Parrot 2.8.0 will be released at "2010-09-21 08:00 UTC" | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 25 tickets; remove deprecated items (especially CodeString); record deprecations; polish for 2.8 release
Set by moderator on 18 September 2010.
00:09 theory left 01:21 Themeruta joined, NotFound left
bacek_at_work aloha, (2*60 + 56) / (1*60 + 32) 01:38
aloha bacek_at_work: 1.91304347826087
bacek_at_work hmm...
01:55 mikehh left 02:04 PhatEddy left 02:07 Chandon left 02:22 kid51 left 02:28 mikehh joined 02:35 janus left
dalek rrot: r49162 | plobsing++ | branches/typesafe_consttable (4 files):
fix find_constants.
 02:41
02:42 janus joined 03:38 necrolyte joined 04:02 davidfetter joined 04:27 necrolyte left
dalek rrot: r49163 | cotto++ | trunk/PBC_COMPAT:
bump PBC_COMPAT for ImageIOFreeze and ImageIOThaw
 05:47
06:10 Themeruta is now known as NotFound 06:33 khisanth_ joined 06:36 Khisanth left 06:40 khisanth_ left 06:44 jsut_ joined 06:49 jsut left 06:53 Khisanth joined 07:09 fperrad joined 07:24 tadzik joined 07:28 davidfetter left 07:57 cotto left 07:59 barney joined
dalek TT #445 closed by mikehh++: [bug] Darwin string buffering segfault in examples/shootout/revcomp.pir 08:12
TT #445: trac.parrot.org/parrot/ticket/445
08:16 cotto joined 08:23 cotto left 08:37 barney left 08:41 cotto joined 08:45 cottoo joined 08:49 cotto left 08:57 cottoo is now known as cotto
cotto tomato++ 08:58
09:02 particle1 joined 09:05 particle left 09:09 wagle_ left
mikehh still getting g++ build failures in trunk - src/pmc/imageiothaw.pmc errors 09:23
bacek aloha, humans 09:31
cotto aloha, bacek 09:32
moritz o/
cotto mikehh, trying to repro
bacek cotto, hi
09:34 wagle joined
cotto repro works 09:38
well, fails to work
mikehh I tried adding casts etc to src/pmc/imageiothaw.pmc but just got into bigger problems 09:47
I am not sure I am happy with the code that is failing g++ there seems to be a underlying problem there 09:48
we have quite a few situations where there are inconsistancies with using const 09:50
09:52 ruoso left
cotto got confused too 09:55
better to let plobsing sort it out
or someone who's not borderline sick and sleep-deprived 09:56
moritz recommends sleep 09:57
mikehh I messaged plobsing a day or so ago about it 09:58
cotto follows moritz's recommendation 09:59
09:59 verb joined
dalek rrot: r49164 | bacek++ | branches/gc_massacre/src/gc/gc_ms2.c:
Compact String pool in GC MS2
 10:00
rrot: r49165 | bacek++ | branches/gc_massacre/src/gc/gc_ms2.c:
Count used memory during allocattion of String storage.
rrot: r49166 | bacek++ | branches/gc_massacre/src/gc/gc_ms2.c:
Ugly hack to prevent to often collecting due String GC
10:03 luben_work joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#125), fulltest) at r49163 - Ubuntu 10.04 i386 (gcc with --optimize) 10:12
luben_work waw, gc_massacre is fast... bacek++ 10:18
bacek luben_work, on which test? 10:19
luben_work here: build of rakudo is 2.50s vs 3.15 is trunk
rakudo startup is 510 ms vs 615 in trunk 10:20
bacek hooray! It was what I looking for :)
What about memory consumption?
luben_work it's twice the trunk as I see 10:21
bacek hmm...
It's not so good. 10:22
moritz that means it's not usable for rakudo developers :( 10:24
whightknight++ # wknight8111.blogspot.com/2010/09/pa...tform.html 10:27
10:30 aloha left 10:32 bacek left 10:44 mikehh left 10:53 bacek joined 10:57 aloha joined
dalek rrot: r49167 | bacek++ | branches/gc_massacre/t/op (2 files):
Invert test for GC.
 11:07
rrot: r49168 | bacek++ | branches/gc_massacre (3 files):
Intriduce GC_string_cb_FLAG to distinguish situation when String GC
rrot: r49169 | bacek++ | branches/gc_massacre/src/gc/gc_ms2.c:
Cleanup number of impatient objects after full GC run
rrot: r49170 | bacek++ | branches/gc_massacre/src/gc/string_gc.c:
Use passed GC_Statisticts in compact_pool
rrot: r49171 | bacek++ | branches/gc_massacre/src/gc/gc_ms2.c:
Call to compact_pool after unblocking GC.
rrot: r49172 | bacek++ | branches/gc_massacre/src/gc/gc_ms2.c:
GC_Subsystem has GC_Statistics now. Drop GC MS2 copy.
rrot: r49173 | bacek++ | branches/gc_massacre/src/gc/api.c:
Fix copy-paste typo.
rrot: r49174 | bacek++ | branches/gc_massacre/src/gc/string_gc.c:
Use block/unblock GC sweep instead of mark in compact_pool which is semantically more correct.
rrot: r49175 | bacek++ | branches/gc_massacre/t/op (2 files):
Use skip_all in disabled tests
bacek luben_work, ping. Can you retest rakudo building on gc_massacre again please? 11:28
luben_work yes 11:29
as point to compare. on parrot trunk, rakudo builds in 3 min 10sec, max memory 590M 11:31
moritz an 32 or 64bit? 11:32
bacek branch will consume about 700-750M for core.pm
32 bits
moritz s/an/on/
so it's twice as much for 64bit
bacek it's about 650 on 64
hmmm. May be I wrong about platform then 11:33
luben_work for gc_massacre - builds in 2m 45s, max memory 915M 11:36
both on linux/amd64
moritz 915 / 590
purl 1.55084745762712
aloha 1.55084745762712
bacek strange. It should be 590 + 256 max... 11:37
590 + 256
purl 846
aloha 846
11:47 JimmyZ joined 11:49 mikehh joined 11:59 JimmyZ left 12:09 whiteknight joined
whiteknight good morning, #parrot 12:20
12:37 bluescreen joined 12:40 ruoso joined 12:46 Coke joined
Coke rt.perl.org/ -> "Parrot Issues". That link should probably just be deleted. (updated to point to trac for bonus points.) 12:46
13:18 Patterner left 13:19 patspam joined 13:28 Psyche^ joined 13:29 Psyche^ is now known as Patterner 13:39 tadzik left 13:40 particle1 is now known as particle 13:42 tadzik joined 13:52 PacoLinux joined 14:03 davidfetter joined 14:23 bluescreen left
dalek ast: 665a134 | KodiB++ | S03-metaops/ (2 files):
Additions and unfudging for arguments ending with * to hypers and zips.
 14:26
kudo: be80e96 | KodiB++ | src/core/ (2 files):
Made zips extend arguments ending with *.
 14:30
whiteknight I had to unwatch diaspora on github. Updates to that were flooding out any parrot-related ones that I'm more interested in 14:35
14:40 bluescreen joined 14:53 davidfetter left 15:04 davidfetter joined 15:37 ruoso left
dalek rrot: r49176 | luben++ | trunk/include/parrot/hash.h:
Fix for TT#1794
 15:39
15:40 Andy joined 15:42 gottreu joined 15:52 bluescreen left 16:02 M_o_C joined 16:03 ruoso joined 16:07 bluescreen joined
dalek rrot: r49177 | luben++ | trunk/include/parrot/hash.h:
Amend fix for TT#1794 to include also cstrings and ptrs
 16:13
16:40 luben_work left
dalek rrot: r49178 | luben++ | trunk/src/pmc/hashiterator.pmc:
Fix HashIterator for NULL keys
 16:45
16:46 M_o_C left 16:52 M_o_C joined 16:54 M_o_C left 16:55 M_o_C joined, ashleyde1 joined 16:58 M_o_C left 17:04 theory joined
whiteknight a compiler on Parrot for Vala (or Genie, or both) would be very interesting. Most interesting would be generating valid C code output from it 17:11
I don't think I've yet seen a Parrot compiler project output anything besides PIR/PASM
atrodo The vala website doesn't do very well explaining it 17:15
NotFound whiteknight: a have the nebulous idea of making winxed able to generate C or C++ some day. 17:18
whiteknight I know it's supposed to be possible if we add a custom code generator step to NQP and HLLCompiler, I just have never seen it done 17:20
Eventually, I would like to see HLLCompiler have a built-in option to output C code that links to libparrot. that would be fantastic 17:21
NotFound whiteknight: maybe easier other idea that I ruminate from time to time: pbc_to_c
whiteknight we have a pbc_to_c tool 17:22
it probably doesn't do what you are thinking about
NotFound What does that thing? 17:23
whiteknight that creates the fakecutables.
NotFound That isn't pbc_to_exe?
whiteknight oh, that's the name. Yes. pbc_to_exe
but, it does create C code first
(just not good C code)
NotFound I mean real C, not a bunch of opcodes pasted together with a launcher, 17:24
whiteknight right
NotFound So many ideas, so little time... 17:25
whiteknight seriously
purl is that for real?
whiteknight maybe we need to start writing down a list of projects we want to work on, so we don't forget 17:26
NotFound Talking about ideas, a guy has suggested me to make winxed be a subset of groovy or phantom, in order to be able to compile winxed programs for jvm or cli. 17:27
Not a bad idea, just is not winxed :D
Tene whiteknight: that sounds like a good project for someone to work on
NotFound There is some parrot groovy project? 17:28
whiteknight I haven't heard of one. 17:29
check the list in the Languages wiki page
NotFound s/phantom/fantom
Interesting times for compiler writers, looks like a new language comes to life every day. 17:30
whiteknight to save effort, they should all use Parrot! 17:31
Coke to SAVE effort? 17:35
I don't think that word means what you think it means </princess bride>
NotFound Looks like Vala uses "good old" reference counting.
dukeleto inconceivable! 17:42
purl You keep using that word. I do not think it means what you think it means.
whiteknight GLib and GObject use refcounting
so I'm sure Vala does
I wonder how hard it would be to create a GObject wrapper PMC, and create a new metamodel that way 17:43
NotFound I'm approaching winxed revision 666... 17:44
dalek nxed: r660 | NotFound++ | trunk/ (3 files):
operators === and !=== (issame and isntsame)
 17:45
cotto has flashbacks to php 17:49
whiteknight, go start a ProjectIdeas page on the wiki 17:51
whiteknight cotto: Will in a bit. Have a meeting now 17:57
cotto excuses, excuses 18:01
dalek rrot: r49179 | jkeenan++ | trunk/include/parrot/hash.h:
[codingstd] wrap c macro args (2 locations).
 18:02
rrot: r49180 | jkeenan++ | trunk/include/parrot/hash.h:
Revert incorrect correction of codingstd.
18:03 verb left
cotto NotFound, do you have some tuits to fix the c++ build? It's broken atm. 18:11
It'd be nice not to cut the release with that broken.
NotFound cotto: unfortunately the mutable/immutable const/non const mix is making things more difficult, and I think that adding more const casts is dangerous for highly optimized builds. 18:16
18:26 fperrad left
cotto That's a shame. 18:28
NotFound cotto: it is, but I prefer to lose the C++ buildability rather than risk to return to the one heisenbug a day times. 18:35
whiteknight we already had a page on the wiki for projects 18:42
BigProjectIdeas
I added the vala compiler idea to that
dalek tracwiki: v11 | whiteknight++ | BigProjectIdeas 18:46
tracwiki: +update description to be more clear about what we want
tracwiki: trac.parrot.org/parrot/wiki/BigProj...ction=diff
tracwiki: v12 | whiteknight++ | BigProjectIdeas
tracwiki: +add Allison's Vala compiler idea
tracwiki: trac.parrot.org/parrot/wiki/BigProj...ction=diff
tracwiki: v13 | whiteknight++ | BigProjectIdeas
tracwiki: trac.parrot.org/parrot/wiki/BigProj...ction=diff
tracwiki: v14 | whiteknight++ | BigProjectIdeas
tracwiki: +idea about creating a new object metamodel using GObject
tracwiki: trac.parrot.org/parrot/wiki/BigProj...ction=diff
NotFound whiteknight: maybe a section "Languages wanted" will be shorter than several itmes "Create a compiler for the XXX language" 18:48
18:49 kid51 joined
nopaste "kid51" at 192.168.1.3 pasted "[codingstd] include/parrot/hash.h fails c_macro_args test" (23 lines) at nopaste.snit.ch/23490 18:50
NotFound kid51: impossible to fix the code, someone should fix the test 18:51
18:52 ruoso left
whiteknight NotFound: "Languages Wanted: All" 18:53
davidfetter SQL!
whiteknight you know what, I've thought about an SQL-like small language that could be used to write queries against objects and arrays 18:56
davidfetter might as well use actual SQL 18:57
NotFound BTW the test result is wrong. _hash is wrapped, the arg used unwrapped is _code
whiteknight I certainly wouldn't want to implement all of SQL. Anything I did would have to be a very small subset
davidfetter the thing about SQL is that once you're using it, you want more 18:58
it's like llello :P
18:58 Khisanth left
davidfetter ...only it keeps doing cool stuff when you're not using it 18:58
whiteknight I'm inspired by LINQ to SQL in C# world
davidfetter is at best underwhelmed by linq
whiteknight or, just LINQ
well yes, LINQ itself is underwhelming. What I like is the idea that I can run an SQL-like query against any arbitary dataset 18:59
cotto llello?
whiteknight once you have that, the implementation details can be tuned
cotto oic
19:00 Khisanth joined
davidfetter thinks the cpan project to make an SQL interface to all tabular system data is great 19:00
...and could probably be done much better with parrot
NotFound All your tabular data are belong to us! 19:01
davidfetter we get signal
purl WHAT! MAIN SCREEN TURN ON.
whiteknight Oh, It's you 19:03
purl How are you gentlemen?
dalek tracwiki: v15 | whiteknight++ | BigProjectIdeas
tracwiki: Break this page up into core development projects and compiler development projects. Add different size charts for each
tracwiki: trac.parrot.org/parrot/wiki/BigProj...ction=diff
tracwiki: v16 | whiteknight++ | BigProjectIdeas
whiteknight you have no chance to survive please make your time
dalek tracwiki: formatting. Now with less suck
tracwiki: trac.parrot.org/parrot/wiki/BigProj...ction=diff
whiteknight ha ha ha 19:04
atrodo Someone set us up the bomb
purl All your base are belong to us
whiteknight to every zig
davidfetter take off every zig
whiteknight needs to find a script
atrodo Really, purl needs updated to include all responses
whiteknight> en.wikiquote.org/wiki/Zero_Wing#Introduction 19:05
cotto tomato++
whiteknight you are on the way to destruction
purl what you say!!
whiteknight take off every 'ZIG'
Move ZIG 19:06
purl Take off every 'zig' !! You know what you are doing. Move 'zig'. For great justice.
atrodo huh, that's the wrong order...
whiteknight hahaha
this brings back so many good memories of the early internetz
atrodo The song is playing in my head now
19:07 ruoso joined 19:19 verb joined 19:29 kid51 is now known as kid51_afk 19:45 ruoso left 19:59 ruoso joined
cotto time for a ram upgrade 20:02
20:02 cotto left 20:04 whiteknight left
GeJ Bonjour everyone. 20:08
20:11 verb left 20:15 cotto joined
dalek rrot: r49181 | chromatic++ | branches/gc_massacre/src/gc/fixed_allocator.c:
[GC] Optimized fixed-size allocation function.

in this hot path; not bad for two minutes of work.		 20:16
rrot: r49182 | chromatic++ | branches/gc_massacre/src/gc/fixed_allocator.c:
[GC] Improved consting in arena_size().

can't handle it, in which case quiet sobbing is appropriate.)
rrot: r49183 | chromatic++ | branches/gc_massacre/src/gc/fixed_allocator.c:
[GC] Optimized pool_allocate()'s branch conditions.

clarifies the function *and* gives another half-percent performance improvement thanks to this being a hot path.
rrot: r49184 | chromatic++ | branches/gc_massacre/include/parrot/list.h:
[include] Optimized LIST_APPEND macro slightly.
dukeleto waves hello 20:54
mikehh howdy dukeleto 20:59
21:04 PacoLinux left, PacoLinux_ joined, PacoLinux_ is now known as PacoLinux
dalek rrot: r49185 | mikehh++ | trunk/t/codingstd/c_macro_args.t:
modify t/codingstd/c_macro_args.t so that it ignores unwrapped _hash
 21:07
21:09 whiteknight joined
cotto 18*19 21:13
purl 342
aloha 342
sorear whiteknight++ # honesty and a positive attitude will get you far 21:14
whiteknight thanks 21:15
we need to be honest. The more honest we are about identifying problem spots, the more we can focus on fixing them
21:20 kid51_afk left 21:24 tadzik left
dalek rrot: r49186 | luben++ | trunk/include/parrot/hash.h:
the fix of code std was right
 21:25
luben good evening everybody
21:29 muixirt joined
dukeleto luben: hola 21:34
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#132), fulltest) at r49185 - Ubuntu 10.10 beta (amd64 gcc-4.5 with --optimize)
luben I remember that there was a time when some subsystems in parrot were using reference counting. In current parrot, is there something like that left? 21:38
21:40 ruoso left
dalek rrot: r49187 | plobsing++ | trunk/src/pmc/imageiothaw.pmc:
add useless casts. C++ is like that.
 21:41
mikehh plobsing: not so much 'useless' as indicative of an underlying problem 21:44
anyway that seems to solve the build problem 21:47
dalek nxed: r661 | NotFound++ | trunk/winxedst1.winxed:
allow use of 'self' in closures
 21:56
rrot: r49188 | mikehh++ | trunk/t/codingstd/c_macro_args.t:
revert r49185 as _hash has now been wrapped (_code is still unwrapped)
 21:59
cotto has more free memory now than he had memory before he upgraded 22:06
22:08 bluescreen left
mikehh cotto: that's what an upgrade is all about :-} 22:13
cotto yes it is
cotto runs rakudo spectest to celebrate 22:16
plobsing upgrades are like GC. usually, you get extra memory afterwards. sometimes it just explodes for no reason. 22:17
dukeleto aloha, msg chromatic i think i am on the brink of finding stress_strings.pir slowdown, more details soon 22:18
aloha dukeleto: OK. I'll deliver the message.
plobsing dukeleto++ # regression detective 22:19
cotto plobsing++
This was possibly the least painful upgrade ever, but I've had the other kind too.
22:31 kid51 joined
dukeleto I HAVE SEEN THE LIGHT. 22:34
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#133), fulltest) at r49188 - Ubuntu 10.10 beta amd64 (g++-4.5) 22:36
dukeleto: hope it didn't damage your eyes any 22:37
cotto dukeleto, care to share with the rest of us? 22:40
Is git secretly terrible? Should we all drop Parrot in favor of .net? Is up down? 22:41
dukeleto strange is familiar 22:43
dukeleto finally has a script to benchmark a file across a set of git commits 22:46
22:48 whiteknight left, whiteknight joined, Andy left
dalek rrot: r49189 | NotFound++ | trunk/t/pmc/stringbuilder.t:
add a StringBuilder test for code coverage completeness
 22:49
sorear I don't think dropping Parrot in favor of .net even makes sense 22:53
dukeleto github.com/leto/Util/blob/master/bin/git-exec <-- dangerous and useful
kid51 msg mikehh Thanks for handling the problems with include/parrot/hash.h and t/codingstd/c_macro_args.t 22:56
purl Message for mikehh stored.
aloha OK. I'll deliver the message.
cotto sorear, I didn't come here to make sense.
kid51 reads chromatic's and whiteknight's blog posts 23:09
23:11 ruoso joined
whiteknight kid51: I hope you're as entertained reading it as I was writing it 23:21
I definitely do not have chromatic's gift of making sharp points in few words 23:24
23:24 elmex_ joined, sri left, _sri joined, x3nU_ left, moritz left, x3nU joined, moritz joined, luben left, luben joined, elmex left, ilbot2 joined
moderator Parrot 2.8.0 will be released at "2010-09-21 08:00 UTC" | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 25 tickets; remove deprecated items (especially CodeString); record deprecations; polish for 2.8 release
bacek_at_work ~~ 23:27
whiteknight hello bacek
bacek_at_work whiteknight, hi 23:28
23:28 x3nU_ joined, _sri left, sri joined
bacek_at_work whiteknight, guess what? opsc implemented in nqp and emits "pure C" :) 23:28
whiteknight bacek_at_work: really? awesome. What does the C look like?
bacek_at_work src/ops/core_ops.c :) 23:29
23:29 x3nU left
whiteknight oh, opsc. Right. 23:30
whiteknight was thinking of something else
I'm going to try to get two win32 smoke reports submitted tonight 23:32
..nevermind, my strawberry perl install does not appear to be sane 23:34
23:34 betterworld left 23:36 betterworld joined
kid51 Does anyone know what formatting is permitted/possible at www.parrot.org/blog? 23:37
whiteknight looks like plain text or html 23:38
23:43 patspam left, davidfetter left 23:45 davidfetter joined
NotFound was about to say that now with closures winxed is starting to be mature, but the discussion about maturity changed his mind X-) 23:48
plobsing NotFound: regarding pbc_to_c, ash_ was working on that in the latter half of gsoc 23:52
NotFound plobsing: I'm too sleepy to talk about that, now, sorry :O 23:53
'night 23:54
23:55 ash_ joined
dalek rrot: r49190 | luben++ | trunk/src/pmc/hash.pmc:
port hash.get_repr to StringBuilder
 23:57
rrot: r49191 | luben++ | trunk/src/pmc/codestring.pmc:
Revert "use concat instead of join when possible in CodeString"

Powered by Cro and the Raku® Programming Language.

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

Thank you!