[Tux] This is Rakudo version 2016.09-142-g605f272 built on MoarVM version 2016.09-27-g0bdbd6e 07:42
csv-ip5xs 3.100
test 16.805
test-t 7.370
csv-parser 18.366
lizmat Files=1146, Tests=53274, 234 wallclock secs (14.15 usr 3.94 sys + 1454.00 cusr 134.71 csys = 1606.80 CPU) 09:35
dalek ast: 03380fe | lizmat++ | S06-signature/types.t:
Make test less specific to accommodate more specificity
09:51
ast/6.c-errata: 9dd5444 | lizmat++ | S06-signature/types.t:
Make test less specific to accommodate more specificity
09:53
nine lizmat: I just love that commit message :) 10:30
I now have a fully automated script for creating updated nqp RPMs on the Open Build Service including writing of the changelog :) Now only automation for rakudo itself is missing 11:57
Soon we'll have fresh openSUSE packages on every release day :)
p3rln00b NeuralAnomaly: stats 12:03
NeuralAnomaly p3rln00b, [āœ˜] Next release will be in 1 day and 1 week. Since last release, there are 36 new still-open tickets (33 unreviewed and 0 blockers) and 142 unreviewed commits. See perl6.fail/release/stats for details
nine Rakudo's ChangeLog will be harder as that's just too much to put into the .spec file :/ Maybe I should just write that it's a new version 12:04
p3rln00b Files=1194, Tests=129670, 133 wallclock secs (20.40 usr 2.73 sys + 2306.51 cusr 220.83 csys = 2550.47 CPU) 14:55
(guess it doesn't hurt for me to post these once in a while, so we get some sort of historical benchmark going in the logs) 14:56
nine lizmat++ does that, too
p3rln00b ZOFVM: Files=1194, Tests=129670, 133 wallclock secs (20.40 usr 2.73 sys + 2306.51 cusr 220.83 csys = 2550.47 CPU) 15:00
(I just realized, it may be hard to grep for them :P, so I'll be using a prefix)
m: sub { 42.return }() 15:17
camelia rakudo-moar 605f27: OUTPUTĀ«Attempt to return outside of any Routineā¤ in sub at <tmp> line 1ā¤ in block <unit> at <tmp> line 1ā¤ā¤Ā»
p3rln00b That's broken innit...
jnthn Looks like 16:10
timotimo it looks for a return in the wrong position? 16:19
p3rln00b m: say (-125)**(1/3) 16:57
camelia rakudo-moar 605f27: OUTPUTĀ«NaNā¤Ā»
p3rln00b m: say (-125+0i)**(1/3)
camelia rakudo-moar 605f27: OUTPUTĀ«2.5+4.33012701892219iā¤Ā»
p3rln00b :S
m: say (-1+0i) ** .5 16:58
camelia rakudo-moar 605f27: OUTPUTĀ«6.12323399573677e-17+1iā¤Ā»
p3rln00b For the curious, that was rt.perl.org/Ticket/Display.html?id...xn-1430565 17:22
p3rln00b puts away his pretend Math Degree back in the drawer.
timotimo do other systems try to give 0+1i for that by finding out if it's that special case? 17:23
p3rln00b Why is that a special case? 17:24
Oh the first one you mean
Perl 5 gives NaN 17:25
timotimo i meant getting the sqrt of -1+0i
giving NaN for -125 is correct. we only give you complex out if you put complex in
p3rln00b m: say sqrt(-1+0i)
camelia rakudo-moar 605f27: OUTPUTĀ«0+1iā¤Ā»
p3rln00b timotimo: but it's not a complex. The result is -5 17:26
heh, python gives (2.5+4.330127018922192j) for it :} 17:28
timotimo oh! 17:29
it's the third root
p3rln00b Yeah
timotimo why didn't i look closely enough ...
dogbert17 o/ 17:32
p3rln00b \o 17:33
dogbert17 is everyone valgrinding?
p3rln00b is working... (well supposed to)
Speaking of which. Perhaps I should do a couple of hours of honest labour this week..
p3rln00b &
:)
dogbert17 :) 17:34
when running valgrind on some perl6 code the Leak Summary often looks like below. Are those the expected numbers? 17:35
==29364== LEAK SUMMARY: definitely lost: 5,454 bytes in 181 blocks; indirectly lost: 14,112 bytes in 668 blocks; possibly lost: 120,032 bytes in 3,751 blocks; still reachable: 13,409 bytes in 9 blocks
most spectests I have valgrinded have given the number above but 't/spec/S17-procasync/stress.t' stands out with 17:43
LEAK SUMMARY: definitely lost: 88,106 bytes in 957 blocks; indirectly lost: 65,868 bytes in 1,229 blocks; possibly lost: 251,384 bytes in 7,228 blocks
dalek p: b258244 | usev6++ | src/vm/jvm/QAST/Compiler.nqp:
Bulletproof against nodes without .orig on JVM

  ... as it was done with d6370eb8be for MoarVM. Makes this type
of code pass (and fixes RT #126899):
my $side-effect = 0; [andthen] Int, ++$side-effect
17:49
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=126899
FROGGS dogbert17: moarvm does an unclean shutdown, so yes, it is expected to leak 18:16
dogbert17: though you can alter your perl6 runner and add a moar switched called: --full-cleanup 18:17
dogbert17 FROGGS: I have --full-cleanup set 18:43
FROGGS well, then it is meant to be very nice to valgrind I suppose 18:44
dogbert17 here's an piece of the leak report for t/spec/S17-procasync/stress.t 18:46
==30680== 4,556 (4,020 direct, 536 indirect) bytes in 67 blocks are definitely lost in loss record 1,413 of 1,780
==30680== at 0x402C109: calloc (in /usr/lib/valgrind/vgpreload_memcheck-x86-linux.so)
==30680== by 0x4121A27: MVM_calloc (alloc.h:11)
==30680== by 0x4121CB3: MVM_io_syncpipe (syncpipe.c:115)
==30680== by 0x40E0D7F: MVM_interp_run (interp.c:4598)
==30680== by 0x41C8A22: MVM_vm_run_file (moar.c:304)
==30680== by 0x8048EA5: main (main.c:191)
AlexDaniel What's the easiest way to provide fallback subroutines if some module is not installed? 19:33
ah, wrong channel
dalek kudo/nom: 0d9b547 | usev6++ | src/core/metaops.pm:
Remove bandaid for JVM

Looks like back in 2015 those unbound closures caused NPE. This seems to be fixed now.
22:24
kudo/nom: 2285d35 | (Zoffix Znet)++ | src/core/metaops.pm:
Merge pull request #898 from usev6/jvm_cleanup

Remove bandaid for JVM