Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_log/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by Zoffix on 27 July 2018.
00:06 nebuchadnezzar joined 00:07 p6bannerbot sets mode: +v nebuchadnezzar 00:11 lizmat left
Geth rakudo: 4307996ae7 | (Samantha McVey)++ | 2 files
Add additional dashed variants for utf16le and utf16be

Add utf16-le, utf-16-le, utf16-be, utf-16-be as variants for utf16le and utf16be. This makes it consistent with the variants for other utf's including utf8-c8.
01:10
samcv hmm Geth seems to be dead 01:13
Zoffix: ping
01:18 cognominal-p6 joined 01:19 p6bannerbot sets mode: +v cognominal-p6
Geth roast: 28516c4272 | (Samantha McVey)++ | S32-encoding/registry.t
Add missing encodings to S32-encoding/registry.t

Add utf32, uf16le, utf16be, windows-932, windows-1251 and windows-1252.
01:19
AlexDaniel samcv: ? 01:36
it seems to be working? 01:37
Zoffix samcv: pong 01:38
samcv: do you have sudo on hack server? Geth is there. `sudo service geth restart`
01:39 cognominal-p6 left, cognominal-p6 joined
Zoffix lol "Blin"? Is that Slavic for crêpe 01:40
01:40 p6bannerbot sets mode: +v cognominal-p6
Zoffix ? 01:40
m: class A {}; subset Meows of Callable where .signature ~~ :(Str, Int, Str --> A); sub a (Meows \handler) { A.new }; a sub (Str, Int, Str --> A) {} 01:46
camelia ( no output )
Zoffix m: class A {}; subset Meows of Callable where .signature ~~ :(Str, Int, Str --> A); sub a (Meows &handler) { A.new }; a sub (Str, Int, Str --> A) {}
camelia Type check failed in binding to parameter '&handler'; expected Meows but got Sub+{Callable[A]} (sub (Str $, Int $, St...)
in sub a at <tmp> line 1
in block <unit> at <tmp> line 1
Zoffix m: class A {}; subset Meows where .signature ~~ :(Str, Int, Str --> A); sub a (Meows &handler) { A.new }; a sub (Str, Int, Str --> A) {}
camelia ( no output )
01:46 MasterDuke left
Zoffix It's kinda weird. I would've thought it to demand a Callable that returns a `Meows` 01:47
AlexDaniel Zoffix: yes 01:51
Zoffix :D 01:52
I'm laughing now at that :) 01:53
AlexDaniel I like that name too :) 01:55
Geth rakudo: 163f6e5dc9 | (Ben Davies)++ | src/core/IO/Socket/Async.pm6
Pass :hint-affinity to create-socket when calling .listen
01:56
rakudo: 9ed57a4c64 | (Ben Davies)++ | 2 files
Use .result instead of await on return value of create-socket

Fixes #2402
synopsebot RAKUDO#2402 [closed]: github.com/rakudo/rakudo/issues/2402 [⚠ blocker ⚠] t/05-messages/02-errors.t crashes
rakudo: f758f4d911 | (Zoffix Znet)++ (committed using GitHub Web editor) | 2 files
Merge pull request #2408 from Kaiepi/async-socket

Fix errors related to async socket introspection
AlexDaniel oh, so no need to revert all that? 01:57
wonderful
Kaiepi++
Kaiepi :) 01:59
now all that's left is dealing with the jvm
AlexDaniel weekly: More tools! github.com/perl6/Blin 02:01
notable6 AlexDaniel, Noted!
Geth rakudo: zoffixznet self-assigned LTA: Heredocs give warning trouble with empty lines that are partially indented github.com/rakudo/rakudo/issues/2409
FCO++ created pull request #2412: Make eqv on signatures test the returns. So, multi method stubs will …
02:10
AlexDaniel Kaiepi: so currently rakudo-jvm is failing to build because of that? 02:21
Kaiepi yeah 02:36
github.com/rakudo/rakudo/issues/2406
02:41 Zoffix left
08:53 lizmat joined, p6bannerbot sets mode: +v lizmat
lizmat Files=1257, Tests=76299, 336 wallclock secs (15.55 usr 5.23 sys + 2361.84 cusr 242.52 csys = 2625.14 CPU) 08:53
08:57 [Tux] left
Geth rakudo: abba67b228 | (Fernando Correa de Oliveira)++ | src/core/Signature.pm6
Make eqv on signatures test the returns. So, multi method stubs will enforce return values. #2385
09:04
rakudo: f0210794dd | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/core/Signature.pm6
Merge pull request #2412 from FCO/master

Make eqv on signatures test the returns. So, multi method stubs will …
synopsebot RAKUDO#2385 [open]: github.com/rakudo/rakudo/issues/2385 [LTA] LTA role multi method stubs enforce arguments but not return value
09:37 cognominal-p6 left 09:44 [Tux] joined 09:45 p6bannerbot sets mode: +v [Tux]
Geth ¦ rakudo: lizmat self-assigned my %*FOO is Set = <a b c> dies github.com/rakudo/rakudo/issues/2360 09:48
lizmat bisectable6: my %*FOO is Set = <a b c> 09:49
bisectable6 lizmat, Bisecting by output (old=2015.12 new=f021079) because on both starting points the exit code is 1
lizmat, bisect log: gist.github.com/ecd1c99dffb494638c...4d4add1560
lizmat, (2017-05-26) github.com/rakudo/rakudo/commit/0c...f1a759afee
12:32 cognominal-p6 joined 12:33 p6bannerbot sets mode: +v cognominal-p6 12:38 cognominal-p6 left, cognominal-p6 joined 12:39 p6bannerbot sets mode: +v cognominal-p6 12:57 cognominal-p6 left
Geth ¦ rakudo: zoffixznet self-unassigned LTA: Heredocs give warning trouble with empty lines that are partially indented github.com/rakudo/rakudo/issues/2409 13:59
¦ rakudo: zoffixznet assigned to lizmat Issue LTA: Heredocs give warning trouble with empty lines that are partially indented github.com/rakudo/rakudo/issues/2409
roast: 7367a04d89 | (Zoffix Znet)++ | S32-str/indent.t
Cover .indent opts and bug

Covers github.com/rakudo/rakudo/issues/2409 and potential optimizations
14:11
roast: 58ac266eea | (Zoffix Znet)++ | S32-str/indent.t
Fudge indent warning test

  github.com/rakudo/rakudo/issues/2409
14:12
15:02 fake_space_whale joined 15:03 p6bannerbot sets mode: +v fake_space_whale 15:50 lizmat left 15:52 cognominal-p6 joined 15:53 p6bannerbot sets mode: +v cognominal-p6 16:04 cognominal-p6 left, cognominal-p6 joined 16:05 p6bannerbot sets mode: +v cognominal-p6 16:20 cognominal-p6 left 16:47 cognominal-p6 joined, p6bannerbot sets mode: +v cognominal-p6 16:54 cognominal-p6 left 16:55 cognominal-p6 joined, p6bannerbot sets mode: +v cognominal-p6
Geth nqp/nqp-mbc: 15 commits pushed by (Stefan Seifert)++
review: github.com/perl6/nqp/compare/6d8c2...c884763b25
17:13
17:23 MasterDuke joined, p6bannerbot sets mode: +v MasterDuke, MasterDuke left, MasterDuke joined, herbert.freenode.net sets mode: +v MasterDuke, p6bannerbot sets mode: +v MasterDuke
timotimo i wonder if "DRY in MAST::Call.new" could benefit from "my int $primspec" instead of "my $primspec", or if that just causes lots of int-to-num-to-int code in nqp again 18:16
18:41 ExtraCrispy left 18:42 ExtraCrispy joined, p6bannerbot sets mode: +v ExtraCrispy
AlexDaniel so if zef says something like this: CONFIGURATION /home/alex/.rakudobrew/moar-master/install/share/perl6/site/resources/4B6FE7F170929FBA2A501CB141A7B073111FA208.json 19:04
where the configuration file is some sort of a sha
does it mean that the config file is different depending on which version of rakudo you use?
19:05 brrt joined
AlexDaniel also, if I'm using zef with disabled auto-update, how is it going to behave when running on different rakudo versions? 19:05
does it need to fetch at least once per rakudo version or not?
19:06 p6bannerbot sets mode: +v brrt
brrt ohai 19:07
yoleaux 19 Oct 2018 16:10Z <Zoffix> brrt: does R#2395 (Proc.in.close hang/long time) sound like anything you may have worked on or do you have any ideas about that ticket? I saw some commits about event loops in MoarVM. Dunno if that's what's involved in handling closing of STDIN for a proc (it uses Proc::Async under the hood)
synopsebot R#2395 [open]: github.com/rakudo/rakudo/issues/2395 [⚠ blocker ⚠] t/spec/S32-io/io-cathandle.t is now 2x-8x slower without precomp
yoleaux 19 Oct 2018 17:51Z <Zoffix> brrt: nevermind, I managed to bisect that ticket to post-release-opts merge in MoarVM, so it wasn't the event loop stuff. Still any ideas on what could be the bug welcome :)
19 Oct 2018 18:02Z <Zoffix> brrt: It's this commit that's causing it: github.com/MoarVM/MoarVM/commit/8f...2935ea61ea
brrt okay Zoffix
timotimo oh hey brrt 19:08
are you interested in checking this out?
brrt ohai timotimo
timotimo i'm just a little baffled by it
brrt i'm silenced in #moarvm :-(
well, to be honest, I can have a look, but I won't be doing much coding this evening
timotimo i have not yet been given privileges to op myself there
AlexDaniel brrt: fixed
brrt ty AlexDaniel :-)
timotimo OK, so the problem with that bug is: attaching anything in the way of debugging tools makes it go away
i tried strace and perf trace, though i have not tried perf record or rr record yet 19:09
additionally, i didn't reproduce it locally, though i think i might know how to do it
what's more, reverting the commit in question fixing the bug doesn't make much sense to me 19:10
brrt lol
timotimo first i thought maybe the spesh facts aren't being marked as required and so there's deopt trouble
but i do see code to do the marking
brrt what commit is the matter 19:11
the commit that Zoffix gave us?
timotimo yes
brrt hmmmmm 19:12
timotimo Zoffix, can you perhaps spin up that vm again?
brrt what do the JIT logs say?
eh, spesh logs, these days
timotimo 17th of august; i'd say there were still jit logs then
no, wait
that was last month
2 months ago
i really need to get the frickin list of month names to numbers and backwards into my bleepin' head 19:13
damn you, selfish roman emperors
[Tux] I cannot explain. I ran three today and all were over 1.75 19:29
Rakudo version 2018.09-488-gf0210794d - MoarVM version 2018.09-137-g4582780a5
csv-ip5xs-207.745 - 8.177
csv-parser22.656 - 24.453
csv-test-xs-200.427 - 0.432
test8.658 - 9.218
test-t1.933 - 2.233
test-t --race0.944 - 1.064
test-t-2031.394 - 36.432
test-t-20 --race11.212 - 11.472
19:34 Zoffix joined, p6bannerbot sets mode: +v Zoffix
Zoffix brrt: timotimo still need me to spin up that VM? 19:34
timotimo if you do, i'll have another look at the bug, and maybe brrt can also have a look with, like, a shared tmux session or so 19:36
Zoffix brrt: timotimo ssh [email@hidden.address] 19:39
timotimo TYVM
Zoffix brrt: info on where's the code and whatnot here: colabti.org/irclogger/irclogger_log...10-19#l955 19:40
timotimo i don't like how fickle tests can be that try to run rakudo 19:47
when i want to instrument it with some stuff
actually, the precompilation attempts fail? 19:49
Zoffix: you didn't tell me that "rr" exists as an alias :D 19:55
i was trying to figure out why the F trying to record the perl6 process would give me "Tests: ...github.../" 19:56
Zoffix has no idea what "rr" is
timotimo it can record and replay and time-traveling-debug program execution 19:57
except it doesn't seem to run on the VM for some reason 19:58
20:00 lizmat joined, p6bannerbot sets mode: +v lizmat 20:02 brrt left
timotimo damn it, i can't even reproduce the bug on the VM now 20:02
Zoffix :| 20:04
Did you remember to declutter the precomp files?
timotimo i think i did it right, but perhaps not
can you connect as cpan@ and "tmux a" to get into my session?
Zoffix The test file has "my $*EXECUTABLE = "./perl6-m";" in it... Could that be affecting the non-reproduction of bug? 20:06
timotimo ah, i did that!
i threw it out just now
did you change something else? 20:07
Zoffix just ran declutter
timotimo regenerated the .rakudo.moarvm file?
i had some dumb debug output hacked in there %) 20:08
Zoffix Yeah
k, now I can repro the bug
cpan@temp-perl6-dev:~/R/rakudo$ declutter; t/fudgeandrun t/spec/S32-io/io-cathandle.t
with that
Zoffix afk
timotimo OK 20:09
I ... i can't reproduce it though 20:12
this bug is *really* slippery
Zoffix I can. You sure you're running `declutter` and the rest in ~/R/rakudo? 20:19
timotimo i think i am. you're not rm-ing the .precomp folder under t/spec/Packages or what it was?
packages not Packages
Zoffix declutter is aliased to `rm -fr */.precomp; rm -fr */*/.precomp; rm -fr */*/*/.precomp; rm -fr */*/*/*/.precomp' 20:20
So presumably yes I am
timotimo oh, ok 20:21
how do you decide whether you reproduced it?
the whole fudgeandrun takes 13 seconds at the moment, but the test i'm looking at in particular, the last one, runs in under a second 20:22
20s user time, 13.5 real time 20:23
Zoffix lemme measure
So after doing t/fudgeandrun so it gens the fudge file, I do `declutter; time ./perl6 t/spec/S32-io/io-cathandle.rakudo.moar` and get 9.864s while on perl6 I get it for 5.5 20:25
Less than 40s vs. 3s I had before but still visible diff. 20:26
timotimo it's not the bug any more where it just stands still for about 15 to 20 seconds for no discernable reason 20:27
that was what i was trying to hunt down
and that is gone
i'd say the other thing is "just" a performance regression
Zoffix FWIW, the other two blockers are far more important IMO. This is just a perf penalty in some arcane code on some machines, while others are clear regressions R#2400 and R#2355
synopsebot R#2400 [open]: github.com/rakudo/rakudo/issues/2400 [fudged tests committed][regression][⚠ blocker ⚠] Module Crane is failing tests on HEAD
R#2355 [open]: github.com/rakudo/rakudo/issues/2355 [SEGV][regression][⚠ blocker ⚠] Concurrency feature race/hyper sometimes failed with Supply
Zoffix I mean, if you have the skills to fix them 20:28
The R#2400 in particular. There's already a fix for a very similar bug, so presumably the new fix would be easy if you know that codepath
timotimo: so did you make some changes to get the bug to go away or did it just disappear randomly?
timotimo not exactly sure
i tried a lot of stuff to hook up diagnostics and tools and such, and that never gave me the hang 20:29
MasterDuke Zoffix: btw, i ran 2355 in a loop for a couple hours and couldn't repro
20:33 robertle left
Zoffix vOv 20:34
20:45 pmurias left
Zoffix timotimo: are you still using the VM? 21:08
Zoffix kills it 21:35
22:12 Zoffix left
Geth nqp: 20f0f10ea3 | (Nick Logan)++ | 66 files
Revert "[jvm] global clean (#507)"

This reverts commit 96714f3be1760750441aef91c00c2b1863cd44c1.
Resolves github.com/rakudo/rakudo/issues/2406
23:31
AlexDaniel heh, github is feeling very bad 23:46
23:57 AlexDaniel left
geekosaur yep 23:57
geekosaur hearing geek screams everywhere
23:58 AlexDaniel joined, p6bannerbot sets mode: +v AlexDaniel
Geth ¦ rakudo: tbrowder self-assigned @.allowed array in class Pod::Block::Code is unneeded github.com/rakudo/rakudo/issues/2413 23:59