|
Parrot 4.6.0 "Wild Parrots of Telegraph Hill" | parrot.org/ | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC Set by moderator on 18 July 2012. |
|||
|
00:20
PacoAir joined
|
|||
| dalek | rrot: f2fb998 | jkeenan++ | t/src/checkdepend.t: [codingstd] Conform to cuddled-else and no-hard-tabs standards. |
01:56 | |
|
02:18
kid51 joined
|
|||
| dalek | kudo/sigilless: bf892d1 | moritz++ | src/core/Int.pm: switch Int to sigilless |
05:38 | |
| kudo/sigilless: 61d67fa | moritz++ | src/core/Rat.pm: use sigilless params in Rat |
|||
| kudo/sigilless: 0d98c32 | moritz++ | src/core/Complex.pm: switch Complex to sigillless params |
|||
| kudo/sigilless: 504e994 | moritz++ | src/core/ (5 files): switch Attribute, Enumeration, Hash, LoL and Order to sigilless params |
|||
| kudo/sigilless: 234ad61 | moritz++ | src/core/List.pm: switch List to sigilless params |
|||
| kudo/sigilless: 1f6fe81 | moritz++ | src/core/ (3 files): switch Numeric to sigilless params |
|||
| kudo/sigilless: 221379e | moritz++ | src/core/ (4 files): switch Numeric, Real and Cursor to sigilless params |
|||
| kudo/sigilless: 3d21d3f | moritz++ | src/core/ (2 files): switch Mu to sigilless params (mostly) |
|||
|
06:00
fperrad joined
07:15
brrt joined
07:56
lucian joined
|
|||
| dalek | rrot/threads: d215d68 | nine++ | src/scheduler.c: Move thrading initialization to Parrot_cx_init_scheduler In embedding situations, control flow may bypass Parrot_cx_begin_execution (using Parrot_api_pmc_invoke directly). Moving threading init to Parrot_cx_init_scheduler makes sure it is run for the first interp. |
08:17 | |
| rrot/threads: 3f147ee | nine++ | lib/Parrot/Pmc2c/PMC/Proxy.pm: Add some description to Parrot::Pmc2c::PMC::Proxy |
|||
| rrot/threads: 276a2f8 | nine++ | src/scheduler.c: Merge branch 'threads' of github.com:niner/parrot into threads Conflicts: \tsrc/scheduler.c |
|||
| rrot/threads: 25acd5a | nine++ | / (10 files): Merge branch 'threads' of github.com:parrot/parrot into threads |
|||
| kudo/sigilless: 30fab97 | moritz++ | src/core/ (8 files): switch more of the setting to sigilless params (Int, Junction, Parcel, Range, Rational, Routine, stubs) |
09:08 | ||
| kudo/sigilless: 018777f | moritz++ | src/core/Array.pm: switch Array to sigilless params |
|||
| kudo/sigilless: ddb2dd2 | moritz++ | src/core/operators.pm: Switch operators.pm to sigilless params |
09:09 | ||
| kudo/sigilless: 76b036d | moritz++ | src/core/metaops.pm: Switch metaops to sigilless params |
|||
| kudo/sigilless: 105cec6 | moritz++ | src/ (2 files): switch SAFE and CORE to sigilless params |
09:57 | ||
| kudo/sigilless: f75de7d | moritz++ | src/core/Mu.pm: remove the last use of \\$foo from the setting |
|||
| kudo/sigilless: dfff492 | moritz++ | docs/ROADMAP: update ROADMAP sigilless params are done; sigilless variables are still TODO |
|||
| kudo/sigilless: 86bc703 | moritz++ | src/Perl6/Grammar.pm: allow typed \\ and | in signatures without a name |
|||
|
11:09
nopaste joined
11:10
tuxit joined
11:27
schmoo joined
|
|||
| dalek | Heuristic branch merge: pushed 34 commits to rakudo/nom by moritz | 11:46 | |
|
11:51
PacoAir joined
|
|||
| dalek | kudo/nom: 449f30f | moritz++ | docs/ChangeLog: update ChangeLog |
11:55 | |
| kudo/nom: de84a76 | jnthn++ | docs/ChangeLog: Some more ChangeLog bits. |
12:03 | ||
|
12:10
whiteknight joined
|
|||
| whiteknight | good morning, #parrot | 12:11 | |
| Very interesting link about Java performance: paulbuchheit.blogspot.com/2007/06/j...an-c.html# | |||
| I don't have a good explanation for the difference, yet | 12:12 | ||
| The timings are clearly ignoring jvm startup time. | 12:14 | ||
| brrt | hi whiteknight | 12:50 | |
|
13:00
PacoAir_ joined
13:11
bluescreen joined
|
|||
| rurban | The jvm caches its compiled code on disk, so it's measuring warm-start. Novae re-did the bench and there java was slower. | 13:26 | |
| whiteknight: We missed you yesterday in #ps. Did you skim through the logs? | 13:51 | ||
| brrt | its not /that/ surprising given the amount of information java has that gcc doesn't | 13:56 | |
| and whats more | |||
| sbenchmarks like these don't measure the 'whole system' thing | 13:57 | ||
| where you may have many different objects / structures | |||
| many different algorithsm | |||
| and the calling conventions / memory layout / etc may make a huge difference | |||
| whiteknight | rurban: Busy with work yesterday. I didn't see the logs. Anything interesting? | 14:05 | |
| brrt: Yeah, but that code has lots of primitive floats and ints. Those kinds of things don't really benefit from type-specialization since the C code was using floats and ints too | 14:06 | ||
| brrt | hmm fair enough | ||
| still | |||
| it isn't /really/ surprising the jvm creates machine code with floats and ints that is as fast that gcc creates | 14:07 | ||
| which is what is happening | |||
| whiteknight | My best guess is that the timing code in the C version causes a slow syscall, while the Date class in the jvm can probably use a cached value and click counts | ||
| I don't know that for certain, but that's my best guess | 14:08 | ||
| of course, if you take that call out, there's no way to compare timings anymore | |||
| rurban: I sent kid51 a message last night. If all his testing looks good, I'll merge io_cleanup1 | 14:11 | ||
| if I get the all-clear, I can merge tonight | 14:12 | ||
| brrt | ... thats actually not unrealistic | 14:15 | |
| rurban | I am testing now nine's threads changes. io_cleanup1 smoked well. | 14:28 | |
| whiteknight | good | 14:35 | |
| I am looking forward to merging io_cleanup1, because as the name implies there's more work to do | 14:36 | ||
| but future branches will be much smaller | |||
| I may try to start 6model first, before I start trying to re-do some of the IO pmcs | |||
| yay! kid51 sent his approval to the list | 14:37 | ||
| time to merge | |||
|
14:38
alester joined
|
|||
| brrt | yay! | 14:39 | |
| alester | rurban: This is awesome. blogs.perl.org/users/rurban/2012/08...tizer.html | ||
| It's great to see all the nitty-gritty details. | |||
| rurban | yes. The new llvm based version is even fast, for huge projects. | 14:40 | |
| But the current version v1 is super-simple to use. | |||
| nine: threads passed ok all tests on solaris. smolder.parrot.org/app/projects/rep...ails/30613 | 14:46 | ||
| whiteknight | rurban: Where are threads still failing? Windows/cygwin? | ||
| rurban | cygwin yes. still a few deadlocks. currently in t/op/time.t in the same loop as ever. | 14:48 | |
| whiteknight | okay | ||
| At this point, I think threads won't merge before the release no matter what | |||
| A week is not a lot of time to get something so big integrated and tested | |||
| rurban | I agree. | ||
| io_cleanup1 should be enough. Did you talk with moritz? | 14:49 | ||
| whiteknight | I have not talked to moritz lately | ||
| rurban | there's one TODO multiseperator test in io_cleanup still failing | 14:50 | |
| whiteknight | in Rakudo? | ||
| rurban | t/pmc/filehandle.t 12 | ||
| moritz | if a rakudo spectest of current nom with the parrot-iocleanup1 branch merged with in passes, I'm fine with merging | ||
| whiteknight | moritz: I'll give that a try tonight | ||
| moritz | s:2nd/with// | 14:51 | |
| rurban | I thought you already implemented record_separator, multiple chars. | ||
|
14:52
davidfetter joined
|
|||
| rurban | Oh, the test is wrong. There is no fh open | 14:52 | |
| whiteknight | rurban: Oh that's fun | ||
| dalek | Heuristic branch merge: pushed 38 commits to parrot/whiteknight/io_cleanup1 by Whiteknight | 14:56 | |
| whiteknight | I'll run Rakudo specttest tonight on that branch, and merge immediately if all looks good | 14:57 | |
| and if my laptop freezes up halfway though, again, I'll say some cursewords and wait till tomorrow | 14:58 | ||
|
14:59
darbelo joined
|
|||
| rurban | I'll try also on my fast build machine. But I'm leaving in a few hours to Europe. | 15:00 | |
| whiteknight | oh fun, what are you doing in Europe? | ||
| rurban | BTW: I already fixed that wrong test t/pmc/filehandle_12.pir | 15:01 | |
| moritz | YAPC::EU maybe? :-) | ||
| whiteknight | oh, I didn't know YAPC:EU was soon | ||
| moritz | upcoming week | 15:02 | |
| Monday to Wednesday | |||
| rurban | p5p is having some meeting before in Perl, Germany | 15:04 | |
| dalek | rrot/whiteknight/io_cleanup1: 0fb668d | rurban++ | t/pmc/filehandle.t: fix t/pmc/filehandle_12.pir: open the filehandle multiple chars record_separator should work IMHO, but do not yet. |
15:08 | |
| moritz now spectests rakudo on io_cleanup1 | 15:09 | ||
|
15:09
dmalcolm joined
|
|||
| rurban | whiteknight: do you want a ticket for the failing t/pmc/filehandle_12.pir case? I thought it should already work. | 15:12 | |
| whiteknight | yes, please | 15:13 | |
| rurban | handle.pmc still uses the old io_readline_s method. That's the problem. | 15:18 | |
| moritz | rakudo fails all 16 socket tests on newest whiteknight/io_cleanup1 | 15:25 | |
| dalek | rrot/whiteknight/io_cleanup1: 34e880a | rurban++ | t/pmc/filehandle.t: [GH #812] Test to support io multi-char record_seperator |
||
| whiteknight | heh, great | 15:26 | |
| moritz | all the tests say "Got: ''" | ||
| whiteknight | okay, I'll fix it | ||
| rurban: io_readline_s is the new method | |||
|
15:26
dngor joined
|
|||
| rurban | Oh, I got it mixed up. | 15:27 | |
| Only filehandle::readline_interactive is missing the delim, and dynoplibs/io.ops | 15:32 | ||
| whiteknight | ah, okay | ||
| that method has slightly different semantics with the prompt and all | |||
| rurban | well, if someone changes the filehandle record_seperator, readline_interactive should use it I think | 15:33 | |
| whiteknight | okay, I can update that | 15:34 | |
| rurban | But there's special CRLF logic afterwards | 15:35 | |
| whiteknight | I think we've got to keep that method the same as it was | ||
| rurban | maybe leave it in asis. | ||
| agreed | |||
| whiteknight | yeah, that's what I'm thinking | ||
| rurban | pastebin.com/TsRXmmji | 15:38 | |
| whiteknight | ah, perfect | 15:39 | |
| rurban | wrong, there's no SELF in .ops | 15:40 | |
| whiteknight | almost perfect | ||
| s/SELF/$2/ | |||
| actually, we can probably move that logic into Parrot_io_readline | |||
| and keep the op body the same, for now | |||
| dalek | d_parrot/new-route: f93ec00 | (Bart Wiegmans)++ | / (4 files): Add test for mime-based loading |
15:51 | |
|
16:01
brrt left
|
|||
| rurban | But that test actually never calls Parrot_io_readline(), only _s() | 16:04 | |
| better check the failing socket tests | 16:06 | ||
| dalek | kudo/nom: e326148 | moritz++ | src/core/Match.pm: avoid codepoints above 255 in the setting reverts the factor ~2 parse slowdown of the setting introduced in 94aa2b7 |
16:08 | |
|
16:15
jashwanth joined
|
|||
| rurban | Parrot_io_readline pastebin.com/6z07rnDt | 16:19 | |
|
16:23
bluescreen joined
16:39
darbelo joined
|
|||
| rurban | oops, the GH #812 test was wrong. | 16:39 | |
| dalek | rrot/whiteknight/io_cleanup1: b694441 | rurban++ | t/pmc/filehandle.t: [GH #812] fix t/pmc/filehandle_12.pir |
16:41 | |
| whiteknight | so that test passes now? | 16:44 | |
| rurban | yes | ||
| forgot a seek | |||
| whiteknight | oh, okay | ||
| I love how some of these tests are wrong, and might have been wrong for years | 16:45 | ||
| rurban | I'm also adding the terminator attr to the old Parrot_io_readline() | ||
| Thats' the problem with skipped or TODO tests | |||
| whiteknight | okay | 16:50 | |
| readline is the worst thing ever | |||
| tadzik | :D | 16:51 | |
| what about it? | |||
| dalek | rrot/whiteknight/io_cleanup1: 8812d14 | rurban++ | src/io/api.c: [GH #812] use record_separator in fallback io.ops readline method also |
16:56 | |
| rurban | done, smoking now. | 16:57 | |
| have you found the socket errors in rakudo? | 16:58 | ||
| whiteknight | I can't look until I get home tonight | ||
| rurban | oki | ||
| I'll also have to leave soon to get my plane. See you in ~17 hrs. | 17:00 | ||
| whiteknight | see you then | 17:01 | |
| dalek | rrot/whiteknight/io_cleanup1: b1171e6 | rurban++ | config/gen/makefiles/root.in: [GH #812] Makefile add pmc/pmc_handle.h dependency to src/io/api.o [CAGE] fixes t/src/checkdepends.t |
17:07 | |
|
17:10
benabik joined
|
|||
| rurban | cygwin is now only failing t/op/time.t in threads smolder.parrot.org/app/projects/rep...ails/30614 | 17:31 | |
|
17:41
benabik_ joined
|
|||
| dalek | kudo/nom: b18ea07 | moritz++ | src/Perl6/ (2 files): sigilless variable declaration Currently only works for 'my' variables, and type constraints are ignored |
17:51 | |
|
18:24
lucian joined
|
|||
| dalek | kudo/nom: c076f5d | moritz++ | src/Perl6/Grammar.pm: fix copy&past-o from STD.pm6 |
18:41 | |
| kudo/nom: eb2f531 | moritz++ | src/Perl6/Actions.pm: fix a copy&past-o in method declare_param |
|||
| kudo/nom: dc19a70 | moritz++ | src/Perl6/Actions.pm: enable type checks for "my Str \\x = ..." term definitions, jnthn++ |
|||
|
19:29
lucian joined
19:32
Khisanth joined
20:01
Khisanth joined
|
|||
| dalek | kudo/nom: 30134cf | coke++ | t/fudgeandrun: add t/fudgeandrun from niecza++ simplifies testing of individual roast files. |
21:01 | |
|
23:28
kid51 joined
23:31
kid51_ joined
23:41
benabik joined
|
|||
| kid51_ wonders if this storm is what's causing my irc connection to continually drop | 23:42 | ||
| Hmm, we don't have aloha for msgs, do we? | 23:43 | ||
| benabik | aloha: aloha? | ||
| aloha | benabik: aloha is simple Bot::BasicBot::Pluggable hosted on github.com/bacek/aloha | ||
| benabik | aloha: ping | ||
| aloha: ping? | |||
| aloha | benabik: ping is to get someone's attention, pong is to indicate that the pingee is paying attention | ||
| kid51_ | msg whiteknight We are getting "uninitialized value" warnings in the 'make' output on io_cleanup1 branch. | 23:44 | |
| aloha | OK. I'll deliver the message. | ||
| dalek | kudo/nom: efde6e5 | tadzik++ | tools/build/Makefile.in: Don't forget to install precompiled Pod::To::Text |
23:56 | |