Camelia, the Perl 6 bug

IRC log for #parrot, 2012-09-20

Parrot | source cross referenced

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:01 whiteknight joined #parrot
00:02 whiteknight good evening, #parrot
00:02 sorear good evening whiteknight.
00:02 whiteknight hello sorear
00:49 dalek parrot/gh837_concat: b9fc29b | dukeleto++ | t/op/stringu.t:
00:49 dalek parrot/gh837_concat: [t] Move concat tests for #837 to a more appropriate file
00:49 dalek parrot/gh837_concat: review: https://github.com/parrot/parrot/commit/b9fc29b091
00:49 dalek parrot/gh837_concat: c552eb6 | dukeleto++ | t/pmc/string.t:
00:49 dalek parrot/gh837_concat: Revert "[t] Add some failing tests for #837 reported by pmichaud++"
00:49 dalek parrot/gh837_concat:
00:49 dalek parrot/gh837_concat: This reverts commit 5d7aa2d0d7489aadc3bc3037271d919634483b90.
00:49 dalek parrot/gh837_concat: review: https://github.com/parrot/parrot/commit/c552eb66d0
00:57 benabik joined #parrot
01:47 mdupont joined #parrot
01:49 l3l1p joined #parrot
01:54 dalek parrot/gh838_cmdline_cleanup: 19183d0 | dukeleto++ | / (2 files):
01:54 dalek parrot/gh838_cmdline_cleanup: Remove mentions of -d/-O/-v from --help output and docs/running.pod #838
01:54 dalek parrot/gh838_cmdline_cleanup: review: https://github.com/parrot/parrot/commit/19183d00be
02:01 MikeFair joined #parrot
02:05 dalek parrot/gh838_cmdline_cleanup: bf60df5 | dukeleto++ | / (3 files):
02:05 dalek parrot/gh838_cmdline_cleanup: Remove more references to options that no longer exist #838
02:05 dalek parrot/gh838_cmdline_cleanup: review: https://github.com/parrot/parrot/commit/bf60df5ce4
02:49 rurban_mobile joined #parrot
02:53 rurban_mobile checking the string encoding bug : it's 837 actually immc which ignores the encoding.
03:05 l3l1p joined #parrot
03:24 rurban_mobile A SymReg cannot hold the encoding, nor stores it the STRING, it just stores the STRING buffer.
03:28 rurban_mobile oh, it can with VT_ENCODED
05:01 l3l1p joined #parrot
06:00 dalek parrot/gh837_concat: 8a56a8e | rurban++ | / (3 files):
06:00 dalek parrot/gh837_concat: [GH #837] Honor encoding in imcc optimizer
06:00 dalek parrot/gh837_concat:
06:00 dalek parrot/gh837_concat: Context registers unfortunately do not store type information,
06:00 dalek parrot/gh837_concat: just the values. We would need at least type 'U" here.
06:00 dalek parrot/gh837_concat: So recreate an proper const'ed encoding string from the register
06:00 dalek parrot/gh837_concat: value.
06:00 dalek parrot/gh837_concat: review: https://github.com/parrot/parrot/commit/8a56a8e79c
06:03 rurban_mobile How we store and reencode our string registers is horribly inefficient. mk_const 'U' throws away the good STRING value and does rather pessimizing than optimizing.
06:04 rurban_mobile SREG's should just hold the STRING pointer, not the char*.
06:05 sorear what.
06:06 MikeFair I'm not sure I see the difference between a string pointer and a char* (and isn't the char* there for the C code that PIR is compiling into).
06:06 MikeFair ?
06:09 dalek parrot/gh837_concat: 3a6acd6 | rurban++ | ChangeLog:
06:09 dalek parrot/gh837_concat: [GH #873] ChangeLog line for imcc optimizer bug
06:09 dalek parrot/gh837_concat: review: https://github.com/parrot/parrot/commit/3a6acd6e9f
06:10 rurban_mobile STRING is more efficient and also holds the encoding ptr.
06:10 rurban_mobile char is basically just the raw information, but not the encoding.
06:10 rurban_mobile and you cannot store \0 for example.
06:12 rurban_mobile Now we escape and unescape a char* mult. times, because a register nor a const cannot hold STRING values.
06:12 rurban_mobile That's embarrassing
06:53 alvis_ joined #parrot
07:08 alvis joined #parrot
07:13 brrt joined #parrot
07:34 lucian joined #parrot
07:47 TonyC joined #parrot
08:20 Psyche^ joined #parrot
09:27 dalek rakudo/nom: 69910dc | masak++ | docs/release_guide.pod:
09:27 dalek rakudo/nom: [docs/release_guide.pod] I'll do September
09:27 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/69910dc761
09:31 dalek rakudo/nom: 94f900b | moritz++ | docs/release_guide.pod:
09:31 dalek rakudo/nom: fix a typo
09:31 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/94f900b3c8
11:33 brrt joined #parrot
11:40 brrt hi #parrot
11:57 JimmyZ joined #parrot
12:18 particle joined #parrot
13:04 PacoAir joined #parrot
13:15 elmex joined #parrot
13:18 PacoAir joined #parrot
14:12 Coke rurban: what are SREGs holding if not STRINGs?
14:12 Coke also, https://github.com/parrot/parrot/commit/​8a56a8e79c636325d646029ba01c16b36f71644 doesn't seem to exist.
14:14 Coke looks like it should have been 8a56a8e79c636325d646a029ba01c16b36f71644
14:25 benabik joined #parrot
14:37 l3l1p joined #parrot
14:51 benabik_ joined #parrot
15:00 dmalcolm joined #parrot
15:00 benabik_ joined #parrot
15:40 estrabd joined #parrot
15:41 estrabd joined #parrot
16:26 alvis_ joined #parrot
17:14 tuxit joined #parrot
17:19 rurban Coke: I'll mail my concerns and ideas to the list. It will be bigger. SREG only holds "encoding:\"encoded escaped string\"" char *
17:58 benabik joined #parrot
18:02 zby_home joined #parrot
18:07 contingencyplan joined #parrot
18:31 dalek parrot/gh837_concat: e8a0ab4 | rurban++ | examples/benchmarks/stress_stringsu.pir:
18:31 dalek parrot/gh837_concat: [GH #837] Add benchmark to test encoding slowdown
18:31 dalek parrot/gh837_concat:
18:31 dalek parrot/gh837_concat: stress_stringsu.pir tests encoding via sprintf and concat.
18:31 dalek parrot/gh837_concat: The fixed version is only a bit slower then the old broken version.
18:31 dalek parrot/gh837_concat: 13.666s (broken) vs 13.506s (fixed) vs 3.022s (without encoding).
18:31 dalek parrot/gh837_concat:
18:31 dalek parrot/gh837_concat: So fixing the encoding representation as STRING* in SREG and consts 'U'
18:31 dalek parrot/gh837_concat: is a worthwile goal.
18:31 dalek parrot/gh837_concat: review: https://github.com/parrot/parrot/commit/e8a0ab4483
18:33 Coke rurban: is SREG the same thing as a $S1 ?
18:33 Coke or is it some internal IMCC thing?
18:45 lucian joined #parrot
18:54 dalek parrot/gh837_concat: 7ecd472 | rurban++ | examples/benchmarks/stress_strings1.pir:
18:54 dalek parrot/gh837_concat: [GH #837] New comparison benchmark against stress_stringsu.pir
18:54 dalek parrot/gh837_concat:
18:54 dalek parrot/gh837_concat: Bad news:
18:54 dalek parrot/gh837_concat: There is almost no measurable speed advantage from ascii strings to encoded strings.
18:54 dalek parrot/gh837_concat: With the fix and without the fix, as sprintf handling slows down the test 4x times.
18:54 dalek parrot/gh837_concat:
18:54 dalek parrot/gh837_concat: So converting a SREG and const to encoded STRING* will not gain much performance.
18:54 dalek parrot/gh837_concat: review: https://github.com/parrot/parrot/commit/7ecd47233a
19:05 dalek rakudo/nom: c13ac5e | masak++ | docs/announce/2012.09:
19:05 dalek rakudo/nom: [docs/announce/2012.09] added
19:05 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/c13ac5e535
19:13 dalek rakudo/nom: 738d1ea | masak++ | docs/announce/2012.09:
19:13 dalek rakudo/nom: [docs/announce/2012.09] mention faster compiler
19:13 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/738d1ea823
19:18 dalek rakudo/nom: 488521b | masak++ | docs/announce/2012.09:
19:18 dalek rakudo/nom: [docs/announce/2012.09] tie-breaking is a feature
19:18 dalek rakudo/nom:
19:18 dalek rakudo/nom: Not a breaking change. :)
19:18 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/488521be0c
19:19 PacoAir joined #parrot
19:21 PacoAir joined #parrot
19:22 dalek nqp: a318bd5 | masak++ | VERSION:
19:22 dalek nqp: [VERSION] bump to 2012.09
19:22 dalek nqp: review: https://github.com/perl6/nqp/commit/a318bd55e4
19:32 PacoAir joined #parrot
19:39 PacoAir joined #parrot
19:43 dalek rakudo/nom: bdb50f4 | masak++ | tools/build/NQP_REVISION:
19:43 dalek rakudo/nom: [release] bump NQP revision
19:43 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/bdb50f47a6
19:43 dalek rakudo/nom: 77defe7 | masak++ | VERSION:
19:43 dalek rakudo/nom: [release] bump VERSION
19:43 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/77defe7332
19:51 dalek nqp: e8a7719 | jonathan++ | docs/qast.markdown:
19:51 dalek nqp: Document more of QAST.
19:51 dalek nqp: review: https://github.com/perl6/nqp/commit/e8a7719544
19:52 sivoais joined #parrot
20:16 dalek rakudo/nom: adc2127 | masak++ | docs/release_guide.pod:
20:16 dalek rakudo/nom: [docs/release_guide.pod] some future release dates
20:16 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/adc21274b3
20:18 dalek rakudo/nom: ee7dfef | duff++ | docs/release_guide.pod:
20:18 dalek rakudo/nom: Move masak's release from the future to the past
20:18 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/ee7dfef791
21:10 sivoais joined #parrot
21:25 dalek rakudo/nom: 00533a4 | duff++ | CREDITS:
21:25 dalek rakudo/nom: Add IRC name to CREDITS
21:25 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/00533a4f3d
21:25 dalek rakudo/nom: ef228dd | duff++ | tools/contributors.pl:
21:25 dalek rakudo/nom: Minor improvement to contributors.pl
21:25 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/ef228dd818
22:14 benabik joined #parrot
22:16 benabik_ joined #parrot
22:21 benabik_ joined #parrot
22:24 chekkaa oh, benabik
22:24 chekkaa I ironed out all the display bugs I could find on the new drupal 6 site
22:24 chekkaa http://www.parrot.org:8082/
22:24 chekkaa let me know if you can find anything else that needs fixing
22:24 chekkaa if there is nothing else, we can push it to production soon
22:25 benabik chekkaa: the list in the sidebar on http://www.parrot.org:8082/resources.html looks odd.  (Not the user sidebar, but I think one for the page.)
22:26 chekkaa I could have sworn I fixed that
22:26 benabik Same on /foundation.
22:26 chekkaa perhaps it's a caching issue
22:26 chekkaa let me look at it
22:27 benabik chekkaa: Other than that, it looks good to me.  I'm not familiar with every corner of the site though.
22:27 chekkaa you're the only one who responded to me though :)
22:28 benabik I lurk a lot and you keep bringing it up when I have a moment of free time.
22:39 chekkaa benabik: should be fixed
22:39 benabik chekkaa++
22:39 benabik (Meaning yes)
22:43 chekkaa all right, so then when you guys are ready, I can update the test site's database to catch some of the recent activity from the last few days and then start pointing the main site at it
22:45 benabik Will post that to -dev
22:48 benabik Sent
22:50 chekkaa thanks
23:21 ilbot2 joined #parrot
23:21 Topic for #parrot is now Parrot 4.8.0 "Spix's Macaw" | http://parrot.org/ | Log: http://irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
23:23 Psyche^ joined #parrot
23:27 rurban chekkaa++  I like it.
23:31 rurban Coke:  SREG is the location of an $S0, see src/call/context.c. The S at the end. REG_STR() accesses them. SREG is REG_STR(interp, cur_opcode[i])
23:32 rurban imcc handles its own copy of registers in memory, the SymReg*
23:33 rurban A SymReg has a type 'U' for example. the SREG not. imcc is much better on representing and optimizing strings than the parrot vm.
23:49 whiteknight joined #parrot

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

Parrot | source cross referenced