| [Coke] | 182 test files aborted. | 02:07 | |
| started tracking failures here: gist.github.com/coke/8250608 | 02:16 | ||
| working on breaking that up into more useful bits. | |||
|
02:38
benabik joined
03:10
ssutch joined
04:00
brother joined
04:11
jnap joined
05:11
jnap joined
07:13
jnap joined
|
|||
| [Coke] | gist.github.com/coke/8250608 - tried to break it down by error, showing the # of aborted tests before the error, and the test files it impacts after. | 07:40 | |
| not quite a standalone script, but should help. | |||
| biggest caused of aborted tests is : [1952] Cannot call method 'dispatch:<.^>' on a null object | 07:47 | ||
| happens in, e.g. S02-magicals/args.t | |||
|
07:59
colomon joined
08:14
jnap joined
09:29
Woodi joined,
ssutch joined,
benabik joined,
JimmyZ joined,
dagurval joined,
woolfy joined,
jnthn joined,
nwc10 joined,
eternaleye joined,
cognominal joined,
masak joined,
dalek joined,
retupmoca joined,
japhb_ joined,
sorear joined,
rurban joined,
timotimo joined,
ingy joined,
wsri joined,
lizmat joined,
lee__ joined,
camelia joined
09:30
_sri joined,
flussence joined,
TimToady joined,
chipdude joined,
moritz joined,
rblackwe joined,
cxreg joined,
brother joined,
lue joined,
harrow joined,
FROGGS joined,
BinGOs joined
09:35
arnsholt joined
09:44
[Coke] joined,
tadzik joined
09:53
ChanServ joined,
rblackwe joined,
colomon joined,
diakopter joined
09:54
rblackwe joined
10:15
jnap joined
11:16
jnap joined
12:08
colomon joined
12:17
jnap joined
13:18
jnap joined
14:18
jnap joined
14:29
ChanServ joined
14:33
jnap joined
|
|||
| nwc10 | jnap: the torture code was itself buggy (not actually torturing) | 15:46 | |
| I believe that the bugs are now fixed, and NQP is being slowly tortured | |||
| and Rakudo will follow | |||
| FROGGS | nwc10: can you keep an eye on the uv__delete assertions? (i.e. if they still appear or not) | 15:51 | |
| nwc10 | had 5 in testing, but not sure what revision. I think last night's HEADs (with MoarVM stuff of mine added) | ||
| both crap data in handle | 15:52 | ||
| and assertion failures | |||
| FROGGS | 67d6ba5 in moarvm fixes it for me... | ||
|
15:53
dalek joined
16:01
jnap joined
|
|||
| jnap | nwc10: although your messages was probably not aimed at me, your use of the word torture has caused me to think deeply again about the types of words we use in IT, and how the possible reveal interesting things about our culture | 16:04 | |
| [Coke] | FROGGS: there were a bunch in yesterday's run. | ||
| er, 2, which caused a bunch of failures: S05-metasyntax/angle-brackets.rakudo.moar and S06-signature/outside-subroutine.t | 16:05 | ||
| FROGGS | [Coke]: my patch is only 16.5 hours old... | ||
| [Coke] | if those 2 work, that's all the (current) cases from roast. | ||
| FROGGS | but surely there are other segfaults which need to be tackled | ||
| timotimo | jnap: agreed. | 16:15 | |
| [Coke] | FROGGS: 1791 individual tests aborted due to segfaults, yes. | ||
| see gist.github.com/coke/8250608 for a list of test files. (//Segmentation fault) | 16:16 | ||
| FROGGS | [Coke]: yes, I want to look into some of these this night | 16:17 | |
| [Coke] | masak: (email) *facepalm* d'oh. | 16:20 | |
| ww | |||
| ah, the unrecognized character name is split up into chunks, but I bet fixing one will fix 'em all. | 20:59 | ||
| FROGGS | ahh, shell() is not working... ym favourite :o) | 21:18 | |
| my* | 21:19 | ||
| timotimo | on moarvm? | ||
| FROGGS | yes | 21:22 | |
| timotimo | get going, then? :D | ||
| FROGGS | ey sir! /o | ||
| timotimo | o> | 21:23 | |
| FROGGS .oO( how do you type the rotated v? ) | |||
| timotimo | the same way you can turn a q into a p, b or d | 21:24 | |
| FROGGS | wow | ||
| :D | |||
| I am reading to much webfails obviously... | 21:25 | ||
| timotimo | it's all good so long as you don't fail yourself :) | ||
| FROGGS | no facebook no fail :o) | ||
| timotimo | :D | 21:27 | |
| should the repr for MVMSharedArray have locking built into its stuff? | 21:31 | ||
| well, at least for when the size needs to change | 21:32 | ||
| diakopter | what is an MVMSharedArray | 21:35 | |
| timotimo | i'm interested in making shmget and friends available to nqp and rakudo on moarvm | ||
| to experiment with representational polymorphy | |||
| diakopter | what is shmget | 21:36 | |
| timotimo | shm_get | ||
| diakopter | what is shm_get | ||
| timotimo | shared memory stuff on linux, system5 ipc | ||
| diakopter | that's for cross-process sharing, I thought | ||
| timotimo | yes, it is | 21:37 | |
| diakopter | (what would you use that for?) | ||
| timotimo | sharing data across processes | ||
| diakopter | I didn't see that discussed here yet | 21:38 | |
| timotimo | yeah. i didn't discuss it with anybody yet | 21:40 | |
| diakopter | what did you want to store in this array? you can't store pointers | ||
| timotimo | it'd have to be a native int/num array | ||
| diakopter | what nqp/perl6 type would use this repr | ||
| timotimo | i was thinking Buf | ||
| diakopter | h | 21:41 | |
| hm | |||
| timotimo: the QAST::WVal.new( :value(nqp::getcodeobj(&p6sort)) ) trick is really powerful, isn't it? :) | 21:44 | ||
| timotimo | i agree | 21:46 | |
| i was amazed myself | |||
| diakopter | I was thinking the other day this is another way to implement macros | 21:47 | |
| (make them actually not "interpolate" and compile until first run) | 21:50 | ||
| FROGGS | that sounds like an indirect lookup, and therefor slow | 21:53 | |
| timotimo | can we do "only the first time it's run"? | 21:54 | |
| diakopter | FROGGS: how would it be slower than compiling them up front? | 21:55 | |
| FROGGS | I dunno, I guess I said rubbish now that I think about it.. | 21:57 | |
| diakopter | I don't doubt it would be slower by some amount due to optimizer overhead, maybe | 21:58 | |
| dalek | arVM: 7cea05b | (Tobias Leich)++ | src/io/procops.c: decont the values of the env hash |
23:03 | |
| diakopter | FROGGS: I think you need some more MVMROOT in that patch's region | 23:11 | |
| FROGGS | diakopter: where exactly? | ||
| diakopter | FROGGS: well, also the declaration of r needs to be at the top of the block ;) | 23:17 | |
| FROGGS | ohh, yes | ||
| diakopter | but you also need to root env_str | 23:18 | |
| since the order of calling MVM_repr_get_str is not guaranteed | |||
| FROGGS | k | ||
| diakopter | (in the args to MVM_string_concatenate) | ||
| FROGGS | yes, I see it | 23:19 | |
| diakopter | well, also that lines 116 and 117 might allocate conceivably | ||
| er 117 | |||
| well, you actually should root iterval and &r.o for the same reason | 23:22 | ||
| but if you root &r.o you need to make sure to NULL it on initialization | 23:23 | ||
| stage parse: 92s | 23:32 | ||
| is that fast? | |||
| FROGGS | slightly faster than my box | 23:33 | |
| diakopter | if I try to fix one of the unicode things, will I need to rebuild the setting to test it? | ||
| perl -MExtUtils::Command -e mkpath C:\Users\mwilson\src\install/languages/perl6/lib/Pod/To | 23:34 | ||
| mkdir C:\Users\mwilson\src\install/languages/perl6/lib: File exists; Cannot create a file when that file already exists at C:/Perl64/lib/ExtUtils/Command.pm lin | |||
| FROGGS | no, usually not | ||
| diakopter | e 288. | ||
| NMAKE : fatal error U1077: 'C:\Perl64\bin\perl.EXE' : return code '0x11' | |||
| is mkpath supposed to care about that? | 23:35 | ||
| FROGGS | not very nice that this is fatal | ||
| IIRC that behaves weird on windows only | |||
| diakopter | === perl6-m -e "say('hi')" === | ||
| hi | |||
| Execution time: 0.677 s | |||
| not horrible! | 23:36 | ||
| FROGGS | :o) | ||
| diakopter | sortof. | ||
| how to parallel nmake test on windows | |||
| er, spectest | 23:37 | ||
| FROGGS | TEST_JOBS=6 | 23:39 | |
| diakopter | well, my windows vm thinks it has 8 cores.. | ||
| :) | |||
| FROGGS | feel free to adjust it :o) | ||
| diakopter | NMAKE : fatal error U1058: teNrminMaAtKEe d: b yf autsaelr | 23:40 | |
| eSrtroorp .U | |||
| hee hee | |||
| FROGGS | yeah, I've see that before | ||
| dalek | arVM: 428cd37 | (Tobias Leich)++ | src/io/procops.c: root things that might move, diakopter++ |
23:43 | |
| diakopter | oh heh. | 23:44 | |
| perl -MExtUtils::Command -e cp lib/Test.moarvm C:\Users\mwilson\src\install/languages/perl6/lib | |||
| that creates the file lib | |||
| since the dir didn't exist beforehand | |||
| FROGGS | umm | 23:45 | |
| now I understand that error message :o) | |||
| add a slash to ...perl6/lib? | |||
| diakopter | can you fix the rakudo makefile.in? | ||
| I don't have a rakudo push flag | |||
| FROGGS | I can | ||
| diakopter | I'd add the mkpath lib thing | 23:46 | |
| well, or move the mkpath of the Pod one up | |||
| FROGGS | I'd just add that at line 312: $(MKPATH) $(PERL6_LANG_DIR)/lib | 23:47 | |
| diakopter | k | ||
| FROGGS | pushed | 23:48 | |
| diakopter | ty :) | 23:49 | |
| heh link /dll /nologo /LTCG /nologo /LTCG | 23:51 | ||
| FROGGS | yeah, we really want to set these options | ||
| diakopter | same with cl /c /nologo /MT /Ox /GL /DWIN32 /DAO_ASSUME_WINDOWS98 /nologo /MT /Ox /GL | 23:52 | |
| I hit this last time: | 23:55 | ||
| perl -MExtUtils::Command -e chmod 755 C:\Users\mwilson\src\install/bin/perl6-m | |||
| annot chmod 755 C:\Users\mwilson\src\install/bin/perl6-m:No such file or directory at C:/Perl64/lib/ExtUtils/Command.pm line 274. | |||
| no need to do that on windows anyway, even if it was the right filename | 23:56 | ||
| [Coke] | oh, here's a good one in the spec tests: Internal error: invalid thread ID in GC work pass | ||
| FROGGS | /o\ | ||
| diakopter | [Coke]: that means GC bug (there shoudl be only 1 threadid, 0) | ||
| but probably just the same old non-rooting | 23:57 | ||
| not a problem with the gc itself, per se | |||
| can someone fix that ^ with the rakudo makefile install? | 23:58 | ||
| also, the line above it is wrong perl -MExtUtils::Command -e cp perl6-m C:\Users\mwilson\src\install/bin | |||
| that should have a .bat I thought | 23:59 | ||