Parrot 2.7.0 "Australian King" Released! | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 13 tickets, merge outstanding branches, help NotFound clean up PBC tests
Set by moderator on 20 August 2010.
00:01 Psyche^ joined
whiteknight yeah 00:05
00:07 Patterner left, Psyche^ is now known as Patterner 00:11 nopaste joined, TonyC joined
dukeleto really quiet. 00:24
00:32 senf_statt_oel left
whiteknight Has anybody else been seeing some weird errors in NQP lately? 00:33
I'm having trouble narrowing-down failing test cases, but I've seen some bad behavior from it recently
I was seeing problems yesterday where the ~ operator was giving me an error about not finding an MMD candidate for PSP->P, and an issue today where a return inside a loop wasn't returning 00:39
Coke nope. been pretty stable lately. 00:42
jnthn whiteknight: You normally need to prefix:<~> one of the argumetns to infix:<~> if you get that error. 00:43
00:43 dngor left, kid51_at_dinner is now known as kid51
whiteknight jnthn: yeah, that's the solution I found 00:43
jnthn whiteknight: I hope to make that issue go away for good in upcoming NQP changes, though. 00:44
It is annoying.
whiteknight jnthn++ 00:48
cotto_work That'll be a nice change.
00:52 theory left
whiteknight goodnight 01:23
cotto_work 'night
01:24 whiteknight left
dalek rrot-linear-algebra: 02f1e10 | Whiteknight++ | setup.nqp:
revamp setup.nqp to not be as bad as it was. I am running into some weird NQP errors so it's not perfect
 01:27
rrot-linear-algebra: 6d76831 | Whiteknight++ | src/include/pla (3 files):
Add two new stub header files for optional bindings to BLAS (non-ATLAS) and LAPACK
rrot-linear-algebra: 476984f | Whiteknight++ | s (5 files):
start refactoring out the BLAS stuff into a wrapper library that will support multiple backends. The build is broken
rrot-linear-algebra: 0ba09c7 | Whiteknight++ | src/lib/pla_blas.c:
...and the file
01:38 dngor joined 01:44 eternaleye left 01:45 eternaleye joined, eternaleye left 01:46 eternaleye joined 01:47 eternaleye left, eternaleye joined
cotto ~~ 01:52
02:07 hercynium joined 02:14 ruoso left 02:18 kid51 left, tcurtis joined 02:25 contingencyplan joined 02:27 ruoso joined
Coke moritz: you in? 02:32
02:35 janus left 02:40 theory joined 02:41 theory_ joined, theory left, theory_ is now known as theory
dalek rtcl-nqp: 58b1c64 | coke++ | / (2 files):
track git revision in .revision for evalbot.
 02:59
rtcl-nqp: 3f60ff6 | coke++ | build/ (2 files):
Configure.pl --gen-parrot has been broken since config_lib.pasm -> .pir
03:07 janus joined
dalek rtcl-nqp: fa21fc2 | coke++ | src/Partcl/ (2 files):
Fixup handling of (top level) globals, ala $::tcl_library.
 03:28
03:40 theory left 04:08 jsut_ joined 04:12 jsut left
dalek rtcl-nqp: 8f4fed2 | coke++ | / (2 files):
Convert [file] to Austin++ style dispatch && implement [file dirname]
 04:49
rtcl-nqp: d55427f | coke++ | src/Partcl/commands/file.pm:
remove debug output, whoops.

note to self, don't push right away. :P
Coke ugh. what scope in PCT corresponds to "get_hll_global" ? "package" is "get_global"... 04:51
... it's package. Ok. Looks like the fact that I'm in an namespace is getting lost somewhere. 04:58
not JFW--
-> abed.
dukeleto can anybody shed light on this coredump? : gist.github.com/546982 05:00
it is caused by rakudo (PL/Perl6) throwing an exception from PL/Parrot 05:28
cotto What's the git term for metadata such as git-svn-id? (or more specifically, how can I get git log to make it show up?) 05:52
sorear "specially formatted commit messages" 05:53
and git log should show them by default
cotto But how would I specify it as part of a format string so it'd be easy to slurp into a db? 05:54
%b is close but it sometimes contains some extra info. I guess it'll be close enough. 05:58
will git log list all commits by default? 06:00
I'll have to read up on that in the morning.
msg cotto git log --quiet --format='===%n%H%n%s%n%b' is a good starting place 06:05
purl Message for cotto stored.
cotto 'night
06:05 uniejo joined 06:17 integral left, integral joined 07:02 tcurtis left
dalek nxed: r600 | NotFound++ | trunk/winxedst1.winxed:
remove a redundant assignment
 07:07
kudo: f8763b8 | moritz++ | src/Perl6/Grammar.pm:
add word boundary assertion after alphabetic infix operators

Fixes RT #77218		 07:13
07:14 fperrad joined 07:41 jsut joined 07:45 jsut_ left
dalek rrot: r48617 | NotFound++ | trunk/src/string/api.c:
fix silly mistake with the escape code
 07:52
07:52 integral left 07:54 payload joined, payload left, integral joined
dalek nxed: r601 | NotFound++ | trunk/token.cpp:
escape \\e in stage 0
 07:56
08:01 bacek joined 08:03 aloha joined
moritz std: any(-> $x {say $x }, -> $x ( say 40 + $x }).(2)' 08:32
p6eval std : ( no output )
moritz std: any(-> $x {say $x }, -> $x ( say 40 + $x }).(2)
p6eval std : ( no output )
moritz rakudo: any(-> $x {say $x }, -> $x ( say 40 + $x }).(2) 08:33
p6eval rakudo 819e2c: OUTPUTĀ«===SORRY!===ā¤Invalid typename in parameter declaration at line 22, near " 40 + $x }"ā¤Ā»
moritz huh?
08:33 lucian joined
moritz rakudo: any(-> $x {say $x }, -> $x { say 40 + $x }).(2) 08:33
p6eval rakudo 819e2c: OUTPUTĀ«2ā¤42ā¤Ā»
moritz www.perlmonks.org/?node_id=856849 another junctions-as-set abuse 08:37
08:43 cogno joined 08:48 zostay left 08:58 cogno left, tadzik1 joined, cogno joined
sorear moritz: I think that belongs over in #perl6 08:59
09:00 tadzik left
moritz oh yes 09:00
09:00 tadzik1 is now known as tadzik
moritz wrong window then :-) 09:00
09:04 lucian_ joined 09:08 lucian left 09:09 cogno left 09:19 lucian_ left 09:36 GeJ_ joined, GeJ left 09:46 GeJ_ is now known as GeJ
dalek tracwiki: v34 | bacek++ | GCTasklist 09:57
tracwiki: trac.parrot.org/parrot/wiki/GCTaskl...ction=diff
10:25 tadzik left
dalek tracwiki: v140 | moritz++ | Languages 10:30
tracwiki: lolcode works on current parrot
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
10:36 ruoso left 10:40 AzureStone left 10:44 nwellnhof joined 10:55 kid51 joined
mikehh kid51: ping 11:38
dalek rrot: r48618 | mikehh++ | trunk/src/call/args.c:
fix codetest failure - parentheses should not have space immediately
 11:46
11:55 ruoso joined
dalek rrot: r48619 | mikehh++ | trunk/t/pmc/testlib:
add svn:ignore for *.pbc in t/pmc/testlib
 12:03
12:05 lucian joined 12:11 bluescreen joined, kid51 left, integral left, TonyC left, Patterner left, dmagnus__ left, x3nU left, jhelwig left, dalek left, PerlJam left, pmichaud left, krunen left, sjn left, jjore left, snarkyboojum left, Ryan52 left, kthakore left, whiteknight joined 12:12 purl left, lucian left, contingencyplan left, perlite left, Coke left, KatrinaTheLamia left, dukeleto left, Util left, NotFound left, p6eval left, Infinoid left, elmex left, atrodo left, TiMBuS left, bacek_at_work left, cotto left, Maddingue left, sri left, Tene left, Khisanth left, confound left, frodwith left, dzoe left, mikegrb left, szbalint left 12:13 jjore joined 12:14 kthakore joined, sjn joined, PerlJam joined, athomason left 12:15 smash joined
smash hello everyone 12:15
mikehh hi smash
12:16 krunen joined 12:17 dmagnus__ joined
mikehh what's with (*.net *.split) and how do we get dalek, purl etc. back? 12:17
12:22 slavorg left, slavorg joined 12:23 Psyche^ joined, Psyche^ is now known as Patterner, integral joined 12:24 jhelwig joined 12:36 hercynium left 12:41 p6eval joined
moritz partcl-nqp: 1 12:41
p6eval partcl-nqp d55427: OUTPUTĀ«PackFile_unpack: This Parrot cannot read bytecode files with version 8.2.ā¤ā¤PackFile header failed during unpackā¤Ā»
moritz swears violently 12:42
anyway, commit hash is in tehre 12:43
12:48 ruoso left, ttbot left, confound joined, athomason joined, Tene joined, szbalint joined, frodwith joined, Maddingue joined, sri joined, dzoe joined, Khisanth joined, bacek_at_work joined, cotto joined, TiMBuS joined, atrodo joined, elmex joined, Infinoid joined, NotFound joined, Util joined, dukeleto joined, KatrinaTheLamia joined, Coke joined, perlite joined, contingencyplan joined, lucian joined, x3nU_ joined, TonyC_ joined, snarkyboojum joined, dalek joined, pmichaud joined, Ryan52 joined, ttbot joined, mikegrb joined, purl joined
moritz partcl-nqp: 1 12:49
p6eval partcl-nqp d55427: OUTPUTĀ«invalid command name "1"ā¤current instr.: '_block17' pc 47504 (src/Partcl/Grammar.pir:29)ā¤Ā»
Coke moritz: would it be possible to change it to partcl, even though I know you already asked that once and I said no? ;)
moritz Coke: sure 12:50
Coke partcl: puts what? 12:51
purl People Under the Stairs, one of the best underground hip hop duos
Coke purl, you crazy bot.
purl Coke: i'm not following you...
Coke purl, lurp?
purl coke: bugger all, i dunno
12:51 p6eval left
Coke purl, lrup? 12:51
purl lrup is yourname backwards
12:51 p6eval joined
moritz partcl: pugs "I don't know TCL" 12:53
p6eval partcl d55427: OUTPUTĀ«invalid command name "pugs"ā¤current instr.: '_block17' pc 47504 (src/Partcl/Grammar.pir:29)ā¤Ā»
moritz partcl: puts "I don't know TCL"
p6eval partcl d55427: OUTPUTĀ«I don't know TCLā¤Ā»
dalek rrot: r48621 | NotFound++ | trunk/t/pmc/exception.t:
more tests for Exception clone
moritz partcl: puts foo 12:54
p6eval partcl d55427: OUTPUTĀ«fooā¤Ā»
moritz partcl: puts (foo bar)
p6eval partcl d55427: OUTPUTĀ«can not find channel named "(foo"ā¤current instr.: '_block17' pc 47504 (src/Partcl/Grammar.pir:29)ā¤Ā»
moritz partcl: puts {foo bar}
p6eval partcl d55427: OUTPUTĀ«foo barā¤Ā»
Coke {} is like '' in perl, "" is comparable. 12:55
puts stderr whee
partcl: puts stderr whee
p6eval partcl d55427: OUTPUTĀ«wheeā¤Ā»
Coke partcl: puts stderr -nonewline whee
p6eval partcl d55427: OUTPUTĀ«-nonewlineā¤Ā»
Coke partcl: puts -nonewline stderr whee
p6eval partcl d55427: OUTPUTĀ«wheeĀ»
moritz partcl: set x foobarā¤puts x 12:56
p6eval partcl d55427: OUTPUTĀ«xā¤Ā»
moritz partcl: set x foobarā¤puts x;
p6eval partcl d55427: OUTPUTĀ«xā¤Ā»
Coke partcl: puts \\u30b3\\u30fc\\u30d2\\u30fc 12:57
p6eval partcl d55427: OUTPUTĀ«ć‚³ćƒ¼ćƒ’ćƒ¼ā¤Ā»
Coke partcl: namespace eval moritz { puts moritz++ }
p6eval partcl d55427: OUTPUTĀ«moritz++ā¤Ā»
12:57 TonyC_ is now known as TonyC 13:04 ruoso joined
Coke moritz: to expand a variable, you can either call [set x] or use the $ sigil. 13:05
mikehh t/codingstd/c_macro_args.t - Failed test 'Check for unwrapped macro arguments' - 2 unsafe macro args found: include/parrot/hash.h : _code 13:09
all other tests PASS (pre/post-config, make corevm/make coretest, test, fulltest) at r48620 - Ubuntu 10.04 amd64 (g++ with --optimize)
13:40 ruoso left
nwellnhof mikehh: my bad 13:43
dalek rrot: r48622 | nwellnhof++ | trunk (2 files):
Clear the live flag on copied strings

I think the handling of flags in str_copy and str_clone still needs some auditing. Also fix a glaring, but probably harmless error in PObj_gc_CLEAR.		 13:44
rrot: r48623 | nwellnhof++ | trunk (2 files):
Add test case for TT#1603
nwellnhof mikehh: no, not my bad 13:45
13:46 uniejo left
nwellnhof But these macro args are code blocks, so they can't be wrapped in parens. 13:47
codetest should probably make an exception for args named '_code'
dalek p-rx: f32c0d3 | moritz++ | t/p6regex/rx_modifiers:
get rid of some regex modifier tests that are (intentionally) no longer supported
 13:48
13:56 khairul joined 14:00 GreenM0nk joined
dalek p-rx: c7b2af5 | moritz++ | t/p6regex/ (2 files):
fudge backtracking tests, and run them again
 14:00
rrot: r48624 | nwellnhof++ | trunk/t/codingstd/c_macro_args.t:
Don't complain about unwrapped args that are code blocks
14:00 GreenM0nk left
14:10 SingAlong joined
SingAlong hi all 14:10
where do i find the docs for HLLCompiler? I'm reading the parrot book on wikibooks and following the exercises. (when "docs" are mentioned, i'm imagining docs like the one javadoc produces or does parrot have only books?) 14:11
Coke HLLCompiler is in compilers/pct, IIRC.
SingAlong Coke: wanted to look at the options it has (like setting the command prompt symbol). where should i be looking at in the hosted docs? 14:12
moritz perldoc compilers/pct/src/PCT/HLLCompiler.pir 14:13
SingAlong moritz: No hosted docs on the site?
moritz idontknow
Coke it's possible that those docs were missed during 'make html', sorry. :( 14:14
hurm. in old school pge, I had to add "is wb(0)" to operators like "ni" and "in". I wonder how to specify that in the new nqp-rx world. :wb(0) seems to have no effect.
SingAlong Coke: So usually hosted docs are available?
or was that a sarcastic line? :) 14:15
moritz Coke: what did :wb do?
SingAlong: docs.parrot.org/parrot/latest/html/ has docs 14:16
Coke moritz: it was "magic" to allow multicharacter non-symbolic ops. 14:17
SingAlong moritz: thanks! forgot to look in the developer docs section. coz the rest sounded like a tutorial-kid of docs to me (which is good actually)
Coke otherwise the op parser would not think I was serious when I said it was an op. =-)
SingAlong docs.parrot.org/parrot/latest/html/...loper.html
moritz "non-symbolic"? as in "alphanumeric"? 14:18
mikehh nwellnhof: yeah I tried to wrap the _code args but that did not work
nwellnhof mikehh: I simply special cased it in the codetest script
Coke moritz: yes. the operator is "ni", as opposed to "<=", which needed no special notation. 14:19
here's the old syntax from PGE:
201:proto 'infix:ni' is wb(0) is precedence('07=') {...}
moritz Coke: I don't think anything special should be needed 14:20
Coke: but i've added a few word boundaries myself to regexes
Coke: docs.parrot.org/parrot/latest/html/ 14:21
purl docs.parrot.org/parrot/latest/html/ is now up to date.
moritz sorry
github.com/rakudo/rakudo/blob/maste...r.pm#L1931
that's the URL I meant
Coke hurm. perhaps I'm not telling it how to find it's implementation properly.
moritz what error do you get? 14:22
Coke for, say, infix:but, do you just define a sub in the right namespace with that name for the impl?
moritz in rakudo, yes
Coke moritz: no error. "foo in bar" returns "f".
what about nqp-rx? ;)
14:23 Andy joined
Coke moritz: I have an ovveride for == called &infix:<==> which /is/ getting invoked. the override for 'ni' is not. 14:27
s/ovveride/nqp-rx sub/ 14:28
Any idea where the sub that implements the .O call here lives: 14:30
Partcl::Grammar.O(':prec<07>', '%list_containment');
kthakore moritz: hello 14:31
moritz Coke: src/cheats/hll-grammar.pir line 65 in nqp-rx
kthakore moritz: Do you know how to get LaTeX to find fonts properly? I have these fonts in /usr/share/fonts and ~/.fonts . But I still get paste.scsys.co.uk/49131 14:32
moritz kthakore: no idea; pronik in #perl6book set up the build system for the Perl 6 book 14:33
dalek rrot: r48625 | mikehh++ | trunk/t/op/gc-active-buffers.t:
add svn properties
 14:34
Coke moritz++ 14:35
ok, looks like it takes all the adverb like elements in the string and stores them in %!ohash, which isn't used anywhere else in nqp-rx. 14:37
(just trying to find a list of possible options to set there to see if there's anything that addresses this. ah well. will poke more later. 14:38
moritz++
moritz drowns in karma 14:39
Coke nwellnhof++ # thanks for fixing that bug. 14:40
nwellnhof Coke++ # for spotting the issue 14:41
whiteknight fperrad: ping 14:44
14:46 plobsing joined 14:53 tadzik joined
dalek TT #1755 created by whiteknight++: Include arbitrary .c and .h files in distutils dynpmc build 14:58
TT #1755: trac.parrot.org/parrot/ticket/1755
14:59 kj joined 15:03 zostay joined
dalek thub-trac: 9f8e0a7 | ++ | TODO:
add TODO for using git log instead of map-svn
 15:22
thub-trac: 7f35256 | ++ | github/github.py:
comment-only code to read git log output into db
dukeleto 'ello 15:25
15:26 kj left
dalek rrot-linear-algebra: f80856e | Whiteknight++ | src/ (6 files):
rearrange some things so that we build again. I've moved the DECLATTR and ALLOCATE_STORAGE macros into the header file, which turns out to be a wasted exercise since my library files can't get access to the Parrot_*_attributes structures, due to the order of the build
 15:29
rrot-linear-algebra: b39eb15 | Whiteknight++ | s (4 files):
Allow binding to the regular (FORTRAN) libblas-3.so. In this case, we don't have cblas.h, so we don't probe for that file in setup.nqp, and we have to define our own bindings in pla_blas.h. Notice that these bindings are different from the bindings provided by cblas/atlas
15:29 tcurtis joined
ttbot Parrot trunk/ r48626 i386-linux-thread-multi make error tt.taptinder.org/file/cmdout/374530.txt ( tt.taptinder.org//buildstatus/pr-Pa.../rp-trunk/ ) 15:32
whiteknight plobsing: ping 15:39
plobsing whiteknight: pong
whiteknight plobsing: I haven't really been keeping track. How did ash_'s GSoC project turn out? 15:40
dukeleto that ttbot error is about: src/packfile/byteorder.h:50: error: 'ASSERT_ARGS_fetch_iv_le' undeclared (first use in this function)
mikehh All tests PASS (pre/post-config, make corevm/make coretest, test, fulltest) at r48625 - Ubuntu 10.04 i386 (g++)
plobsing dukeleto: already fixed and commited
dukeleto plobsing++
dalek rrot: r48626 | plobsing++ | trunk (6 files):
inline byteorder functions in pf_items.c
 15:41
plobsing he met the conservative goal of getting a libffi frame builder functional (it's sitting in an svn branch ATM).
dalek rrot: r48627 | plobsing++ | trunk/src/packfile/pf_items.c:
remove now-redundant i386-specific inlining
rrot: r48628 | plobsing++ | trunk/src/packfile/byteorder.h:
fix unoptimized build
plobsing the more aggressive objective of some form of LLVM-based runtime system ran into issues. He identified a few ways of going about a simple runloop-type of native executing engine and I think he'll continue work along those lines. 15:43
kthakore How do I get LaTeX to find fonts properly? I have these fonts in /usr/share/fonts and ~/.fonts . But I still get paste.scsys.co.uk/49131 15:44
cotto_work ~~ 15:48
15:52 tadzik left
SingAlong There's a problem about variables in interactive mode mentioned in the wikibooks article. Has that been fixed? 15:56
moritz for what? parrot? rakudo? some other compiler?
SingAlong moritz: parrot 15:57
PCT
moritz uhm, parrot reads only PIR and PASM
those are line oriented
I don't think an interactive mode makes much sense
nqp-rx and rakudo both have a working, interactive mode, not sure how general it is though 15:58
dalek rrot: r48629 | mikehh++ | trunk/src/packfile/byteorder.h:
add svn properties
plobsing closest thing has to interactive mode is pdb. it will crash on you interactively :p
SingAlong moritz: this is what I was refering to en.wikibooks.org/wiki/Parrot_Virtua...ive_Squaak 15:59
tcurtis squaak?
purl somebody said squaak was examples/languages/squaak or the language in kj's tutorial at parrotblog.org
tcurtis SingAlong: that's out of date. docs.parrot.org/parrot/latest/html/...orial.html is the up-to-date version. 16:00
SingAlong tcurtis: thanks a lot! :)
tcurtis SingAlong: if there's anything that doesn't work or is hard to understand, let me know. 16:01
kthakore moritz: OMG! I figured it out
SingAlong tcurtis: thank you! the latest docs dont seem to have the warning mentioned in wikibooks. Sounds that it's been fixed already 16:02
moritz tcurtis: so, what's the fix?
sorry, meant kthakore
kthakore moritz: in ubuntu for a differening version of xelatex 16:03
you have to use the name provided by fc-list
moritz: please put that in README because I can finally make perl6book on my ubuntu
or should I?
lets see if debian still likes it 16:04
moritz kthakore: please do
SingAlong whats the channel #parrotsketch for?
moritz SingAlong: weekly design meetings 16:05
mikehh SingAlong: developer weekly meeting - today
kthakore moritz: testing it first
SingAlong I'm trying to register my irc nickname and nickserv doesnt exist 16:06
moritz irc.perl.org doesn't have such mechanisms
but it's much smaller than freenode, and doesn't usually need it
16:06 kthakore is now known as SangAlong
SangAlong SingAlong: I am you evil twin :D 16:07
SingAlong ah so i cant register my nick :P
16:07 SangAlong is now known as kthakore
kthakore SingAlong: nope 16:07
Coke moritz: (interactive mode) partcl is using it too.
SingAlong kthakore: I just read a superman comic last week in which Bizzaro is featured (he's superman's evil twin or look-alike or whatever. actually a phantom) 16:08
moritz Coke: does it preserve lexicals from the previous line?
Coke moritz: I don't use lexicals because <?> sucks. =-)
moritz wait... does TCL have lexicals?
Coke sort of.
but not really. 16:09
16:09 cogno joined
Coke partcl-nqp is using a lexical called %LEXPAD and using that to handle variable storage. 16:09
I /think/ partcl did it the "right" way from parrot's POV.
so, $foo is always the local foo; $::foo is always foo in the global namespace... but you can use uplevel/upvar to walk up the chain and do what you want, or alias them to vars in the current scope, etc. 16:11
no idea how much of that is still left in pnqp
(but to answer your question:)
(yes, the repl remembers my variables in teh REPL) 16:12
dalek rrot: r48630 | mikehh++ | trunk/src/packfile/byteorder.h:
add PARROT_BYTEORDER_H_GUARD
 16:15
16:15 khairul left
Coke moritz: wierd. apparently my 'ni' and 'in' subs /are/ getting called, at least in some cases. need to deparse the cases where not, presumably some other misparse was affecting it. 16:16
moritz++ #again1 16:19
16:19 jsut_ joined 16:24 jsut left 16:28 pyrimidine joined
dalek rrot: r48631 | plobsing++ | trunk (2 files):
remove less than sane "sanity" check from LexInfo
 16:31
16:35 cogno left
dukeleto looks like we have closed 14 tickets so far since last #ps, good job peeps 16:36
dalek rrot: r48632 | plobsing++ | trunk (3 files):
remove unneeded visit vtable from Hash
 16:48
16:54 kj joined 16:55 davidfetter joined 17:04 kj left
dalek rrot: r48633 | plobsing++ | trunk (3 files):
consolidate hash freeze/thaw in src/hash.c
 17:05
17:10 ruoso joined
Coke fsdo? 17:12
purl well, fsdo is for some definition of
17:13 jan left 17:15 Paul_the_Greek joined
Paul_the_Greek Good afternoon, Parroteers. 17:15
mikehh greetings and all
dalek rrot: r48634 | plobsing++ | trunk (4 files):
inline undocumented, used-once static functions
 17:22
rrot: r48635 | nwellnhof++ | trunk/src/string/api.c:
Clarify handling of flags in str_copy
17:31 mikehh left 17:37 cogno joined
plobsing why do we test that freeze/thaw and clone preserve internal order in hashes? 17:39
dalek rrot: r48636 | nwellnhof++ | trunk/src/string/api.c:
Undo my last commit
17:41 nwellnhof left 17:42 mikehh joined 17:46 davidfetter left 17:52 cogno left 17:53 rurban joined
dalek rrot: r48637 | plobsing++ | trunk (3 files):
pre-allocate buckets when thawing hashes
 17:56
ttbot Parrot trunk/ r48637 i386-linux-thread-multi make error tt.taptinder.org/file/cmdout/374917.txt ( tt.taptinder.org//buildstatus/pr-Pa.../rp-trunk/ ) 17:57
jnthn plobsing++ 18:00
18:12 theory joined 18:15 lucian left
Coke mikehh: holy crap. hope everything is ok with the DIL. 18:35
if anyone ever needs to drop the release for an issue like that, by all means, someone can pick it up. mikehh - thanks for putting in the extra effort there. 18:36
18:37 jan joined 18:40 ruoso left 18:42 nwellnhof joined
whiteknight what the DIL? 18:46
18:47 lucian joined
mikehh Coke: she has a little memory loss, but no apparent other damage, but it has been very traumatic 18:47
whiteknight: Daughter-In-Law 18:48
Coke points at mikehh's ps report. 18:49
mikehh plobsing: oing 18:54
sorry
plobsing: ping
18:55 TimToady joined
plobsing mikehh: pong 18:56
mikehh plobsing: I am getting build failures with g++, gcc ok, with src/hash.c and then src/pmc/hash.pmc 18:58
plobsing hmmm... maybe my changes aren't correct C++. I'll look into it. 18:59
mikehh All tests PASS (pre/post-config, make corevm/make coretest, test, fulltest) at r48637 - Ubuntu 10.04 i386 (gcc with --optimize) 19:00
19:13 lucian left
particle1 why on earth are our checksums tgz'd? ftp.parrot.org/releases/devel/2.7.0/ 19:17
19:22 senf_statt_oel joined
mikehh particle1: they aren't - sha256 checksums for tar.gz and tar.bz2 files 19:23
19:24 particle1 is now known as particle
particle oh, i see 19:24
i didn't even look inside. duh
19:26 Paul_the_Greek left
cotto_work #ps in 60 19:30
19:38 antivigilante joined
nwellnhof I just realized how awesome 'git rebase --interactive' is 19:40
git++
moritz it is indeed 19:42
and -i is so much shorter than --interactive :-)
dukeleto nwellnhof++ # really nice work closing some long standing gc tickets 19:49
nwellnhof: as for git rebase -i, you might have a laugh at twitter.com/rebastiality 19:50
nwellnhof dukeleto: I fixed two them by accident, actually
But I also wrote some test cases
dukeleto nwellnhof: that is pretty awesome. and always write test cases :) 19:51
nwellnhof: take a look at this ticket: trac.parrot.org/parrot/ticket/1746 19:52
nwellnhof: see if you can fix it by accident :) 19:53
nwellnhof: i think masak++ would buy you a delicious beverage if you did
nwellnhof hmm, in hash_mark 19:54
also a very deep stack
dukeleto nwellnhof: yeah. i think it is a bug tickled by recent hash optimizations
dalek rrot: r48638 | plobsing++ | trunk/src (2 files):
fix C++ build and unoptimized build
dukeleto nwellnhof: yes, i think it is an infinite loop that corrupts memory
nwellnhof it doesn't seem to be a stack overflow 19:56
so it's not a duplicate of #1723
moritz 8 19:58
sorry
nwellnhof definitely looks like some kind of meomry corruption
dukeleto 30 mins or so until #ps, right? 20:00
atrodo dukeleto> that
dukeleto> that's what my clock says
(stuipd ' and enter being too close together) 20:01
nwellnhof dukeleto: but i don't think it's an infinite loop. these kind of stack traces are possible in the GC marking phase. see also #1723 20:02
20:05 bluescreen left
dukeleto nwellnhof: you are right, it is memory corruption, but not necessarily an infinite loop 20:06
20:06 whiteknight left
nwellnhof dukeleto: do you know if masak tried an unoptimized build? 20:08
cotto_work atrodo: PROTIP: remove your ' or enter keys 20:10
atrodo cotto_work++
cotto_work dukeleto++ for the git work 20:11
20:11 masak joined
masak o/ 20:11
no, I didn't try an unoptimized build.
how do I do that?
cotto_work don't pass --optimize to parrot's Configure.pl 20:13
masak er. I never do that. 20:16
so if that's all it takes; yes, I have tried it on an unoptimized build.
every time.
purl well, every time is peak time somewhere
masak purl: forget every time
purl masak: I forgot every time
20:17 bluescreen joined 20:19 lucian joined
cotto_work If you're building Rakudo with --gen-parrot, it defaults to --optimize 20:23
masak I'm not building Rakudo with --gen-parrot. 20:28
I have an old-fashiened Rakudo-inside-Parrot setup. 20:29
s/ie/io/
cotto_work ok
#ps in 1
antivigilante ok I can't find the squaak files 20:33
anyone know where I can get them 20:34
purl m is the new i - for example, "mlife". (i was the new e; for example, ibook and ebay.) or a cartoon
cotto_work examples/languages/squaak/ in svn trunk
squaak?
purl i guess squaak is examples/languages/squaak or the language in kj's tutorial at parrotblog.org
NotFound Oooops
cotto_work seen smash 20:39
purl smash was last seen on #catalyst 5 hours, 27 minutes and 10 seconds ago, saying: thks everyone for input
cotto_work smash: ping
20:40 fperrad left
dalek website: tcurtis++ | PAST Optimization: GSoC is over 20:45
website: www.parrot.org/content/past-optimiz...-gsoc-over
smash cotto_work: pong 20:49
cotto_work smash: should there be something about PaFo voting on the parrot.org front page? 20:50
smash cotto_work: not usually, since it's something just for members, the traditional announcements for membser are on parrot-members 20:51
cotto_work ok
20:58 antivigilante left 21:01 whiteknight joined 21:20 nwellnhof left 21:21 nwellnhof joined
dukeleto how do people want to use github? 21:22
masak isn't sure he understands the question 21:26
tcurtis Pull requests are a nice feature, at least from the perspective of the person requesting. 21:27
masak Rakudo tends to favour RT patches, but has been known to notice pull requests too. 21:28
pyrimidine dukeleto: is parrot set up as a github organization? 21:30
dukeleto My question is: How does the Parrot community want to use Github? Do we want to use it the same way as Rakudo does? Do we want it to be the main mirror for our git repo? A backup repo? Do we want to use pull requests?
pyrimidine: yes
pyrimidine dukeleto: last I looked, pull requests for orgs are not working as expected
support.github.com/discussions/orga...anizations 21:31
but, that should be fixed at some point... 21:32
dukeleto pyrimidine: good to know, but the parrot community still needs to figure out how it wants to use github
I am pretty sure nobody wants to use github issues or the wiki, since we have Trac. But there are other questions
nwellnhof dukeleto: I think pull requests would be great for new developers or occasional contributors 21:33
pyrimidine pull requests are good (so I would vote for having it as the main repo), but that will be up to the Parrot devs. 21:35
21:35 bluescreen left
dukeleto one big question: github has a primary mirror or backup mirror 21:36
atrodo A naive question, does the CLA complicate pull requests? 21:37
sorear dukeleto: with git, we'll have around 10k backups anyway... 21:38
21:42 lucian left
dukeleto sorear: yes, good point. but we still need to decide as a community what will be the canonical place 21:43
atrodo: that is not so naive. only people with CLAs can commit, but someone can do a pull request and be applied by someone with a CLA, i think 21:44
cotto_work dukeleto: sorry. I have a meeting from 2:00 to 2:45-ish that usually interrupts #ps 21:46
When the github plugin is "ready", it'll create wiki links to the parrot repo on github from svn revision numbers and git hashes (similar to how r12345 on a trac wiki creates a link to Trac's source/changeset browser). 21:48
It
It'll also correctly deal with github's hooks so that any changes to the github repo get propagated and can be linked from trac. 21:49
sorear I think CLAs are going to be pretty meaningless in a DVCS world 21:50
since there's almost no operational difference between being given a commit bit, and being trusted/good enough to have a 100% patch acceptance rate
dukeleto sorear: i hear that, but I think CLAs mostly exist as a protection against patent trolls. 21:52
cotto_work I don't see that as being a difficulty. We can just say "This is a good fork. We'll pull once we get a CLA."
sorear cotto_work: right now, we don't require CLAs for accepting patches in any form 21:53
cotto_work That's true.
sorear a CLA lets you apply patches without a human rubber-stamping them 21:54
particle it's a prerequisite
the commit bit lets you apply patches.
dukeleto cotto_work: no worries. One point is this: right now, the parrot github mirror is just a mirror of svn. nothing should (currently) be on our github mirror that is not in svn
particle we should have CLAs from all contributors, ideally.
we must have them to accept large contributions
or, many contributions over time. 21:55
cotto_work Sure. Once the plugin is "ready" and enough Parrot people are on board, I'll draft a timeline for the details of the migration. 21:56
21:57 tadzik joined, dduncan joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, test, fulltest) at r48638 - Ubuntu 10.04 i386 (g++) 22:00
22:01 pyrimidine left
particle smolder.parrot.org/app 22:03
22:03 senf_statt_oel left, dduncan left
particle dukeleto: ^^ 22:04
22:07 tadzik left
dukeleto particle: /me looks 22:08
particle: sweet! 22:09
particle: is that brand new?
particle i'm trying to get a user created, so i can create a project etc
aye
just born
dukeleto particle: awesome!
particle: if you tell me the user/pass whenever you know it, I can change the code so that "make smoke" submits to that 22:10
particle: is it smolder running under apache? under it's own issue? 22:11
particle: own user, i meant. brainfart
particle smolder runs its own webserver 22:12
nwellnhof masak: ping
masak nwellnhof: pong
nwellnhof masak: i tried to run your file from #1746
masak: i get an error message "Unable to find module 'A' in the @*INC directories." 22:13
masak nwellnhof: you need A.pm in your current directory.
and B.pm
nwellnhof masak: then i get segfault in the code trying to find out the line number
masak they're right above the actual perl6 invocation.
nwellnhof where can i find A.pm and B.pm? 22:14
masak they're right above the actual perl6 invocation.
in my paste in the ticket.
nwellnhof can't find it 22:15
masak looks
of course, the fact that you get a segfault without A.pm and B.pm is grave enough to merit investigation. but it may be another issue altogether.
nwellnhof indeed 22:16
masak nwellnhof: ok, so we're both looking at gist.github.com/530197 , right?
do you see 'cat A.pm' and 'cat B.pm'?
nwellnhof ah, now i see it 22:17
masak oh, good.
22:17 mikehh left
masak you may use any means you like to actually populate those two files with that content locally. 22:18
Coke (github issues or the wiki) DEAR FSM, WE ARE NOT MOVING THAT PART OF OUR INFRASTRUCTURE AGAIN.
masak +1 22:19
purl 1
masak trac works nicely for that.
nwellnhof hmm, "perl6 -e 'use A; use B'" runs fine on my machine
masak nwellnhof: and you applied the patch in the ticket?
and ran 'make install' on Rakudo?
(just checking) 22:20
nwellnhof of course not. sorry.
masak let me know how I can make the steps-to-reproduce in that ticket more clear. :)
Coke ... I just realized I'm still at work. 22:21
nwellnhof masak: your steps are perfectly clear. i'm just stupid. 22:22
masak nwellnhof: I refuse to believe that. you're looking at my ticket, so obviously you're an intelligent fellow. :) 22:23
nwellnhof OK, applied the patch, building rakudo
22:23 kid51 joined
masak but maybe I assumed too much familiarity with the Rakudo build process while laying out the steps. 22:23
nwellnhof i was confused by the foo.pir you posted in the ticket 22:24
i thought it was a self-contained test case
but it is, actually, because it segfaults, too 22:25
*sigh*
jnthn You're in a twisty maze of segfaults all different.
nwellnhof OK, now i got the original segfault. 22:26
masak \\o/ 22:27
:(
nwellnhof But it's in a different place
masak nwellnhof: the foo.pir is sorta a self-contained segfault. but it still requires the patched Rakudo.
nwellnhof And A.pm and B.pm? 22:28
masak foo.pir is a distillation of A.pm and B.pm
er, no wait. foo.pir is 'use A; use B' compiled down to PIR code. so A.pm and B.pm are still needed. 22:29
22:30 Andy left
nwellnhof OK, I added a comment with my segfault. 22:33
masak thank you. 22:34
jnthn It's not an obvious segfault, sadly. :( 22:35
If it's the one that showed up from the enums patch, anyway.
masak my next step is to patch up older releases of Rakudo/Parrot with the same patch, and try to find the newest Parrot release that doesn't segfault.
plobsing msg Coke can you still reproduce TT #1459? I am unable to.
purl Message for coke stored.
jnthn masak: Did you manage to get it down to a minimal Rakudo addition that triggers it? 22:36
masak: That is, is the ticket already golfed from the Rakudo side?
masak jnthn: no, it's as-is.
tcurtis Wow. Niecza supports threads. 22:37
Oops. Wrong channel.
masak jnthn: if I had to guess, I'd guess the anon role or anon class did it. that's my spider sense saying that.
tcurtis: but oddly appropriate here too...
22:40 mikehh joined
nwellnhof masak: it seems to be a GC issue. using 'parrot --no-gc' works. 22:42
dalek rrot: r48639 | jkeenan++ | branches/tt677_toolsdirs:
Creating branch to work on ļæ½trac.parrot.org/parrot/ticket/677.
rrot: r48640 | jkeenan++ | branches/tt677_toolsdirs/tools/release:
Add new directory to hold programs, templates and configuration files useful during the release process.
masak nwellnhof: good to know.
masak nwellnhof: the stacktrace I got also seemed to indicate the GC being the culprit. 22:43
nwellnhof maybe something not marked correctly during GC
masak seriously considers running Parrot with --no-gc on by default from now on
would run faster and with fewer segfaults :)
nwellnhof until -ENOMEM 22:44
masak well, that tends to happen anyway...
I'm being unfair. but not very.
22:54 tetragon joined
dalek rrot: r48641 | jkeenan++ | branches/tt677_toolsdirs/tools/release/README:
Initial commit of README for tools/release/.
 22:59
rrot: r48642 | jkeenan++ | branches/tt677_toolsdirs/tools (0 files):
Per plan outlined in TT #677, move addoptags.pl and headerizer.pl from tools/build/ to tools/dev/. Move 7 files from tools/util/ to new directory tools/release/. Add two READMEs.
rrot: r48643 | jkeenan++ | branches/tt677_toolsdirs/tools/dev/README:
Add one more README.
rrot: r48644 | jkeenan++ | branches/tt677_toolsdirs/MANIFEST:
Update MANIFEST.
rrot: r48645 | jkeenan++ | branches/tt677_toolsdirs (1 files):
Move remaining files from tools/util/ to tools/release/ and update MANIFEST.
whiteknight masak: If rakudo would play nice, and follow supported releases instead of attempting to track svn HEAD, there would be fewer segfaults 23:01
jnthn [Citation needed]
whiteknight trunk breaks more frequently than it should, but releases do tend to be much better tested 23:02
cotto_work It's good to have a user who shouts at us more often than every three months.
23:02 davidfetter joined
masak whiteknight: if what you say is correct, then I won't be able to reproduce the segfault with any of the Rakudo releases, which always target Parrot releases. 23:02
whiteknight cotto_work: that may be, but you can't complain about instability if you are tracking HEAD
masak whiteknight: though I fear you're being... a bit optimistic. 23:03
whiteknight masak: I said "fewer" and "tends to"
rblackwe Robert from PPW here asking if there will be a Parrot/Rakudo hackathon and such at PPW 2010 and if so what are your needs?
cotto_work ppw?
purl it has been said that ppw is the Pittsburgh Perl Workshop pghpw.org or perl paddy wagon or gonna have mothafuckin sartak
whiteknight and I also suggest that more effort would go into releases if we thought anybody were making real use of them
rblackwe cotto_work: pghpw.org 23:04
masak whiteknight: one of the reasons I have have the setup I do (always blead Parrot/Rakudo) is to be able to pick up on these errors early on. I'm a whatsitbird in the coalmine.
whiteknight: Rakudo releases do target Parrot releases.
whiteknight: in that sense, many many people are making 'real use' of them.
jnthn whiteknight: I suspect all that would result is it'd take longer between a segfault causing bug being introduced and the Parrot team finding out about it. 23:05
whiteknight: The other thing is that nqp-rx updates at the moment force us to track Parrot head.
So it's not always that we want to track the latest Parrot changes so much as nqp-rx ones 23:06
Maybe that link needs breaking at some point.
whiteknight in any case, regardless of the relationship if you track HEAD, you get the instability of HEAD 23:08
dukeleto I also try PL/Parrot on Parrot trunk and Rakudo master, to make sure I am not going to get hit by a huge upgrade tax. masak++ has a good point 23:13
davidfetter is there a parrot master? 23:16
davidfetter has been using dukeleto++'s
dukeleto davidfetter: yeah, there is a github mirror, where the master branch tracks parrot's svn trunk
davidfetter an official parrot master*
masak thinks allison++ is the Parrot master :) 23:17
davidfetter heh
good point :)
dukeleto davidfetter: there is github.com/parrot/parrot, which is more officialer than github.com/leto/parrot, but the word "official" is mostly a joke in the perl community
davidfetter sufficiently aware of the cultural norms not to get bollixed by this 23:18
actually, i'd like to see about buildfarming parrot, pl/parrot, etc.
23:21 smash left 23:29 jan left
dalek rrot: r48646 | jkeenan++ | branches/tt677_toolsdirs (12 files):
Correct re-placement of dump_pbc.pl; update MANIFEST. Inside files, update mentions of paths under tools/.
 23:33
23:34 jan joined
whiteknight buildfarming would be nice 23:36
I wonder if we could chat with e.g. the GCC people about that 23:37
davidfetter pgbuildfarm.org/ it's not really a compile farm for distribution 23:43
cotto_work Is there a nice way to get git show to show me the latest n commits? 23:51
dukeleto davidfetter: i would very much like to see PL/Parrot get tested against a few versions of PG and Parrot/Rakudo 23:53
davidfetter yeah
dukeleto cotto_work: git log -n ?
23:53 Paul_the_Greek joined
dukeleto cotto_work: git log -p -n, i mean 23:53
davidfetter unsure how to generalize the way the buildfarm works
cotto_work dukeleto++ 23:54
just what I was looking for
will git log give me a list of every commit to parrot, including e.g. branches that never got merged 23:55
dukeleto cotto_work: you can give git log a branch name, or more generally, a "commit-ish" 23:56
cotto_work: by default, git log and most git commands work on "master"
cotto_work: so if a branch was never merged to master, it will not show up in "git log -p -3" for example 23:57
cotto_work: what are you wanting this command for?
cotto_work how much finagling would it take to get a list of all commits?
getting svn revision numbers and commit messages for the plugin
dukeleto cotto_work: i think i know, one sec 23:58

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!