Parrot 2.8.0 released | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | smoke GC-related branches and attack GC tickets
Set by moderator on 1 October 2010.
00:13 TonyC left
cotto ascii art incoming 00:21
Is dmitry here (the one who just posted to parrot-dev)? 00:22
00:25 chromatic left, theory left
dalek rrot: r49459 | cotto++ | trunk/src/runcore/profiling.c:
[profiling] attempt to clarify how timing information for inner runloops is handled
 00:27
kid51 Darwin/PPC: r49456 trunk once again PASS when using g++ for --link and --ld; see this Smolder: smolder.parrot.org/app/projects/rep...etails/459 00:28
00:28 kid51 is now known as kid51_at_dinner
cotto plobsing, does r49459 make the code easier to understand? 00:44
or anyone else who cares to read the diff 00:46
plobsing cotto: yes, it makes more sense now. 00:53
I didn't need the ascii art so much as the explanatory static functions 00:54
cotto now, on to the heisenbug 00:59
01:03 nopaste joined 01:04 TonyC joined 01:06 hercynium joined 01:07 ruoso joined
plobsing heisenbug eh? so do you know where the bug is or how long it'll take to fix? 01:08
cotto Um... I *wish* I knew one of those things. 01:10
01:12 bluescreen joined 01:28 theory joined
dalek rrot: r49460 | jkeenan++ | branches/tt1810_missing_step_tests (2 files):
Introduce and test get_steps_missing_tests().
 01:44
kid51_at_dinner r49459 cotto++ 01:45
01:45 kid51_at_dinner is now known as kid51
cotto What'd I do? ascii art? 01:49
02:00 kid51 left 02:12 kid51 joined
dalek rrot: r49461 | jkeenan++ | branches/tt1810_missing_step_tests/lib/Parrot/Configure/Options/Test.pm:
Call get_steps_missing_tests() before postconfiguration tests.
 02:14
02:35 janus left 02:41 janus joined 02:54 Drossel is now known as Kulag 03:05 kid51 left 03:09 theory left 03:20 mikehh left 03:29 ruoso left 04:03 ash_ left 04:07 theory joined 04:16 jsut_ joined 04:21 jsut left
cotto seen khairul 04:22
aloha Sorry, I haven't seen khairul.
cotto Sorry, I haven't seen khairul either. 04:30
bacek_at_work 44 days ago according to purl 04:49
05:16 sjn left 05:17 sjn joined 05:53 cottoo joined, cotto left 05:54 cottoo left, cottoo joined, fperrad joined
cottoo xine-- 05:55
06:31 fperrad left 06:38 fperrad joined 06:54 cottoo is now known as cotto 07:03 theory left 07:13 jsut joined 07:18 jsut_ left 07:30 bacek left
cotto clock? 07:31
allison, clock?
aloha, clock?
aloha cotto: cotto: LAX: Wed, 00:20 PDT / CHI: Wed, 02:20 CDT / NYC: Wed, 03:20 EDT / UTC: Wed, 07:20 UTC / LON: Wed, 08:20 BST / BER: Wed, 09:20 CEST / TOK: Wed, 16:20 JST / SYD: Wed, 18:20 EST
cotto: Sorry, I don't know.
07:35 he joined 07:36 M_o_C joined 08:01 bacek joined
dukeleto bacek: hola 08:01
bacek aloha, dukeleto
dukeleto bacek: how goes it? what are you workin' on today? 08:02
bacek dukeleto, $dayjob or parrot?
cotto hio bacek 08:10
bacek hi cotto 08:11
08:16 allison left
dukeleto bacek: parroty bits 08:18
dukeleto is about to pass out
bacek dukeleto, gen_gc
horribly broken atm
dukeleto bacek: let me know if you need branches smoked
bacek dukeleto, not yet. It's few weeks work before I can get gen_gc into proper shape. 08:19
cotto only for bacekbot is fixing gc bugs a reasonably deterministic process 08:20
08:27 M_o_C left 08:29 M_o_C joined
dalek rrot: r49462 | cotto++ | trunk/src/runcore/profiling.c:
[profiling] move version and cli output code into a separate function
 08:38
08:44 PerlPilot joined 08:48 szabgab left, PerlJam left 09:25 M_o_C left 10:08 bacek left 10:09 bacek joined 11:06 bluescreen left 11:11 Patterner left, contingencyplan left 11:58 Patterner joined 12:21 bluescreen joined 13:20 Patterner left 13:30 Psyche^ joined, Psyche^ is now known as Patterner 14:04 particle1 left 14:05 bkuhn` joined 14:10 PacoLinux joined 14:11 particle joined 14:14 bkuhn` left 14:23 hatseflats left 14:24 Util left 14:28 hatseflats joined 14:29 Util joined 14:41 Util left, hatseflats left, Util joined 14:42 hatseflats joined 14:49 hercynium left 15:03 ruoso joined 15:06 tadzik joined 15:10 pmichaud joined 15:11 theory joined
dukeleto 'ello 15:27
15:38 PerlPilot is now known as PerlJam 15:41 tadzik left 15:49 ash_ joined 15:52 tadzik joined 16:24 theory left
ash_ what is needed for BigInt to work on ubuntu? 16:26
moritz would guess libgpm-dev
ash_ i installed libgmp3-dev but when it runs the test it dies on new ["BigInt"] 16:27
it dies saying "no bigint lib loaded" 16:28
weird, the BigInt.pmc has that hard coded if it doesn't have the bigint library at configure time, but it says auto::gmp......yes. when i did config, i'll re-check it out, since i just installed libgmp3-dev 16:32
ya, that fixed it 16:37
16:42 mikehh joined
dukeleto ash_: yeah, you need to reconfig if you install gmp after a build 16:49
ash_: do you need any more testing on your branch? I have a darwin ppc box that I can smoke it from, if you need 16:50
ash_: as well as 32bit and 64bit ubuntu's
ash_ dukeleto: let me try to work through the amd64 linux bug right now
i have a 64bit linux vm i setup today to have easier access at tracking down the issue with my nci stuff 16:51
dukeleto ash_: ok, good luck and let us know if you need help. A lot of us are really excited about merging that branch.
dalek rrot: r49463 | mikehh++ | trunk/src/runcore/profiling.c:
fix codetest failures - linelength and c_function_docs
dukeleto ash_: awesome. have the appropriate amount of fun.
ash_ ya, sorry its been so long :-( been busy with school and midterms are this week
16:54 ruoso left
ash_ dukeleto: i just re-made parrot with clang-2.8 (clang-2.8 was released like 2 days ago) instead of gcc-4.4 on linux and i do not get the error when using clang. i hate these kind of bugs :-( 17:03
dukeleto ash_: yes, indeed. Compiler variations sure do make our jobs more fun ;) 17:05
ash_ os x uses gcc 4.2, i am downloading gcc 4.3 to see what kind of error it makes :P 17:06
gcc-4.3 gives the same error as gcc-4.4 17:12
which makes sense, clang is modeled after gcc-4.2 and gcc-4.2 doesn't give the errors on OS X so i see no reason they would show up with any other compiler 17:13
ash_ checking gcc-4.4 on OS X, it probably gives the same error as linux, which makes testing easier because no vm is required...
moritz it seems that r49406 broke a rakudo test 17:14
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#480) fulltest) at r49463 - Ubuntu 10.10 RC amd64 (g++-4.5) 17:18
ash_ hmmm weird, gcc-4.4 doesn't give the error on OS X, now that is surprising 17:21
17:21 fperrad left
ash_ looks like i am going to have to do some digging 17:22
17:24 NotFound left 17:38 contingencyplan joined
dukeleto ash_: what is the nature of the bug? what is different across the compilers and versions you have tested? 17:47
ash_ its a 'double free or corruption' bug, its from freeing something twice or from freeing something thats not malloced 17:48
17:48 Andy joined
ash_ it does not appear on OS X, tested with gcc-4.2(apples), clang 2.7, clang 2.8, and gcc-4.4 (macports), it also does not appear to happen on 32-bit linux, but 64-bit linux it does appear when using gcc-4.3 or gcc-4.4 17:49
17:49 plobsing_ joined
ash_ on linux-64 with clang2.8 it does not happen 17:50
plobsing_ moritz: (re: r49406 breaking rakudo test) does rakudo somehow depend on the trace core? how? 17:51
ash_ i think i narrowed it down some 17:52
moritz plobsing_: it seems my bisecting was wrong; it's a different revision that broke it
plobsing_: I'm doing a second run now
plobsing_ OK, seemed fishy
moritz indeed 17:53
17:55 plobsing_ left
moritz fears his bisecting might be inconclusive 17:56
the second run identified r49459, which only changes comments 18:02
18:25 masak joined 18:30 davidfetter joined 18:40 fperrad joined
cotto moritz, could it be a bug dependent on hash ordering? 18:40
18:47 fperrad left
ash_ i think i fixed the bug on amd64 in my branch doing a bit more testing to be sure 18:49
dukeleto: alright, i got that fixed and now my branch only fails in 2 places, packfileannotations.t and packfileconstanttable.t for some reason both stop in the middle of the tests 18:58
dukeleto ash_: those tests may need new pbc generated to pass 19:01
ash_ ah
how do you do that?
dukeleto ash_: make pbctestfiles or run tools/dev/mk_packfile_pbc 19:02
ash_: not sure which is better
ash_ trying that 19:03
make pbctestfiles didn't do anything so i am doing make pbctestfiles -B
dukeleto: no good, still dies with: too few positional arguments: 2 passed, 3 (or more) expected 19:08
current instr.: 'test_pack_unpack' pc 326 (t/pmc/testlib/packfile_common.pir:67)
19:09 bluescreen left 19:10 bluescreen joined
ash_ anyone know about the PackfileConstantTable PMC ? 19:35
or PackfileAnnotations PMC ?
cotto what about them/ 19:37
?
ash_ in my gsoc_nci branch i now only fail tests for both of those
gist.github.com/613936 is what happens when i run one of the tests with --trace=2 19:38
i got a "too few positional arguments: 2 passed, 3 (or more) expected" error
for some reason
cotto Is there anything special I need to play with that branch? 19:41
ash_ libffi headers, thats all
wait, let me push my changes 19:42
so your working with the same things i am if you check it out
okay, i pushed my changes, so if you checkout the branch it should be good to go 19:45
dukeleto ash_: do you want me to smoke the branch? 19:51
ash_: can you paste the full output of the failing tests, or do they hang?
ash_ the output of prove t/pmc/packfileannotations.t + packfileconstants.t?
or the output of ./parrot /path 19:52
i am seeing the same errors on OS X and 64 bit linux
cotto me too
ash_ so i would expect the same results on other systems
cotto which is confusing since that method appears to take only two arguments (including self) 19:53
ash_ ya, i noticed that
i thought it might be because it was multi, but other multi's are working
so i am not sure what exactly is the issue
i am trying to merge with the trunk to see if that fixes it 19:54
dukeleto ash_: worth a shot
dalek rrot: r49464 | ash++ | branches/gsoc_nci (23 files):
Merge branch 'master' into gsoc_nci
 19:56
rrot: r49465 | ash++ | branches/gsoc_nci/config/gen/libffi/nci-ffi.pmc.in:
Fixing a bug for amd64 linux where it tried to free corrupted memory.
ash_ merging with trunk has the same result 19:57
TimToady phone 20:00
dukeleto ash_: is the problem happening on all platforms or just a few/ 20:02
ash_: asking people to smoke the branch on parrot-dev might get some extra eyes on it 20:03
ash_ its happening the exact same on OS X and linux for me, those are all i have tried
dukeleto ash_: and you are getting a hang in those tests?
ash_ no, parrot throws an error say: too few positional arguments: 2 passed, 3 (or more) expected
dukeleto ash_: what is the full error? 20:04
ash_: which function is complaining about that?
GeJ Bonjour everyone.
Otitises(?) suck, just so you know. 20:05
ash_ PackfileAnnotation set_name (which takes self and 1 STRING), and PackfileConstantTable get_or_create_constant which is a multi and takes self + 1 param for all versions of the multi
ash_ attempting to build rakudo with my branch to see if it has any problems... 20:13
20:19 cognominal joined
ash_ passes rakudo's make test, trying spectest 20:22
20:24 tadzik left 20:33 Andy left 20:35 bluescreen left, bluescreen joined 20:38 aloha left, bacek_at_work left 20:46 bacek_at_work joined 20:47 aloha joined 20:50 masak left
cotto At what point is it more efficient to use StringBuilder rather than concatenating STRINGs? 21:04
dalek rrot: r49466 | cotto++ | trunk/src/runcore/profiling.c:
[profiling] make get_ns_cstr a bit less convoluted
 21:13
21:29 bluescreen left
cotto nm. For my purpose I'd need to add unshift, and I'm not frustrated enough with my heisenbug to get distracted and do that yet. 21:36
dalek rrot: r49467 | cotto++ | trunk/src/runcore/profiling.c:
[profiling] minor cleanup
 21:44
22:35 szabgab joined 22:37 ruoso joined 22:56 Andy joined 23:08 whiteknight joined
whiteknight good evening, #parrot 23:13
23:14 kid51 joined 23:37 dngor left 23:38 dngor joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!