yoleaux AlexDaniel`: RT/GH weekly 07:17
ugexe jenkins.p6.nu/job/rakudo/label=linux/1/console 07:43
jenkins.p6.nu/job/rakudo/label=win.../1/console
normally tests (spectests in this case) would be in their own thing, but throwing this together it just runs make spectest during build (so you have to view the tests through the console) 07:48
slaves are dynamically provisioned and torn down on azure cloud, so still need to figure out how to hook up to an osx cloud provider the same way 08:01
Geth rakudo: 100198a3c0 | usev6++ | src/core/ThreadPoolScheduler.pm
Bandaid to unbust JVM build

For some reasons using a native results in a NullPointerException during stage parse.		 08:47
travis-ci Rakudo build failed. usev6 'Bandaid to unbust JVM build 09:12
travis-ci.org/rakudo/rakudo/builds/300871581 github.com/rakudo/rakudo/compare/8...0198a3c074
buggable [travis build above] ☠ Did not recognize some failures. Check results manually.
bartolin oh f* I messed that one up. Could someone revert my last commit, please. I have no access to my GH account atm 09:43
Geth rakudo: eebbcca1ca | (Elizabeth Mattijsen)++ | src/core/ThreadPoolScheduler.pm
Revert "Bandaid to unbust JVM build"

This reverts commit 100198a3c074bb99babde2c3ee152bd279e398da.
As requested by author on:
   irclog.perlgeek.de/perl6-dev/2017-...i_15437142		 11:27
lizmat Files=1229, Tests=75833, 319 wallclock secs (14.86 usr 5.42 sys + 2192.27 cusr 218.85 csys = 2431.40 CPU) 11:39
Geth star: virtualsue++ created pull request #107:
Minor README.md rewording		 11:48
star: 85a9abd81a | (sue spence)++ | README.md
Reworded to help people like me who stumble onto this repo by accident

when searching for rakudo star related info.		 11:50
star: 7f3a9964bb | lizmat++ (committed using GitHub Web editor) | README.md
Merge pull request #107 from virtualsue/master

Minor README.md rewording
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Revert "Bandaid to unbust JVM build" 12:10
travis-ci.org/rakudo/rakudo/builds/300908700 github.com/rakudo/rakudo/compare/1...bbcca1cad4
Geth rakudo: f87d8ef8e8 | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
Resolve confusion about short/long names in T:I:Usage

Spotted while writing tests		 14:16
lizmat afk&
[Tux] Rakudo version 2017.10-164-gf87d8ef8e - MoarVM version 2017.10-58-gad8618468
csv-ip5xs1.103 - 1.125
csv-ip5xs-2013.019 - 13.179
csv-parser12.017 - 12.851
csv-test-xs-200.439 - 0.456
test12.300 - 12.606
test-t3.282 - 3.381
test-t --race1.416 - 1.448
test-t-2060.681 - 61.515
test-t-20 --race20.764 - 22.167
 15:44
Geth rakudo: 38f51db9ff | usev6++ | src/core/ThreadPoolScheduler.pm
Bandaid to unbust JVM build

For some reasons using a native results in a NullPointerException during stage parse.		 15:56
bartolin lizmat: thanks a lot for helping out with the revert earlier today!
lizmat bartolin: yw 17:56
Geth rakudo: 6d8ed28743 | (Elizabeth Mattijsen)++ | src/core/ThreadPoolScheduler.pm
Remove $!exhausted attribute altogether

  - it is not needed, because we only need it in the supervisor loop
  - the idea of making it an attribute was to allow Telemetry to access it
   - but that is probably better served by putting that in Thread itself		 18:15
AlexDaniel oh wow… 18:57
found a segfault in a program that has just one comment 18:58
lizmat ? cool :-)
Geth rakudo: 8c5bc57ea7 | (Elizabeth Mattijsen)++ | src/core/Thread.pm
The app_lifetime of the initial thread is True

FWIW, this is only a correction for purely informational purposes
AlexDaniel two characters in total :) 18:59
u: 񭺪
unicodable6 AlexDaniel, U+6DEAA <reserved-6DEAA> [] (񭺪)
AlexDaniel hmm
let's see if we're lucky enough 19:00
m: #񭺪
camelia ( no output )
AlexDaniel m: #񭺪
camelia ( no output )
AlexDaniel m: #񭺪
camelia ( no output )
AlexDaniel m: #񭺪
camelia ( no output )
AlexDaniel m: #񭺪
camelia ( no output )
AlexDaniel m: #񭺪
camelia ( no output )
AlexDaniel e: #񭺪
evalable6
AlexDaniel e: #񭺪
evalable6
AlexDaniel nope.
samcv where is the segfault AlexDaniel 19:04
or how do i trigger it myself
m: "#\x[6DEAA]".EVAL 19:05
camelia ( no output )
Geth roast: 7a6630a9ca | usev6++ | 6 files
[jvm] Fudge some failing tests
 19:09
AlexDaniel samcv: GH#1247 19:10
synopsebot GH#1247 [open]: github.com/rakudo/rakudo/issues/1247 [SEGV] Reserved characters in the source code cause segfaults (#򫳞)
AlexDaniel samcv: let me know if you can't reproduce it
Geth rakudo: 928e24306b | (Elizabeth Mattijsen)++ | src/core/Kernel.pm
Expose getrusage<max-rss> as Kernel.memory

This is the 3rd part of exposing nqp::cpucores and nqp::getrusage to the world. We should probably decide if we want to keep these or not this way before the next release. This applies to:
  - Kernel.cpu-cores # number of cores in system
  - Kernel.cpu-usage # number of microseconds of CPU used by process
  - Kernel.memory # max resident set size in Kbytes
samcv AlexDaniel: i cannot
AlexDaniel in a loop like this? Does not segfault? 19:11
:'(
samcv: can you leave it running for a few minutes?
it takes about 5 iterations here 19:13
lizmat afk& 19:16
dogbert17 samcv: run the example in valgrind 19:21
AlexDaniel what is so special about my machine? Similarly to R#1202, people find it hard to reproduce stuff that I get in a few seconds 19:23
synopsebot R#1202 [open]: github.com/rakudo/rakudo/issues/1202 [severe] Async qqx sometimes hangs or dies ( await (^5).map({start { say qqx{… …} } }) )
AlexDaniel debian unstable 64-bit, no CPU load 19:24
dogbert17 AlexDaniel: your example did trigger valgrind on the first run, that's enough for me :)
==15104== Use of uninitialised value of size 4
==15104== at 0x41BD4F8: MVM_unicode_get_property_str (unicode.c:71883)
==15104== by 0x41BD4F8: MVM_unicode_codepoint_get_property_cstr (unicode.c:82197)
AlexDaniel dogbert17: ah, ok. Good. Good.
Geth rakudo: dbce04e6a2 | (Elizabeth Mattijsen)++ | src/core/Thread.pm
Revert "The app_lifetime of the initial thread is True"

This reverts commit 8c5bc57ea70193d57698fa42545d8e625b03829f.		 19:25
AlexDaniel dogbert17: what about R#1202 ?
dogbert17 that always hangs after a while for me. Would have preferred a SEGV though :) 19:26
AlexDaniel dogbert17: ah-ha, ok.
jnthn As another data pointer, no crash or valgrind triggering in my Linux VM on my not-so-fast home machine; This is Rakudo version 2017.09-479-g7cf5ce7 built on MoarVM version 2017.09.1-614-g1952356 19:27
Which is some way back from HEAD, I notice :) 19:29
samcv AlexDaniel: it doesn't trigger valgrind for me. are you using the latest moarvm? 19:30
i fixed an issue in MVM_unicode_get_property_str a few days ago
AlexDaniel This is Rakudo version 2017.10-165-g38f51db9f built on MoarVM version 2017.10-57-gd257df23
samcv commit 11b12a268
AlexDaniel: yeah that's too old 19:31
moarvm may or may not be bumped though
AlexDaniel it's not, as far as I can see
samcv that explains why i can't reproduce :) 19:32
ugexe is there a way to run the `make test` and `make spectest` using prove? 19:33
samcv not really i mean it won't fudge the results
ugexe thats what i was afraid of
samcv i mean maybe you could
Geth nqp/master: 14 commits pushed by pmurias++
review: github.com/perl6/nqp/compare/5dba8...854baa6795		 19:34
samcv first start up make spectest. then get a listing of files in t/spec if there's a TESTNAME.rakudo.moar file, run that, if not run TESTNAME.t
then you'll get the fudged tests if there are fudges
and still run the tests that don't todo/skip anything
ugexe running make spectest would be problematic as its automation 19:36
pmurias bartolin: github.com/rakudo/rakudo/issues/1248 - new js/jvm shared bug 19:38
Geth nqp: 693656ba32 | pmurias++ | src/vm/js/RegexCompiler.nqp
[js] Stop eol anchor from matching between \r\n
 19:41
pmurias_ ugexe: you can run fudged tests with prove 19:42
ugexe im trying to find how to fudge the tests... notably those for `make test` without first cloning the spectest 19:43
pmurias ugexe: perl t/spec/fudgeall rakudo.js t/spec/S01-perl-5-integration/*.t returns a list of fudged filenames
ugexe any suggestion for make test? if a build agent fails make test I could avoid cloning the spectest 19:44
pmurias ugexe: do you need anything special for 'make test' 19:45
?
ugexe: it seems that those are just normal test files 19:46
ugexe t/03-jvm
pmurias ugexe: the test seems to compile the .java file itself 19:48
pmurias_ ugexe: and I wouldn't run this on backends that aren't jvm 19:49
travis-ci NQP build failed. pmurias '[js] Fix bug in frugal quant' 19:56
travis-ci.org/perl6/nqp/builds/301051659 github.com/perl6/nqp/compare/5dba8...4baa679541
Geth nqp: c8519551ec | pmurias++ | t/nqp/109-coercions.t
[jvm] Skip failing test on the JVM
 20:04
travis-ci NQP build passed. pmurias '[jvm] Skip failing test on the JVM' 20:27
travis-ci.org/perl6/nqp/builds/301061397 github.com/perl6/nqp/compare/69365...519551ecec
dogbert17 AlexDaniel: wrt #1202, I managed to get a tiny bit of information from gdb after the program hung, whether it's useful I'll leave to jnthn. I could see a bit of what *one* thread was up to 20:35
(gdb) bt
#0 0xb7783cb0 in ?? ()
#1 0xb73b3859 in MVM_reentrantmutex_lock_checked (tc=0xb34a58f8, lock=0xb3f51d10) at src/6model/reprs/ReentrantMutex.c:121
#2 0xb73367ad in MVM_interp_run (tc=0xb34a58f8, initial_invoke=0xb73516aa <thread_initial_invoke>, invoke_data=0xb35c6630) at src/core/interp.c:4071
#3 0xb73517e4 in start_thread (data=0xb35c6630) at src/core/threads.c:85
AlexDaniel releasable6: status 21:26
releasable6 AlexDaniel, Next release in 5 days and ≈21 hours. No blockers. 0 out of 169 commits logged
AlexDaniel, Details: gist.github.com/66686c5231ccb7f3e7...a1a3831a5d
japhb buggable: speed :4 21:35
buggable japhb, ▃ ▂ ▃ ▅ █ ▃↑▃ ▃ ▅ █ dates: 2017-10-26–2017-11-12
japhb, █ ▂ █ ▁█ ▆█ █▇ ▆███ █▅ ▇ █ █▁ range: 19.341s–22.639s
japhb, ▆█▄█ █▆▆▅▇██ ██▄████████▇ ██▆█▇█ ██ speed: 4% slower (widths: 9/3)
japhb, ▁▂████▅▇▇███████▆▇████████████▄██████▃██
japhb Hmmm, highly variable 21:36
buggable: speed :4 test-t-20
buggable japhb, ▄ ▂▇ ↑ ▅ ▂ ▅ dates: 2017-10-26–2017-11-12
japhb, ▂█ ▂ ▃ ▆▅▆ ██▅▅▄▇ █▄ ██ ▂ █ ██ range: 56.192s–62.623s
japhb, ██ █▃█▆▇▃███▃▅█████████▆██▇ ▇█▆█▂▂██ speed: 6% slower (widths: 9/3)
japhb, ↓▄██▁▄▄████████████████████████▇████████
japhb Not much better. :-(
BenGoldberg One of those has a face in it! 22:04

Powered by Cro and the Raku® Programming Language.

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

Thank you!