Parrot 2.3.0 Released | parrot.org/ | Channel log: irclog.perlgeek.de/parrot/today | Priority: apply deprecations, merge branches, testexceptions_refactor branch | GSoC students, please read trac.parrot.org/parrot/wiki/GSoCersStartHere
Set by moderator on 6 May 2010.
Coke I guess one of leo, dan, or melvin. 00:38
just the way it is.
er, "historical reasons". 00:41
sorear are any of leo, dan, or melvin still with us? 00:47
plobsing seen melvin 00:48
purl melvin was last seen on #parrot 5 years, 42 days, 8 hours, 15 minutes and 16 seconds ago, saying: Coke: don't be ashamed of PIR, there are many production compilers using IR [Mar 29 16:33:39 2005]
plobsing seen dan
purl dan was last seen on #catalyst 4 days, 14 hours, 21 minutes and 47 seconds ago, saying: if it's your own box, stick with fastcgi [May 5 10:27:12 2010]
plobsing seen leo 00:49
purl leo was last seen on #perl 121 days, 16 hours, 56 minutes and 38 seconds ago, saying: and dont say read the topic [Jan 8 07:52:26 2010]
plobsing I'm going to go with no on that one.
00:54 abqar joined
dalek rrot: r46448 | coke++ | trunk/docs/book/pct/ch03_compiler_tools.pod:
Fix path to file, Courtesy John W. Emerson
 01:00
01:05 bacek_at_work joined
dalek TT #1073 closed by plobsing++: Lexical redeclarations in subs should throw exceptions 01:15
TT #1073: trac.parrot.org/parrot/ticket/1073
plobsing Coke: got any more IMCC bugs for me? I've exhausted the TODOs in t/compilers/imcc
01:16 Mokurai joined
dalek rrot: r46449 | plobsing++ | trunk (2 files):
croak on multiple declarations of the same lexical
 01:16
rrot: r46450 | plobsing++ | trunk/t/op/lexicals.t:
remove test that asserted the opposite behaviour to that specified in TT #1073
rrot: r46451 | petdance++ | trunk (2 files):
updating annotations
01:28 theory joined
dalek rrot: r46452 | mikehh++ | trunk/compilers/imcc/pbc.c:
fix codetest failure - trailing spaces
 01:33
sorear According to svn, awk, find, comm, xargs, and ack (there has got to be a simpler way), no less than 4 non-generated files contain lists of dynpmcs that I would presumably have to update 01:37
5 if you count DEPRECATED.pod
(MANIFEST, MANIFEST.generated, src/dynpmc/Defines.in, src/dynpmc/Rules.in) 01:38
and that's not counting the testsuite
01:40 Andy joined
plobsing adding functionality to core is not a thing to be taken lightly 01:40
the work is appropriate IMHO
01:40 snarkyboojum joined
sorear The idea here isn't to add functionality per se but to rearrange it 01:41
"lineof" being a method on CodeString has unfortunate interactions with branches/codestring 01:42
plobsing btw, if you look at src/dynpmc{Defines,Rules}.in I think you'll find they get generated from the contents of the src/dynpmc directory 01:43
sorear they're under version control 01:44
plobsing whoops, I was thinking of the .mak files they generate 01:45
01:56 kurahaupo joined 02:00 Psyche^ joined
mikehh if I add the ASSERT_ARGS() to src/embed.c I get test failures in make test - t/tools/dump_pbc.t - Failed test: 1 and t/tools/pbc_disassemble.t - Failed tests: 1-4 - All with failed assertion 'outfile' 02:13
plobsing mikehh: which function? 02:14
ttbot Parrot trunk/ r46454 i386-linux-thread-multi make error tt.taptinder.org/file/cmdout/302587.txt ( tt.taptinder.org//buildstatus/pr-Pa.../rp-trunk/ ) 02:16
plobsing mikehh: Parrot_disassemble appears to do explicit null handling of outfile. The argument should be guarded with 'ARGIN_NULLOK' 02:17
sorear hey, I just figured out the source of a mysterious recent Parrot speed regression
... I installed a non --optimize build 02:18
mikehh plobsing - looks like pbc_disassemble
02:21 JimmyZ joined
dalek rrot: r46453 | petdance++ | trunk/include/parrot/hash.h:
removed unused typedef
 02:24
rrot: r46454 | petdance++ | trunk (3 files):
updating function pointer annotations
sorear msg Coke, Hi. Rakudo Grammar.pm just built in 151MB and 92 seconds on the codestring branch (changes were required to NQP-rx; see branch topic/no-codestring-target in the nqprx repo) 02:31
purl Sorry, I've never seen coke, before.
sorear msg Coke Hi. Rakudo Grammar.pm just built in 151MB and 92 seconds on the codestring branch (changes were required to NQP-rx; see branch topic/no-codestring-target in the nqprx repo)
purl Message for coke stored.
mikehh plobsing - ok that seems to work - had to run make headerizer again 02:34
sorear msg Coke rakudobuild completed in 10m14s, the fastest I've ever seen (second best around 11m30s). Peak memory usage 323MB. 5m15s spent compiling the setting. Let's ship. 02:39
purl Message for coke stored.
ttbot Parrot trunk/ r46455 i386-linux-thread-multi make error tt.taptinder.org/file/cmdout/302666.txt ( tt.taptinder.org//buildstatus/pr-Pa.../rp-trunk/ ) 02:40
dalek rrot: r46455 | mikehh++ | trunk (4 files):
add ASSERT_ARGS, change Parrot_disassemble ARGIN to ARGIN_NULLOK, re-run make headerizer
 02:41
mikehh damnit somethging broke there 02:45
02:49 snarkyboojum joined 02:54 janus joined
ttbot Parrot trunk/ r46456 i386-linux-thread-multi make error tt.taptinder.org/file/cmdout/302708.txt ( tt.taptinder.org//buildstatus/pr-Pa.../rp-trunk/ ) 02:55
dalek rrot: r46456 | plobsing++ | trunk/compilers/imcc (8 files):
all static variables in IMCC are now either marked "XXX non-reentrant" or are const
 02:57
plobsing mikehh: I'll have a look at it.
03:04 davidfetter joined 03:08 theory joined
plobsing mikehh: fixed at r46457. 03:10
dalek rrot: r46457 | plobsing++ | trunk (2 files):
this function VERY CLEARLY handles NULL arguments
 03:13
03:19 bacek_at_work joined
mikehh phew - that's better 03:28
dalek rrot: r46458 | mikehh++ | trunk/src/embed.c:
fix codetest failure - line length
 03:29
Tene I think I like this laptop. I just built parrot in: 03:35
real\t0m29.844s
user\t1m35.268s
03:45 LoganLK joined
plobsing do we have any tools for finding dead code? 03:48
bacek_at_work code coverage? 03:50
purl code coverage is cv.perl6.cz
bacek_at_work plobsing, this one
sigh...
msg mj41 Looks like code coverage is slightly broken. Where "slightly" == "totally for last (few) days" :) 03:51
purl Message for mj41 stored.
03:51 snarkyboojum joined
bacek_at_work msg mj41 Last one was on 22nd of April. 03:52
purl Message for mj41 stored.
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33733), fulltest) at r46458 - Ubuntu 10.04 amd64 (g++) 03:54
dalek kudo: 3d3893a | (Martin Berends)++ | tools/test_summary.pl:
tools/test_summary.pl now logs Parrot and Rakudo versions, and has a much better report example
 04:03
04:42 Andy joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33733), fulltest) at r46458 - Ubuntu 10.04 amd64 (gcc with --optimize) 04:43
that should be smoke (#33735) 04:44
mikehh needs a break - bbl 04:46
sorear Legally speaking, how does ext/nqp-rx work? 05:02
NQP-rx doesn't require CLAs. Parrot does.
So there's code being shipped with Parrot that never went through the CLA process. 05:03
dalek website: Chandon++ | Hybrid Threads 05:09
website: www.parrot.org/content/hybrid-threads
plobsing sorear: ext/ contains code we don't own. Maybe we should document that better somewhere... 05:11
dalek TT #1244 closed by plobsing++: compilers/imcc/cfg.c: emit warning in propagate_need() 05:21
TT #1244: trac.parrot.org/parrot/ticket/1244
sorear Fixes like that are nice. 05:31
is it bad that trac.parrot.org/parrot/timeline is the fastest way to review svn commits? 05:36
plobsing I think that's a "feature" of centralized version control 05:37
05:54 uniejo joined
dalek rrot: r46459 | plobsing++ | trunk/runtime/parrot/library/libpcre.pir:
remove some long-dead code
 05:57
rrot: r46460 | plobsing++ | trunk/examples/nci/sdl_blue_rectangle.pir:
eliminate .nci_call statements
rrot: r46461 | plobsing++ | trunk/examples/sdl/lcd/clock.pir:
get example to run
dukeleto Parrot_str_is_null and Parrot_load_bytecode should be (and are not) in embed.h
dalek parrot: 9208b3f | dukeleto++ | plparrot.c:
Convert STRING_is_null to Parrot_str_is_null
 05:59
06:08 aukjan joined
dalek TT #1624 created by plobsing++: [DEPRECATED] .meth_call 06:10
TT #1624: trac.parrot.org/parrot/ticket/1624
rrot: r46462 | plobsing++ | trunk/DEPRECATED.pod:
deprecate .meth_call
 06:13
cotto hio 06:15
sorear Hello.
cotto seen jimmy 06:16
purl jimmy was last seen on #parrot 148 days, 4 hours, 51 minutes and 32 seconds ago, saying: good moring, #parrot [Dec 13 01:24:31 2009]
cotto seen jimmyz
purl jimmyz was last seen on purl 14 hours, 53 minutes and 32 seconds ago, saying: <private message>
cotto much better
purl i heard much better was to design each table in isolation with the table name providing context
sorear purl, no much better is <reply>
purl OK, sorear.
sorear much better
purl rumour has it much better is to design each table in isolation with the table name providing context
cotto not much better
sorear purl, forget much better
purl sorear: I forgot much better
sorear purl, die
purl ?4$??1???H????MD$FLNO CARRIER
cotto It's a losing battle.
bacek_at_work Don't offend my little stpid sister! 06:19
cotto purl can't be offended 06:23
purl cotto: excuse me?
sorear bacek_at_work: Are you involved with codestring? 06:26
bacek_at_work sorear, little bit... I broke it (as you mention early) 06:27
sorear has codestring working. Well.
06:28 JimmyZ joined
JimmyZ hello cotto 06:28
sorear Where do I report PCT bugs? 06:29
JimmyZ trac
sorear whose trac? 06:30
cotto hi JimmyZ
sorear, trac.parrot.org
JimmyZ cotto: there is already macro __C before r46434 06:31
I renamed it to CONTEXT_STRUCT
then I use it. 06:32
cotto: see r46434
plobsing do any of our tests run 'parrot -O'? 06:34
cotto I added a version of make test for -O2 but it's super broken. 06:35
plobsing how do I run it?
cotto I took it out.
It was testO2 06:36
plobsing :-(
cotto chromatic said he couldn't remember when imcc optimizations didn't break something.
plobsing I think I have a fix for TT #1209, but I'm concerned I may have a regression somewhere.
changing things without tests is worrysome to me 06:37
JimmyZ cotto: Is it persuasive?
sorear imcc optimizes?
plobsing sorear: you can have it right or you can have it fast. pick 1. 06:38
cotto sorear, only if you tell it to
sorear didn't realize PIR was optimizable
cotto JimmyZ, you mean your reply to chromatic?
plobsing sorear: you can run dead code elimination, better register allocation, opcode selection, etc 06:39
sorear other than register numbering, what can a PIR compiler /change/?
ah
I thought opcode selection was fixed
JimmyZ cotto: either
plobsing sorear: compare '$I0 = $I0 + 1' and 'inc $I0' 06:40
JimmyZ cotto: I think I should recall 'C89 doesn't like it'
sorear plobsing: the first should always generate add_i_i_i, imho
cotto Yeah. That needs some expansion or correction. 06:41
sorear it's going to be very suprising if it's ever changed...
JimmyZ cotto: It's disputed.
plobsing sorear: if you ask it to optimize... it'll optimize. Haven't you ever stepped through optimized code with gdb and gotten dizzy? 06:42
sorear no, I don't trust source-level debuggers 06:43
(display/i $eip)++
I suppose that works, yes, but ... I dunno, optimization just feels wrong in an assembler 06:45
plobsing I could say the same thing about assemblers in virtual machines 06:46
moritz wow, latest parrot causes some rakudo regressions 06:52
./perl6 t/spec/S06-signature/positional.rakudo
Multiple declarations of lexical '$'
moritz verifies that nobody touched the test 06:53
last touched in Feb
plobsing moritz: that's a result of a longstanding parrotbug I recently fixed
moritz oh :-) 06:54
plobsing You can't declare a .lex multiple times in one sub
moritz and that broke mulitple anonymous parameters in rakudo
multi sub two($, $) { ... }
kinda makes sense, more or less 06:55
plobsing++ # fixing bugs - even if they break Rakudo :-) 06:58
dalek TT #1209 closed by plobsing++: parrot -O Fails Tests 06:59
TT #1209: trac.parrot.org/parrot/ticket/1209
plobsing does it even make sense for rakudo to be putting *anonymous* arguments into the lexpad? 07:00
moritz I think it was easier to bind them than to special-case them in the binder
jonathan++ is the one who implemented it 07:01
my ($, $, $foo) = 1, 2, 3, 4; also breaks 07:02
dalek rrot: r46463 | plobsing++ | trunk/compilers/imcc/pbc.c:
Unconditionally generate debug information. Before turning on optimization
moritz and we have random segfaults, still or again (not sure) 07:08
plobsing moritz: random in location, time, or both? 07:12
dalek TT #1625 created by sorear++: PCT generates usage message too early 07:16
TT #1625: trac.parrot.org/parrot/ticket/1625
p-rx: 9d4c8a9 | sorear++ | src/NQP/Compiler.pir:
Add to command options, not overwrite them

accessible because the "larger" list of command options we used didn't include it).		 07:17
sorear after .lex "foo", $P17; is $P0 = find_lex "foo" and store_lex "foo", $P0 in every way equivalent to $P0 = $P17 and $P17 = $P0 respectively? 07:21
plobsing AFAIK, yes. Unless your HLL maps LexInfo in an evil way. 07:25
sorear hmm, that's a suprise 07:27
POST::Compiler is almost 4x slower than PAST::Compiler 07:28
(I'm on branches/codestring so it's using StringBuilder internally)
07:31 viklund joined 07:32 fperrad joined 07:43 fperrad_ joined
dalek TT #1142 reopened by plobsing++: test 12 of t/pmc/eval.t fails in testr (passes in other runcores) 07:49
TT #1142: trac.parrot.org/parrot/ticket/1142
rrot: r46464 | plobsing++ | trunk/t/pmc/eval.t:
Intentionally break eval tests wrt TT #1142 for everyone.
 07:52
07:52 Maddingue joined 08:35 kurahaupo joined
mj41 msg bacek_at_work Code coverage probably fixed. 08:50
purl Message for bacek_at_work stored.
08:59 bacek joined
bacek mj41++ # Hooray! We've got TapTInder back 09:03
09:09 snarkyboojum left
dalek rrot: r46465 | bacek++ | trunk/t/oo/metamodel.t:
Uncomment proper todo test for TT #1619 and instanly claim it as done.
 09:13
TT #1619 closed by bacek++: t/oo/metamodel.t: implement todo-ed item: new opcode makes working ... 09:27
TT #1619: trac.parrot.org/parrot/ticket/1619
tracwiki: v29 | bacek++ | GCTasklist 09:44
tracwiki: Add link to GCSweepFreeImplementation
tracwiki: trac.parrot.org/parrot/wiki/GCTaskl...ction=diff
tracwiki: v30 | bacek++ | GCTasklist
tracwiki: trac.parrot.org/parrot/wiki/GCTaskl...ction=diff
bacek msg Coke trac.parrot.org/parrot/ticket/1556 this one is for you :) 09:58
purl Message for coke stored.
dalek TT #978 closed by bacek++: Remove _synchronize from PMC Struct 10:00
TT #978: trac.parrot.org/parrot/ticket/978
TT #1584 closed by bacek++: STRING Out parameters in Parrot_str_* functions
TT #1584: trac.parrot.org/parrot/ticket/1584
tracwiki: v31 | bacek++ | GCTasklist
tracwiki: _synchronize is gone
tracwiki: trac.parrot.org/parrot/wiki/GCTaskl...ction=diff
rrot: r46466 | bacek++ | trunk/t/pmc/class.t:
Use test_more.pir instead of hand-crafted imports from Test::More.
 10:02
10:02 rurban joined 10:09 mikehh joined
dalek TT #1539 closed by bacek++: Remove deprecated Parrot_string_cstring function 10:16
TT #1539: trac.parrot.org/parrot/ticket/1539
TT #234 closed by bacek++: mk_gitignore.pl (lib/Parrot/Manifest.pm) needs svk 10:33
TT #234: trac.parrot.org/parrot/ticket/234
TT #170 closed by bacek++: Intermittent failures in t/op/arithmetics_27.pir
TT #170: trac.parrot.org/parrot/ticket/170
rrot: r46467 | bacek++ | trunk (2 files):
Rewrite mk_gitignore to generate .gitgignore from MANIFEST.SKIP and regenerate .gitignore
 10:35
rrot: r46468 | bacek++ | trunk/t/op/calling.t:
Add regression test for TT#133.
10:36 jsut joined
dalek TT #133 closed by bacek++: methodtailcall broken 10:49
TT #133: trac.parrot.org/parrot/ticket/133
TT #508 closed by bacek++: JSON libraries should be renamed to something less confusing
TT #508: trac.parrot.org/parrot/ticket/508
10:52 khairul joined 11:09 clinton joined
dalek TT #972 closed by bacek++: Remove "old" context related function such as "parrot_gc_context" 11:22
TT #972: trac.parrot.org/parrot/ticket/972
11:23 khairul joined 11:51 whiteknight joined
bacek whiteknight, aloha 12:10
dalek TT #894 closed by bacek++: Grant commit access to /languages to bacek 12:11
TT #894: trac.parrot.org/parrot/ticket/894
TT #835 closed by bacek++: Crash when appending to subclasses of ResizablePMCArray
TT #835: trac.parrot.org/parrot/ticket/835
12:12 szabgab joined
dalek rrot: r46469 | bacek++ | trunk/t/pmc/resizablepmcarray.t:
Add regression test for TT#835
 12:12
whiteknight good morning bacek 12:13
bacek whiteknight, fsvo "good" and "morning" :)
whiteknight bad afternoon, bacek 12:14
bacek much better!
pmichaud ...why is the parrot release tomorrow? Did Parrot's schedule change? 12:15
bacek wave from future to whiteknight
pmichaud (good morning, #parrot) 12:16
bacek pmichaud, hi
whiteknight isn't tomorrow the second tuesday of the month?
moritz yes
whiteknight ...is parrot released on the second tuesday of the month?
pmichaud I thought parrot released on the third tuesday of each month.
moritz and traditionally the parrot relesae is on the 3rd, no?
whiteknight is it? Am I going completely stupid?
bacek someone have to learn how to count to 3 :)
whiteknight thinks he has answered all his own questions
moritz just absent-minded :-) 12:17
whiteknight damnit
fart, crap, damnit
bacek whiteknight, hint - there is calendar on www.parrot.org with release date :)
12:20 bluescreen joined
dukeleto msg darbelo could you give the definition of NFG in www.parrot.org/content/what-nfg-why...ot-have-it ? :) 12:22
purl Message for darbelo stored.
dukeleto 'ello
12:27 ruoso joined
dalek TT #813 closed by bacek++: [RFC] Deprecate (parts) of OrderedHash. 12:27
TT #813: trac.parrot.org/parrot/ticket/813
12:40 bluescreen joined
pmichaud ...who did TT #1595? 12:42
is downcase now a three-argument op?
bacek pmichaud, yes. 12:46
(And probably I did it)
pmichaud ???
why would downcase need *three* arguments?
bacek all in-place ops are deprecated
pmichaud I'm confused.
dukeleto pmichaud: JimmyZ did that
pmichaud a two-argument downcase isn't inplace. 12:47
moritz $x = downcase $z
bacek ah
moritz would work out-of-place
with two args
bacek sorry, I misread it
pmichaud downcase $S0, $S1 # not inplace
shall I fix or will someone else?
dukeleto pmichaud: i don't like the look of the patch attached to that ticket. it needs tests
pmichaud: please fix, if you have time
pmichaud dukeleto: it would be difficult to come up with a test for that particular issue.
dukeleto pmichaud: difficult or impossible? 12:48
bacek pmichaud, TT#1595 is not about downcase. It's "substr" and "replace"
pmichaud bacek: it also changes downcase
- piropsig['downcase'] = 'Ss' + piropsig['downcase'] = 'SSs'
dukeleto pmichaud: trac.parrot.org/parrot/changeset/46100 12:49
bacek pmichaud, it's in ticket. Actual commit doesn't change it
pmichaud dukeleto: the new code "works", but it causes an unneeded conversion of constant strings into string registers
dukeleto pmichaud: doesn't look like it actually touched downcase
pmichaud bacek: okay, the ticket says it was applied though
dukeleto pmichaud: i understand now
pmichaud ticket should say "applied the changes for substr and replace, rejected downcase" :-)
bacek pmichaud, "language barrier"
pmichaud np
dukeleto just fixed the TT description 12:50
pmichaud: check now
12:51 iblechbot joined
pmichaud dukeleto: looks funny on my system, now. 12:52
moritz seems to miss {{{...}}} for verbatim text 12:53
better now
pmichaud still looks weird to me. but instead of changing the description, would be better to leave the original description in place and simply say that the downcase part of the patch was not applied. 12:54
dukeleto pmichaud: the original description was unreadable to me
pmichaud that way it's more obvious that downcase needed to be left alone.
dukeleto pmichaud: trac.parrot.org/parrot/ticket/1595#comment:3 12:56
pmichaud weird, my page isn't showing a comment #3
Coke msg plobsing - I'll see about assigning some trac tickets...
purl Message for plobsing stored.
Coke plobsing++ 12:57
moritz pmichaud: it seems reload is a bit unreliable
pmichaud moritz: indeed, it does.
that's...... unfortunate.
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33753), fulltest) at r46469 - Ubuntu 10.04 i386 (g++) 13:00
t/pmc/packfile.t - TODO passed: 34 in make coretest, make test, testb, testf and testr
t/op/exit.t - TODO passed: 6 in testf
don't get the TODO passes on amd64
dalek rrot: r46470 | coke++ | trunk/docs/book/pct/ch03_compiler_tools.pod:
doc patch courtesy tcurtis++ (TT #1556)
 13:01
dukeleto pmichaud: holding shift while reloading usually makes most browsers not use cache 13:02
pmichaud dukeleto: I did shift+reload, yes.
my browser (firefox) isn't reading from cache, afaict
it's definitely sending anothe request.
dukeleto I want a "byte array/bit vector" PMC in parrot. 13:05
Coke so, I got a good report from our most memory challenged builder of rakudo for branches/codestring. I'll merge it back today. 13:06
13:11 JimmyZ joined
JimmyZ Sorry, TT #1595 was wrong description 13:11
Did I hurt PCT? 13:12
Coke wonders if having plobsing means we don't need pirc. 13:14
dukeleto JimmyZ: nope, the description was just a bit confusing
JimmyZ: no harm done :)
Coke: why so? 13:15
Coke dukeleto: because he just closed a dozen or so IMCC tickets that have been been open for years. 13:16
moritz somebody with a mind just as crazy as the original author? :-) 13:17
Coke (and recent efforts to integrate pirc have failed)
free karma - go through DEPRECATED.pod and remove tickets listed that are eligible where the ticket is closed.
dalek TT #1556 closed by coke++: Wrong path and description of mk_language_shell.pl in PCT Book
TT #1556: trac.parrot.org/parrot/ticket/1556
TT #1626 created by coke++: Re: [parrot-tickets] Newbie reporting doc inconsistency
Coke (ala #1584)
dalek TT #1626: trac.parrot.org/parrot/ticket/1626
Coke and close out 1626 as an inadvertent opening. Apparently you CAN open tickets via email. :P 13:18
(er, close it as fixed)
dalek rrot: r46471 | coke++ | branches/codestring (68 files):
merge latest changes from trunk
 13:19
13:21 davidfetter joined
Coke msg sorear: on #perl6 you mention an nqp-rx patch - is that applied yet? If not, where is it? 13:23
purl Message for sorear stored.
JimmyZ dukeleto: thanks 13:25
Coke: github.com/perl6/nqp-rx/tree/topic/...ing-target 13:27
13:29 JimmyZ joined
bacek wear "Better to ask for excuse" and dcommiting some stuff 13:33
dalek TT #508 reopened by coke++: JSON libraries should be renamed to something less confusing 13:34
TT #508: trac.parrot.org/parrot/ticket/508
rrot: r46472 | bacek++ | trunk (4 files):
Implement Test::More.done_testing. Not perfect and require
 13:35
bacek seen chromatic 13:37
purl chromatic was last seen on #parrot 3 days, 9 hours, 37 seconds ago, saying: Looks reasonable. [May 7 04:36:25 2010]
bacek msg chromatic Please review r46472. I'm not sure that I grok Test::Builder fully.
purl Message for chromatic stored.
dukeleto bacek: r46472 needs some tests. I will attemp to write some if chromatic doesn't beat me to it 13:38
bacek dukeleto, I added simple test. But more tests are usually good :) 13:40
dukeleto bacek: yes, they are :) 13:41
davidfetter oh hai
bacek davidfetter, o/
dukeleto davidfetter: hola 13:42
davidfetter ¿que paso?
dukeleto is working on three laptops and a desktop at the same time, before 7am. There was a screaming across the sky. 13:49
13:50 Andy joined
moritz finds two keyboards confusing enough 13:50
dalek TT #1627 created by dukeleto++: tests for Test::More::done_testing
TT #1627: trac.parrot.org/parrot/ticket/1627
moritz I have that reflex to hit a more "dangerous" key when a program doesn't react 13:51
first Ctrl+Q, then Ctrl+D, then Ctrl+C
so I've killed various programs because I stared at the wrong monitor (or used the wrong keyboard) and wondered why programs wouldn't react
14:00 rurban_ joined 14:02 PacoLinux joined 14:15 atrodo joined
Coke ahhh. ok. so pmichaud was right, and this is one of the places where assuming a Codestring is really a String is expensive. 14:16
sorear++ 14:17
msg pmichaud any objections to merging sorear's topic/no-codestring-target nqp-rx branch to master along with parrot's codestring branch to trunk ? 14:22
purl Message for pmichaud stored.
Coke msg pmichaud (I may just do both together later today) 14:23
purl Message for pmichaud stored.
dalek rrot: r46473 | gerd++ | trunk/tools/docs/filename_and_chapter.pl:
add a chapter; make the upper bound changeable
 14:24
Coke sorear: yup, that does it; rakudo builds for me here with: user 4m23.465s 14:26
moritz as opposed to how long normally? 14:27
Coke as opposed to pretty much INFINITE with branches/codestring. =-)
let me try against a pristine trunk.
14:30 bubaflub joined
Coke It's faster in trunk: user 3m51.563s 14:37
I didn't compare memory usage, however.
(nor can I ratchet down ulimit on my OS X box, for some reason, it fails to do anything.)
lemme patch ext/nqp-rx in branch... 14:38
msg andy what was that brainstorming visualization tool you were using? 14:43
purl Message for andy stored.
moritz seen andy 14:44
purl andy was last seen on #parrot 3 days, 12 hours, 14 minutes and 31 seconds ago, saying: embed.h is going to get headerized. It's the last holdout [May 7 02:29:48 2010]
moritz seen andya
purl andya was last seen on #parrot 73 days, 19 hours, 23 minutes and 23 seconds ago, saying: Back's fine, thanks :) [Feb 25 19:20:59 2010]
Coke seen adougherty? 14:45
purl I haven't seen 'adougherty', Coke
Coke supposes he will test the low memory build of rakudo on feather later if no one beats him to it.
Coke added the modified nqp-rx to branches/codestring if someone wants to try it out. 14:46
14:55 ruoso joined
Andy Coke: freemind 14:57
purl somebody said freemind was a mind mapping tool or freemind.sf.net/ or ...your ass will follow...
dalek rrot: r46474 | coke++ | branches/codestring/ext/nqp-rx/src/stage0 (4 files):
Use sorear++ 's no-codestring-target branch of nqp-rx.
rrot: r46475 | dukeleto++ | trunk/NEWS:
Improve our NEWS for 2.4.0
Coke thanks, andy. 15:00
dalek rrot: r46476 | dukeleto++ | trunk/NEWS:
Improve NEWS some more for 2.4.0
 15:14
15:50 lee_ joined 15:53 krunen joined 16:01 patspam joined
cotto_work hio 16:03
16:11 lucian joined, theory joined 16:40 iblechbot joined 17:00 darbelo joined 17:12 leejr joined
leejr I just spent half an hour looking for the languages repository. I found it, but it would have been much easier if the downloads page mentioned the languages repo 17:22
the trac site doesn't mention how to get the languages either. 17:23
I found it by inferring, from the trac website, where the languages repo was. 17:24
svn list wasn't any help from the parrot repo
Coke I just linked www.parrot.org/download to trac.parrot.org/parrot/wiki/Languages 17:26
so, for lolcode, e.g., that would then link to: trac.parrot.org/languages/browser/lolcode/trunk
dalek tracwiki: v128 | coke++ | Languages 17:28
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
Coke also added a link at trac.parrot.org/parrot/wiki/Languages to trac.parrot.org/languages
Thanks.
cotto_work leejr, thanks for taking the time to report being annoyed by something of ours. 17:32
leejr Am I supposed to know that the trac url mimics the svn url? 17:33
I could find the trac, but I didn't know how to translate that into getting the software. 17:34
Coke I'll add a note on the Languages page about that. 17:37
leejr oh, I'm supposed to git it
Coke ... no, =-) 17:39
leejr languages sold separately. ;-)
Coke all the stuff hosted on trac.parrot.org (languages or core) is all svn yet.
trac.parrot.org/parrot/wiki/Languages updated. 17:40
(if those were completely separate projects, they'd all have their own page.)
(with their own "get the source" instructions). I guess I should actually update the root page of the languages trac instance.
leejr the relationship between trac and svn is not made explicit 17:41
Coke leejr: trac.parrot.org/languages/wiki/WikiStart 17:44
how's that?
17:44 khairul_ joined
Coke (i'll remove the svn info from parrot's Languages page and just leave the pointer there.) 17:44
dalek tracwiki: v129 | coke++ | Languages 17:45
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
leejr In other words, nobody told me explicitly to type 'svn co svn.parrot.org/languages' 17:46
Coke msg sorear - at .25G, rakudo doesn't build. at 0.5G (ulimit), trunk is faster for me. I can't win. =-) 17:47
purl Message for sorear stored.
Coke leejr: yes, and trac.parrot.org/languages now tells you this.
is that sufficient?
leejr eggsellent
btw, cool project 17:53
I looked at it because I was wondering if instead of a C++ DOM.document object at its core, Firefox had a Parrot DOM.document 17:57
maybe dumb idea.
Coke that's come up before, mainly as a "wouldn't it be nice if you could write client side perl6" 18:01
dalek tracwiki: v130 | coke++ | Languages
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
leejr that was my thinking as well 18:02
in other words, the document would really be a parrot file... a local proxy url that any language could work on 18:03
the browser downloads the webpage, turns it into a parrot object then gives the tmp filename as an argument to any script 18:07
must reflect more on this. but I think it could be done using the parrot API. and implementing the api in each language. 18:09
with a dom api wrapper
18:13 joeri joined 18:19 khairul joined
darbelo o/ 18:21
Tene \\o
treed o/ 18:22
darbelo o/~ o/~
19:05 shockwave joined
sorear messages erase 19:12
Coke sorear: hio. 19:24
sorear yes? 19:27
Coke applied your nqp-rx branch to parrot's branches/codestring - I find that at a virtual memory limit of 0.5G, the branch is slower in user: time than trunk. 19:30
(at 0.25G, it fails to compile in both.)
sorear try an rss limit of 0.35G 19:33
also, are you comparing it to trunk-now or trunk-when-branched? 19:34
oh, you merged
Coke both at HEAD. 19:35
dalek rrot: r46477 | darbelo++ | trunk/MANIFEST:
Regenerate MANIFEST
 19:37
Coke with a ulimit -v of 367001, parrot trunk runs OOM. 19:45
(building rakudo)
Coke must remember to ulimit -c 0 when running something that could memory PANIC. :P 19:51
sorear does ulimit -m do anything useful? 19:52
GeJ Good morning everyone.
sorear Good morning.
purl For you maybe.
cotto_work good moring GeJ
19:53 dalek joined
shockwave purl paste 20:09
purl paste is (see: nopaste) or like glue but a little safer to sniff. or nopaste.snit.ch:8001/ or scsys.co.uk:8001/ anywhere shadowpaste is or mmm, sticky paste or You there! Eating the paste. or <see> 2 girls, 1 paste or App::Nopaste or toxic for bots and humans
Coke sorear: ok. at ulimit -v 367001, I see the sweet spot where branches/codestring beats trunk. 20:11
(3m5s user time to build rakudo vs. "BOOM")
sorear Coke: Does ulimit -m not work?
Coke it might. bash warns "many systems may not honor this limit" 20:13
nopaste "shockwave" at 192.168.1.3 pasted "code to illustrate question." (44 lines) at nopaste.snit.ch/20520
Coke as I understand it, -v overrides that anyway. (you can't have a -m > than -v)
sorear -m doesn't make programs go BOOM though
-v doesn't make programs slower, in general 20:14
-v either has no effect, or makes them crash
-m makes them slower
shockwave I have 2 simple questions about declaring .local directives. I'll reffer to the paste for illustration.
Coke shockwave: shoot.
sorear: k. I'll try it that way. 20:15
shockwave 1) Does it make any memory/speed difference if .local directives are declared in a position where they could be skipped by code, as in the .locals's _i, _hier, ... or does it not matter where in the body they are declared?
sorear it does not matter where in the body 20:16
.local does not generate any code
shockwave I probably pasted more code than I needed for the answer, but I just grabbed the first sub that came to mind with it.
darbelo shockwave: .local defines an alias to a register.
shockwave ah, that kinda answers my next question then:
2) Is there any speed difference between .local int _i, and $I0 20:17
I guess the answer is no?
Coke certainly not at runtime.
shockwave Coke, sweet. Runtime is what matters to me, in this case.
Thanks, guys.
Coke at compile time, it might be theoretically faster. (but not practically so.)
shockwave Coke, I hear that. 20:18
sorear can't use isa?
Coke sorear: and that was going to be my question about his example. =-) 20:19
shockwave sorear, I don't think so because... hold on. Let me remember the reason.
Coke I assume that his type hierarchy is outside of parrot's.
sorear there is no "Parrot type hierarchy" 20:20
there's a default implementation of isa, sure
shockwave still thinking...
purl it has been said that thinking is see `made of meat`
20:20 smash joined
smash hello everyone 20:20
Coke sorear: Depending on how you define type hierarchy, sure. 20:21
sorear but there are no rules when it comes to using better vtables
shockwave oh, right. I remember,
Coke sorear: I don't know what you mean by "better vtables" 20:22
shockwave Because I'm comparing the class' typeid agains an instance, as: var a = someInstance is SomeClass
someInstance is SomeClass translates to:
sorear isa someInstance, someClass doesn't work? 20:23
shockwave '%TypeOf'(11, $P#), where 11 is the typeid of SomeClass.
Coke er... where are you getting that type id?
shockwave Coke: compiler.
purl compiler is, like, a controversial feature of Perl5.005 which will probably be used for evil ends or a way for the unenlightened to make themselves think their programs will run faster
Coke shockwave: so it's a unique ID you're tracking about about your own pmcs/objects?
shockwave Coke, it's a statically typed language, so the compiler has the type info. 20:24
Coke (not parrot's internal integral type id?)
shockwave Coke, nope.
Coke hokay.
shockwave Coke, the compiler creates an unique type-id for each class.
dalek rrot: r46478 | darbelo++ | trunk/t/library/test_builder_done_testing.t:
[CAGE] Add svn props to t/library/test_builder_done_testing.t
shockwave Don't laugh: The main reason behind that is because: 20:25
The compiler's symbol table is actually an embeeded database. Sqlite, to be exact.
You would think that it would run slow as hell, but, I'm confident it will be fast enough. And, as it's a database, you can do all kinds of neat things with it. 20:26
Each entry of a class within the symbol table (database) has an unique-id, which works wonders in so manny situations. 20:27
20:28 Themeruta joined 20:32 ash_ joined
Coke sorear: with -m at that level, trunk builds in user: 2m35s, branch in 3m6s. 20:40
so I really don't see any benefit anymore, sadly. 20:41
(granted, I'm testing this on feather, which is a VM with mannnnny other things going on.)
I see similar ratios with my OS X box (which doesn't seem to respect any ulimits) 20:43
so the only time I see any benefit to this branch is on feather, with a ulimit -v of 0.35G, where trunk explodes and branch doesn't. 20:44
20:54 bluescreen joined
Coke OTOH, glad that trunk is faster. 21:00
nopaste "shockwave" at 192.168.1.3 pasted "Lengthy question." (50 lines) at nopaste.snit.ch/20521
Coke seen chromatic?
purl chromatic was last seen on #parrot 3 days, 16 hours, 23 minutes and 55 seconds ago, saying: Looks reasonable. [May 7 04:36:25 2010]
shockwave I posted a question as a paste, since it's lengthy. If someone wants to take a stab at it, I'll be around.
Coke shockwave: yes, you can't tell in parrot if a $I0 is really just a 0/1 bitval. 21:01
21:01 Whiteknight joined
Coke (well, aside from checking to see if its value is 0 or 1) 21:01
21:01 fperrad joined
Coke if you wanted to know this at compile time, you could check. 21:01
er, s/check/keep trck/
if you want to know arbitrarily at runtime, yah, I think you need a PMC. 21:02
shockwave Coke, the compiler knows. The problem is that the info is lost at runtime.
Coke you can't make an I register more specific.
21:02 gpw joined
shockwave Coke. Yeah. I may have to live with that ambiguity for now. I kinda figured. :( 21:02
Coke are you willing to consider an Integer that happens to have a value of 0 or 1 at runtime a Boolean? 21:03
(I'm guessing now.)
(no)
shockwave Coke, that's going to be the closest aproximation that I can make. Excelent guess, btw.
Coke =-) 21:04
shockwave I wonder how languages like Java, C# and ActionScript deal with this.
Coke I ended up dodging all this in partcl by saying "everything's a PMC" and hoping that by the time I got back to it, something like PCT could track that for me.
shockwave Coke. Having simple I registers is so sweet, though. 21:05
That's about as fast as something is going to be, I would assume.
21:06 fperrad_ joined
shockwave I'll keep thinking about it for some time in the back of my mind. If I think of something, I'll mention it. 21:09
err, is $I0 ... the same as: .local pmc _i = new 'Integer' ? 21:14
Or is $I0 more efficient. 21:15
Coke $I0 is not an Integer.
it's an int.
more efficient.
$I0 is .local int _i
(also, you sadly can't do .local and = on the same line. =-) 21:16
shockwave Coke: Cool, I ask because of the type of error message generated by $I0:
Method '%getType' not found for invocant of class 'Integer'
Coke no methods on int register. 21:17
only on PMCs.
shockwave Coke: Yep, I found about the not being able to initialize .local declarations the first day. I was surprised.
Coke is this on your previous nopaste?
shockwave Coke: Yeah. I know why the error happens. It's just the message that made me think about the question. 21:18
Coke shockwave: github.com/partcl/partcl/blob/maste...os.pir#L62
if you're hand-rolling PIR that can come in handy.
(you'd invoke it with .int(foo, 3), .g.
er, e.g. 21:19
shockwave Coke, That code is too sweet. I'm gonna have to steal it.
Coke distillation of years of fighting PIR. enjoy. =-) 21:20
shockwave :-) 21:21
Coke (I finally gave up once nqp-rx existed.) (pmichaud++)
see also hll_macros.pir in core.
shockwave What was partcl? I'm only vaguely familiar with nqp (I don't Perl).
Coke partcl? 21:22
purl i think partcl is tcl on parrot or code.google.com/p/partcl
Coke partcl-nqp?
purl well, partcl-nqp is github.com/partcl/partcl-nqp
shockwave I though partcl was the name of a project you worked on.
Coke yes. 21:23
those are mine.
shockwave Nice.
Coke danke. -> 21:24
shockwave Latest commit was last week. Are you still working on it?
On and off, maybe? 21:25
21:29 tetragon joined
shockwave I guess my question was dumb. Sorry about that. 21:31
21:38 tcurtis joined
shockwave I'll end up not using the macros file, since, at least for the first version of the compiler, I would like the satisfaction of claiming "100% original code" :-) 21:40
darbelo is a big fan of having 100% original bugs. 21:41
Debugging my code has always been more fun thatn debugging someone else's.
shockwave It's relieving to know that the bug in the program was created by you. 21:42
cotto_work seen khairul 21:43
purl khairul was last seen on #parrot 1 days, 13 hours, 40 minutes and 18 seconds ago, saying: hi sorear, i think the one doing the RTEMS port is bubaflub? [May 9 08:02:46 2010]
darbelo "Oh, for the love of $deity. This was coded by complete morons^W^Wme." 21:44
cotto_work msg khairul Next time you need something, ping me in #parrot. It's too easy for me to miss privmsg windows if I'm not expecting them. 21:50
purl Message for khairul stored.
smash anyone knows who is mentoring the parrot on rterms GSOC project ? 21:51
ash_ shouldn't it be on the wiki or in the accepted purposal? 21:52
darbelo It got accepted with RTEMS, but that info should be available in the gsoc page.
socghop.appspot.com/gsoc/program/li...e/gsoc2010 21:53
If your browser can handle that javascript.
Jonathan leto is listed, with Chris Johns and joel sherrill co-mentoring. 21:54
socghop.appspot.com/gsoc/student_pr...7230763613
Chris Johns is the one that had gotten parrot running on RTEMS the first time around. 21:56
He goes by kiwichris on IRC.
smash darbelo: great, thks 21:59
22:00 rurban_ joined
darbelo smash: Oh, and in case it was unclear Jonathan Leto is out very own dukeleto. Pinging him on all matter GSoC is generally a good Idea. 22:06
22:07 snarkyboojum joined 22:11 shockwave left
smash darbelo: yeap, i kknow dukeleto from irc :) thks again 22:13
darbelo Excellent. Just making sure. 22:14
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33760), fulltest) at r46478 - Ubuntu 10.04 i386 (gcc with --optimize) 22:17
t/pmc/packfile.t - TODO passed: 34 in make coretest, make test, testb, testf and testr
t/op/exit.t - TODO passed: 6 in testf
I think I mentioned before - the TODOs do not pass on amd64 22:20
dalek TT #1628 created by NotFound++: Unescape parrot string to parrot string 22:25
TT #1628: trac.parrot.org/parrot/ticket/1628
bacek ~~ 22:28
mikehh howdy bacek
darbelo bacek: I have another Crazy String Idea for you. 22:29
bacek mikehh, partially recharged and ready for new day :)
darbelo, do tell!
darbelo Chop off the 'encoding' pointer.
bacek darbelo, I like it 22:30
darbelo Probably, the way to do that is to make the encoding a property of the charset.
bacek darbelo, hang on. encoding? Not chartset?
s/chartset/charset/ 22:31
darbelo I keep mixing them up. The functions in charset call the ones in encoding?
If not, then reverse my reasoning ;) 22:32
bacek They are hardly twisted...
encoding is one of utf-*, ucs and fixed-8
charsets are ascii, unicode, etc
darbelo Then charset is the one that has to stay. 22:33
NotFound_b I proposed several times to drop the charset thing, charset is always unicode, but been rejected.
bacek NotFound_b, let's try it again. 22:34
darbelo My idea is a little more conservative.
bacek darbelo, that I don't like it :) 22:35
darbelo ascii, binary and iso-8859-1 all imply a single encoding, if we chop up unicode into several 'this wide' sub-encodings we get to save a pointer. 22:36
dalek rrot: r46479 | bacek++ | trunk/NEWS:
Add news about string immutablity.
rrot: r46480 | NotFound++ | trunk (4 files):
experimental function Parrot_str_unescpae_string and String.unescape method, TT #1628
darbelo And we get to kill utf8 with fire.
bacek darbelo, no way.
NotFound_b darbelo: ascii and iso-8859-1 are unicode subsets with a very simpleencoding. 22:37
And for binary we can have encoding/decoding functions that throws unless explicitly asked for some transcoding.
darbelo NotFound_b: Exactly. I'm arguing we should try to do the same with other 'wider' subsets of unicode. Then the 'encoding' is a property of the (sub?)charset. 22:38
sorear Let me get this straight. shockwave is avoiding NQP-rx and doing straight PIR for political reasons?
bacek NotFound_b, looks about all right.
NotFound_b darbelo: but calling it encoding is more on the line of the unicode documents. 22:39
bacek darbelo, if all strings are unicode we don't need "charset"
22:39 shockwave joined
NotFound_b If we fully support unicode, using unicode jargon in our docs and code will be helpful. 22:40
shockwave I've been looking on docs.parrot.org and Google, but can't find it: Where the info for addproperty opcode?
darbelo bacek: True, but I think allison was against all-unicode. She wants parrot to be able to handle strings in 'native' format as well. Even if it isn't unicode. 22:41
NotFound_b shockwave: I think there is no such op.
Native format is an encondig, 22:42
shockwave NotFound_b: I'm looking for the opcode that's like addattribute, but it's not inherited and (I think) can be added to pmc's at runtime.
bacek shockwave, trac.parrot.org/parrot/wiki/Parrot%20Dictionary
darbelo, it can use "fixed_8". 22:43
shockwave woot! I did not know that page existed. Nice!
Thanks, bacek
NotFound_b shockwave: properties aren't added, just set.
bacek darbelo, actually I don't see any advantages of having "native" strings. We can use some kind of BinaryStream PMC for bloody binary data. 22:44
shockwave NotFound_b, do you remember the opcode to set them?
NotFound_b shockwave: setprop, getprop, delprop 22:45
shockwave Those aren't on the page bacek linked.
Thanks, NotFound 22:46
NotFound_b "native string" is shit. I can have a latin-1 locale now and a utf8 locale one minute later.
sorear shockwave: Let me get this straight. You are deliberately avoiding NQP-rx?
mikehh NotFound_b: I agree completely
shockwave sorear. Yes.
sorear I propose we rename NQP to Not Quite PIR. 22:47
NotFound_b The idea of native string leads to confusion between text and binary IO 22:48
sorear It's currently the one and only maintained parser and compiler engine for Parrot
Maybe if we remove the "Perl" from the name, people will 1. use it 2. have more fun with Parrot?
bacek NotFound_b, +1
NotFound_b, actually, +100500 22:49
darbelo sorear: So? We're not apple. He can use whatever toolkit he likes.
shockwave sorear, you may not be that off on that comment. I would have gone with C++ anyway, but the name Perl in the name didn't help.
Of course, that's just me. 22:50
It's just I've never written code in Perl.
NotFound_b sorear: NQP is a language, the parser and compiler engine tools are called PCT. NQP is just language commonly used with PCT. 22:52
You can use only PIR with PCT if you want to.
darbelo sorear: Also, even without 'perl' in the name it still has the syntax. It is a subset of perl 6, calling it anything else is pointless. 22:55
22:56 ash_ joined
sorear NotFound_b: How do you use PCT for parsing? 22:56
bacek PCT Grammar is subset of Perl6 too
sorear Oh, you consider PGE to be part of PCT 22:57
Tene PCT is more than just the Grammar. I've used the PAST and POST compilers from PIR and from other HLLs.
NotFound_b sorear: well, the grammar engine is also based on perl6, so you aren't completely free of perl6'nes anyway.
purl okay, NotFound_b.
cotto_work the grammar engine? 22:59
purl i think the grammar engine is just going to be another parrot compiler or based on perl6, so you aren't completely free of perl6'nes anyway.
dalek tracwiki: v2 | bacek++ | GCSweepFreeImplementation
tracwiki: Minor fix
tracwiki: trac.parrot.org/parrot/wiki/GCSweep...ction=diff
tcurtis shockwave, if you're writing a compiler to Parrot without using PCT(or at least the PAST stuff), you're going to be missing out on the PAST-level optimization stuff I'll be doing this summer for GSoC.
NotFound_b cotto_work: Isn't that what PGE stand for?
darbelo tcurtis++ 23:00
NotFound_b Oh, you talekd with purl :D
shockwave tcurtis, dang!
darbelo Of course, I've had a 'PAST' compiler on my TODO since last summer. So anything that outputs PAST will eventually benefit. 23:01
23:01 Psyche^ joined
bacek E.g PCT based PIR compiler :) 23:01
shockwave attempts to influence tcurtis to add an optimizer to imcc instead: Optimize IMCC instead, tcurtis.
bacek Optimize IMCC with fire! 23:02
sorear ==bacek
darbelo bacek++
shockwave Poor IMCC. It does it's job, yet no one is fond of it. (I don't hate it, yet). 23:03
darbelo "The code that runs fastest is the code that does not run at all."
darbelo goes all zen and stuff.
sorear shockwave: well that's the thing, it *doesn't*
darbelo shockwave: Read the source, you will.
23:03 bluescreen joined
shockwave darbelo. I may actually do that soon, just for kicks. 23:04
sorear IMCC crashes regularly and generates wrong bytecodes if you look at it from a bad angle
shockwave sorear, is there something about it that doesn't work?
:S
Tene pirc was pretty close to working for a while.
sorear And don't even bother trying to enable the IMCC optimizer
NotFound_b shockwave: look at the IMCC related track tickets and skipped and todoed tests.
shockwave Damn. I was reading some of the PDDs yesterday, and it mentioned something about an optimizer. I was like, hell yeah! 23:05
darbelo It is a bit like Hell. Yeah.
Coke (imcc crashes regularly) please open tickets when that happens, ok?
(generates wrong bytecodes) also, tickets. 23:06
sorear Coke: I know it crashes regularly because I'm subscribed to parrot-tickets
Coke RAWR YOUR STUFF IS BORKED is not helpful.
sorear This is also how I know about the miscompilations
Opening tickets for tickets I read about is less helpful
shockwave It hasn't crashed on me. Though, I don't think I've gotten too crazy with the code.
Tene Is there a ticket for too many open tickets?
darbelo Tene: Just open one, we can merge duplicates ;) 23:07
shockwave lol
Coke FYI, 14 active tickets with "segf" in the summary, of which 1 is identified as an IMCC bug.
msg plobsing: here's an IMCC ticket for you: #162. 23:08
purl Message for plobsing stored.
Coke (and that's not even segfaulting anymore. no one updated the ticket summary.)
Whiteknight: mind if I reassign #162 to plobsing? 23:10
Whiteknight ?
yes, please do
shockwave Obviously, a program shouldn't crash. But compilers shouldn't be generating that code anyway. 23:11
ticket 162 ^
Oops, what I meant to say is that IMCC shouldn't allow that syntax. 23:12
Tene There are many X for which IMCC shouldn't X
darbelo Coke: It's slightly wider than "We all have IMCC with no reason". I'm not seing anything wrong with IMCC *now*. But I've found (and then fixed, no ticket needed) one case of stack corruption that's been there since who knows when. I also don't know how many other bugs are there lurking waiting for me to step on the wrong stone. 23:13
I just do not like the feeling of impending doom associated that IMCC instills in me.
Coke darbelo: "there are undiscovered bugs" is not the same as "ZOMG ITZ TEH SUXX0RZ" (hyperbole added for emphasis)
darbelo Well, ITZ TEH SUXX0RZ as well ;) 23:14
Coke I'm not saying it's all hugs and puppies.
dalek TT #1118 closed by coke++: [PATCH] Added __END__ to pir
TT #1118: trac.parrot.org/parrot/ticket/1118
Tene "We keep finding bugs here again and again over time, and nothing has changed such that we now believe it to be sufficiently correct." is also different. 23:15
shockwave Speaking about tickets. I added ticket 1621, about Parrot not compiling under Windows. 23:17
Coke ugh. svn-- # tripping over an unversioned copy of a file during an svn switch seems to be fatal.
shockwave Compilation broken. (I updated 2 seconds ago).
Coke checks.
darbelo Awww. Again? We need more windows folks in here.
shockwave This is the basic error (a few of these, the same): src/debug.c:3249: error: structure has no member named `to_ctx' 23:18
Coke shockwave: are you using strawberry perl?
shockwave Coke, yep.
Coke k. I'll see if I can duplicate here.
shockwave I tried using the other method, MSVC
same thing. 23:19
Coke probably a missing PARROT_EXPORT
darbelo: we have taptinder
taptinder?
purl taptinder is continues integration tool - taptinder.org . For Parrot project running on tt.taptinder.org/ and reporting build failures to #parrot channel as ttbot.
Coke shockwave: did you realclean? 23:20
shockwave Not today. Let me try it.
Coke also, what svn revision?
(we're green on taptinder)
shockwave 46480, trunk
darbelo taptinder squawks when we break it, it doesn't fix it for us ;)
Coke there were some errors back on r46454 or so 23:21
Coke runs an svn update on his 3 month old windows parrot checkout. woof. 23:22
shockwave Coke. Did distclea. Still busted.
Coke (ah, svn is not borked on that svn switch. just took ages to resolve.) 23:23
shockwave: k. I'll try a build as soon as this svn update finishes.
shockwave Coke, cool.
Coke did you pass any options to Configure.pl?
shockwave no
Coke k. 23:24
shockwave? 23:25
purl shockwave is bad huh? or www.shockwave.com/bin/shockwave/err...atform.jsp
Coke forget shockwave
purl Coke: I forgot shockwave
shockwave shockwave, yeah?
23:25 gpw joined
Coke just wondering if purl knew who you were. 23:26
note that 'make distclean' doesn't seem to remove everything on my strawberry instance. 23:29
(svn st and svn st --no-ignore still show a lot of junk) cleaning up by hand.. 23:30
shockwave Coke, checking out a freshly minted copy. 23:31
Coke also, did you build with make -j?
mingw32-make -j, that is? 23:32
shockwave no
nope
perl Configure.pl
mingw32-make
Coke k.
23:33 janus joined
sorear svn st --no-ignore | cut -c 8- | xargs rm -f 23:34
shockwave Coke, it's still going. I think it's gonna finish. 23:35
(building)
Coke Yah, that doesn't help in cmd.exe. =)
shockwave Yey!
It did build.
You're right, it's distclean's fault. 23:36
Coke ... so something was hanging around somewhere. that sucks.
shockwave: yah, builds here too. 23:38
NotFound_b Maybe include/pmc isn't cleaned?
Coke should be easy enough to check on a fresh checkout. 23:39
shockwave Well, at least it compiles fine, and the problem is with the build system.
sorear Coke: What's the status of codestring? Abandoned and I should switch my testing back to trunk? 23:40
shockwave Thanks, Coke 23:41
Coke sorear: there is a single case where it's better in my testing. I was going to through it out to the list for more feedback than yours system and my system. 23:42
shockwave Someone please update ticket #1621
Coke I'll do that now, sorear.
shockwave (I don't have modify access)
Tene or someone give shockwave access to update tickets 23:43
:)
shockwave :p
When you're used to working with dual monitors, it sucks when you go somewhere and have to use only a little laptop monitor. 23:49
You gotta open windows often just to check status of things; as opposed to just glancing at the other monitor. 23:50
I code at least 300% slower with my laptop along.
s/along/alone
Coke sorear: sent 23:54
sorear excellent. 23:57
dalek rrot: r46481 | mikehh++ | trunk/src/string/api.c:
add missing ASSERT_ARGS
 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!