it seems like that's the most "natural" place for it
japhb Whiteknight, IIRC, it's the deprecation policy.
pmichaud_ deprecation policy is one reason
japhb DP == scares people
NotFound A problem is that there is no way to annotate the sub declaration, so for example bad number of params report can't show the hll line number.
pmichaud_ another is that I expect nqp-rx to potentially target other backends someday
darbelo Austin: parrot HEAD can track nqp-rx HEAD, for almost no tax. 22:05
japhb NotFound, AH! I always wondered why that error was so "barren"
cotto_work pmichaud_, like jvm?
Whiteknight I worry that the deprecation policy seems to cause more harm then good
pmichaud_ cotto_work: yes, or clr
moritz or perl 5 ;-)
pmichaud_ or p5
japhb Whiteknight, *cough* *cough*
cotto_work or c even
bacek chromatic, Whiteknight, cotto_work nopaste.snit.ch/18422 All tests apart from packfiles passed. Feel free to commit in trunk
pmichaud_ anyway, Parrot needs to come to terms with being able to bundle external modules, and nqp-rx is a natural place to start
japhb Well, that and plumage. 22:06
Since according to last #ps, parrot will ship a snapshot of PIR-compiled plumage.
Whiteknight pmichaud_: so long as HLL compilers can get to it and make use of it without jumping through huge hurdles, I wont worry about it
pmichaud_ Whiteknight: I think it'll be easy to do, yes.
and I'm somewhat expecting parrot to ship a snapshot of PIR-compiled nqp 22:07
thus the "one week before parrot release" timing
anyway, separating projects and modules offers much more degrees of freedom
Whiteknight okay
NotFound We can even ship pir-compiled HLL written tests 22:08
A lot of tests will be easier to write in a HLL 22:09
moritz if you are sure the HLL is implemented correctly, that works 22:10
darbelo Yes, but you are restricted to languages with a small-ish runtime.
NotFound That's one of the reason why Winxed has 0 runtime ;) 22:11
dalek rrot: r42026 | chromatic++ | trunk/src/string/api.c:
[string] Worked around C++-foiling constness-losing assignment in
darbelo Also "good for testing" ne "Good for debugging" 22:12
Ask coke about it some day :)
He can segfault parrot in 2 lines of TCL, and we still have no way to fix it. 22:13
NotFound darbelo: that's one of the reasons why Winxed does not use HLL map X-)
darbelo NotFound: What's teh reason winxed doesn't have a Configure stage? 22:16
:)
dalek TT #965 closed by coke++: segfault in Parrot_convert_arg
NotFound darbelo: lazyness
purl lazyness is, like, not appreciated here. or weeshcomic.com/archive.php?strip=2009-04-20
darbelo NotFound: Pick an implementation language and I'll contribute one.
chromatic mikehh, r42046 may help you. 22:17
ttbot Parrot trunk/ r42026 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/119118.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 22:18
NotFound darbelo: I'm not sure I want to depend on some external tool... maybe autoconf will be appropiate 22:19
Austin I think there needs to be a "plumage remove" option.
Whiteknight I think I'll try to get :call_sig implemented tonight
Austin I've got it jammed already - a git merge required - and there's no way out without going into .parrot by hand. 22:20
chromatic Didn't you promise to write tests for me first, Whiteknight?
Coke chromatic: I tried a ulimit -s 1 (in kb) and saw no difference in failure mode.
chromatic The difference should be that the backtrace should be much shorter.
It'll still fail the same way (running out of stack space), but it'll be easier to see how it started that loop. 22:21
The end of the loop isn't significant for debugging purposes.
darbelo NotFound: Hm, you are probably right. Even if I hate autoconf. 22:22
Austin W00t. All the way to make install. 22:23
NotFound darbelo: anyway is not an urgence ATM. Even if someone wants to use it on a platform not supported by the compiler, the generated pir can be used.
darbelo NotFound: Yeah, but I'm lazy enough that removing -Wextra from the makfile seemed like work. 22:25
Austin: have you added kakapo to the ParrotModules wiki page? 22:26
Austin Nope.
darbelo Want me to?
Coke chromatic: the backtrace was still several thousand lines. 22:27
NotFound Mmm... Maybe a fisrt step to bootstrap can be used. Write a simple Configure stage with self, and store the pir generated by it.
Whiteknight chromatic: did you assign me that ticket?
(and I don't make promises)
darbelo PASSED 33 tests in 10 files
NotFound: That's for OpenBSD -current i386 22:28
NotFound darbelo: good :)
I've tested only on linux.
darbelo After the mentined -Wextra removal. OpenBSD ships a 6 year old gcc as it's system compiler. 22:29
chromatic I didn't assign it.
Whiteknight well I certainly don't remember the TT #
darbelo Whiteknight: irclogs ;) 22:30
japhb bak, backlogging
chromatic TT #800
NotFound darbelo: I initially planned to write in C++ 0x, but discarded the idea because few system support that
cotto_work NotFound, do any systems support that? 22:31
NotFound So instead of using smart pointers I just leak lots of memory
Whiteknight darbelo: so not only do I need to fix the ticket, but I also need to find it myself?
NotFound cotto_work: gcc 4.4, I think
darbelo Hm. Count me out then. The most recent gcc in ports is 4.2.4 22:33
japhb Austin, 'plumage remove' is (already) in the TODO. :-)
cotto_work darbelo, he said he's *not* using C++ 0x 22:34
NotFound Anyway, the infinite memory model is good enough as a start
darbelo cotto_work: I know, I'm his first tester ;)
cotto_work: I was refering to c++0x
Which is going to need a new name in afew monts :) 22:35
22:37 kid51 joined
dalek rrot-plumage: 374ab5f | austin++ | :
Added metadata/kakapo.json describing kakapo library.
 22:43
dukeleto mikehh: what did I mess up? 22:44
chromatic C++0x has another five years to release.
22:44 nopaste joined
dukeleto japhb: no time to backlog when I am not highlighted 22:44
msg me plz
me messages
NotFound chromatic: yes, but there is consensus about several usable things 22:45
darbelo Austin: You got a plumage bit? congratulations! 22:46
Austin I got a whaT?
japhb dukeleto, what did I forget to highlight you about?
japhb confused
purl You won't be after this episode of Soap!
darbelo plumage commit bit?
Austin Not me. No CLA.
dukeleto <@japhb> darbelo, dukeleto, Tene? Any ideas on my quandry above? 22:47
dalek tracwiki: v4 | darbelo++ | Modules
tracwiki: Add Austin's kakapo library.
tracwiki: trac.parrot.org/parrot/wiki/Module...ction=diff
Whiteknight that reminds me, I'm supposed to fill out an updated CLA
darbelo Hm, last plumage commit said ' Austin++ ' on top.
japhb dukeleto, ah. Um. That's why I highlighted you. ;-)
Austin It was a clone, probably.
22:48 Limbic_Region joined
darbelo slaps his forehead. 22:48
Right, on git committer != author.
Austin DVCS, baby. Nothing is what you think it is. 22:49
chromatic I meant only for the version number. C++0xA is 2010.... 22:50
Austin One of you guys should create a plumage 'team' on gitorious.
plumage-birds, or some such.
(Although -birds might not fly in the uk)
darbelo teap;) 22:51
eh. ;)
Austin plumage-psittacines ?
darbelo plumage-team ?
Austin lame
If you can't spell it, you can't join. 22:52
What a slogan!
purl it has been said that a slogan is annoying
Austin purl, parrot?
purl well, parrot is our teacher, our mother, our secret lover or the reason Dan started or the reason Dan left or pretty onionish:) or www.youtube.com/watch?v=4vuW6tQ0218...re=related or regular releases and carefully following the deprecation policy
cotto_work s/following/trying to work around/ 22:53
22:57 darbelo left
dalek rrot: r42027 | jkeenan++ | trunk/config/auto/gc.pm:
Correct spelling error in documentation.
 23:01
ttbot Parrot trunk/ r42027 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/119171.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 23:09
23:10 theory joined
japhb Austin, the error message for mistyped metadata instructions types should now be SMA (Significantly More Awesome) 23:15
Austin Sweet. I hope to never see it, of course. :)
japhb ... but still not perfect. Due to control flow issues, you will see the error twice in some cases.
:-)
Austin That's fine. 23:16
I hope to never see it twice, also.
japhb heh
23:16 plobsing joined
dalek rrot-plumage: 393057e | japhb++ | :
[plumage] Add additional stage to metadata_valid and refactor; refactor ...
 23:16
23:17 Zak joined
japhb pmichaud_, what are you planning vis-a-vis exceptions and exception handling in nqp-rx? 23:18
23:18 Austin_Hastings joined
japhb Plumage control flow would be a lot less of a mess if I had proper exceptions. 23:19
plobsing hi #parrot
cotto_work hi plobsing
Austin_Hastings :)
I thought you had implemented throw and try ? 23:20
japhb Or rather, syntax for same, and support for some way to determine different groups of exceptions without checking the string value of error messages.
Austin_Hastings Hello, plobsing
Ah.
You need a multisub.
:)
And a tree of exception types, all alike. 23:21
japhb Austin_Hastings, yeah, I do have die and try, but they are not real syntax, and thus ... kinda suck. In particular, without even closures, try is a humongous hack.
treed Watch out for those catch-grues.
Austin_Hastings Hmm. Where's my nick?
23:22 Austin_Hastings joined
japhb I wonder how many guys whose real name is 'Nick' use that for their IRC nick. 23:22
Austin_Hastings Meh.
cotto_work Austin_Hastings, do you need your duplicate kicked?
Austin_Hastings If possible, yes. 23:23
But maybe not.
Austin Ahh.
cotto_work I guess not.
Austin So much less formal.
Whiteknight needs to find a new editor 23:24
Kate is so buggy that it's unusable
particle suggests chromatic
Austin notepad++ on windows.
dukeleto suggests particle
Austin Scite on Linux.
japhb Whiteknight, really? I'm kinda surprised. I (used to) hear good things about Kate. 23:25
Whiteknight I use notepad++ on windows. That's not a question
Austin Ed! Ed is the standard!
japhb Mmm, edlin ....
Whiteknight japhb: I heard good things too, that's why I switched
chromatic My only qualm with Vim is that sometimes when I hit backspace in insert mode, it reindents the current line which is exactly what I don't want it to do, and I don't know why.
Whiteknight I refuse to use vim or emacs
tried them, hate them, don't want to talk about them
dukeleto Whiteknight: HERESY!
cotto_work I find it quite amusing picturing Whiteknight doing all his typing via chromatic. 23:26
s/typing/editing/
Whiteknight :call_sig is turning out to be trickier then I thought
Austin Here's a question: is "null" an instance of type X ?
dukeleto Whiteknight: i use 'ed' religiously
japhb One of my previous bosses required his entire staff to switch to emacs. I never switched away, but never became expert at it either. Kinda waiting for Padre to become really excellent for all the languages I care about.
Whiteknight japhb: if Padre supported C and had better syntax highlighting, I would switch today 23:27
I'm very picky about my highlighting. It's a personality flaw
Austin Does padre run under linux, too?
dukeleto Whiteknight: go say that in #padre and it might happen
mikehh Whiteknight: Kate is my main editor - I don't seem to have many problems with it 23:28
Whiteknight mikehh: maybe it's just my installation, or my particular set of features
dukeleto Whiteknight: highlighting is extremely important, i agree
japhb Whiteknight, Yeah, I need C, C++, JavaScript, HTML, and Perl 5 done better than emacs to even consider it.
chromatic padre runs pretty well under Linux.
Whiteknight but it spits out error messages all the time to my console, there are graphical glitches, sometimes the cursor turns the same color as the background and I can't see it, etc
Austin: You say scite? 23:30
japhb dukeleto, darbelo, Tene: Any of you have unmerged branches that edit plumage.nqp? I'm about to move a lot of code around.
mikehh Kate - version 3.3.2 using KDE 4.3.2 23:31
Tene No.
japhb Tene: thanks
Austin whiteknight: It's what I use on linux, but there's a reason I stick with windows. :$
Tene japhb: that is, i don't; dunno about th eothers.
japhb Tene, understood.
Whiteknight I like scintilla, mostly
Tene I tried Kate a few times; couldn't stand it.
Same with gedit, a few other graphic editors... 23:32
Whiteknight I was using medit, which was simplistic but solid
I switched to Kate because I wanted a few more features, but then I got a bunch of bugs, so I want to switch again 23:33
Austin Japhb: There's a merge request for you. It has nothing to do with code. :)
mikehh I generally prefer the KDE stuff to Gnome
japhb Austin, ah, lemme check
Whiteknight wknight8111.blogspot.com/2009/10/text-edior.html
(I wrote a blog post about my editor search)
Austin *ediTor
Whiteknight damnit. my callsig patch is failing some tests 23:34
mikehh When I used M$ stuff my favorite editor was Crimson Editor - haven't used M$ Windows for a couple of years now 23:35
dalek rrot-plumage: 0e2470b | japhb++ | :
Merge commit 'refs/merge-requests/1891' of git://gitorious.org/parrot-pl...
 23:38
Austin Is there a single non-string way to store type identifiers? 23:39
IOW: If an object is an Object, then there is a Class PMC that can be looked up. But for RPA and Integer, it's a PMC type. What's the unification of those? 23:40
japhb OK, assuming I didn't completely mess that up, your merge request has been merged, Austin. Thanks!
Tene Whiteknight: I take it you're completely uninterested in comments on your blog post about how great vim is? ;) 23:44
23:45 davidfetter joined 23:51 darbelo joined
japhb Is there a safe cross-platform make syntax for extension rules ("turn any .pir file into a .pbc by doing this")? 23:55
darbelo No
There is not a safe cross-platform make syntax for anything. 23:56
japhb damn it
Austin You might try basic suffix rules.
.c.o and .SUFFIXES
That's what I'm using.
(And on that subject, you can look in kakapo/build/Makefile.in) 23:57
darbelo Some makes allow it. GNU make and BSD make, which gets everyone and me, in that order. But might not be handled correctly by other makes.
Austin My lowest-common denominators are HPUX make and MSFT nmake, and they both take it, I think. 23:58
japhb Needs to run on any make that Parrot cares about.
Austin (I'm not sure about .suffixes on HPUX, but it should be ignored if it doesn't work.)
japhb GNU make, BSD make, and nmake are different enough that signs are good ....
Austin What makes does Parrot care about?
japhb And HP-UX is from a different planet, so that's another plus 23:59
Austin Hell, dukeleto almost pulled the trigger on 1.7 with failing Windows tests the other night.
japhb Austin, I don't actually know for sure.
Austin Yeah.
darbelo Austin: That was an accident ;)
Austin Everybody makes all this noise about platforms, but I really don't think anyone is going to port Parrot to the playstation anytime soon. (And if they do, there's already linux there.. :)

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!