timotimo only a ton of "TODO passed" messages 00:00
AlexDaniel alright then 00:05
travis-ci Rakudo build passed. Timo Paulssen 'teach X::Syntax::Number::LiteralType about natives 00:16
travis-ci.org/rakudo/rakudo/builds/341675265 github.com/rakudo/rakudo/compare/1...30e2537f87
samcv timotimo: i agree. a make target would be great 00:17
travis-ci Rakudo build passed. Timo Paulssen 'missed the case where Rat was assigned to int' 00:56
travis-ci.org/rakudo/rakudo/builds/341678887 github.com/rakudo/rakudo/compare/e...a1d25ac672
samcv hmm it seems doing the XOR trick to swap variables is faster than doing (@a[$i], @a[$j]) = (@a[$j], @a[$i]); 07:26
did not expect that 07:27
though using a temp variable is slightly faster than doing the XOR method 07:28
[Tux] Rakudo version 2018.01-199-g63a1d25ac - MoarVM version 2018.01-91-g1c79ca35d
csv-ip5xs0.821 - 0.950
csv-ip5xs-207.522 - 7.877
csv-parser11.522 - 12.752
csv-test-xs-200.459 - 0.472
test9.152 - 9.270
test-t2.543 - 2.578
test-t --race1.093 - 1.114
test-t-2046.807 - 47.667
test-t-20 --race16.096 - 16.739
08:04
psch o/ 08:43
as suspected, making each p6bindsig_no_autothread label unique by giving them IDs fixes GH #1526 08:55
i'm not quite sure it's the right fix though -- the fact that they didn't need to be unique makes me assume that we only want one p6bindsig call per QB (i.e. java method) 08:56
psch adds that to the ticket as well
the underlying issue seems to be that we emit multiple p6bindsig declarations in a single InstructionList, and p6bindsig declares a Label with a constant name, which thus throws the redeclaration 09:15
s/emit/apparently emit/
GH #1431 is apparently JDK9-specific, which we don't have on hack yet 09:20
also, nqp/docs/jvminterop-goals.md contains POD 09:52
this is a bit silly, seeing as it's named .md :l
jnthn huh, why'd one block ever have more than one p6bindsig, given signatures are per block? 10:17
psch jnthn: i haven't looked deeply into what we emit during codegen, but considering we get a redecl for a label name for a label defined inside p6bindsig this seems to be what's happening 10:21
github.com/rakudo/rakudo/issues/1526 is the issue fwiw
jnthn Don't have time to look now, alas, but my point is that it sounds like we're generating a bad QAST tree in the first place 10:22
psch right 10:23
the change that brought this about was about Whatever currying, but i'd guess that it's a difference in MAST vs JAST that we only now see 10:24
don't really have time to look deeply into it right now either unfortunately
jnthn I suspect the MAST code-gen is just more forgiving so lets it pass 10:28
psch i can't find the MAST::Label declaration, but i don't see us giving it a name (or anything really) in p6bindsig 10:43
but i have no clue how labels work on moar... :) 10:44
jnthn I don't think p6bindsig compiles into anything involving a label on moar 10:51
tbrowder \o good [AP]M all 12:23
it looks like the stderr stream on the jvm has some problems 12:24
adding an nqp spawnprocasync test for stderr works on moarvm but not on jvm 12:26
it looks like jvm issues are filed on github:perl6 so i will file there unless told otherwise 12:28
psch huh, nqp::sayfh is gone? 12:33
well, from nqp-m at least it seems
r: use nqp; nqp::sayfh(nqp::getstderr(), "foo")
camelia foo
===SORRY!===
No registered operation handler for 'sayfh'
Zoffix "Need at least JVM 1.8 (got java version "1.7.0_151" what package do I neeed to install to get better version? 12:38
on debian/buntu/bodhi
psch Zoffix: openjdk-8-jdk i think? 12:39
Zoffix hm, all I got is -7- 12:40
Zoffix tries another VM
psch packages.debian.org/de/sid/openjdk-8-jdk
i haven't used debian in quite a while...
Zoffix this box is bodhi linux, which is a fork of ubuntu, (which is a fork of debian). I guess its repos just lag behind 12:41
psch maybe there's a ppa somewhere? 12:42
or yeah, just use a VM with a different *nix
Zoffix yeah, my debian VM got -8- in it
ilmari or the -headless version, if you dont need gui stuff 12:43
stretch (debian 9), released last june, has jdk 8. it's also in backports for debian 8 (jessie) 12:45
ubuntu xenial (16.04 LTS) also has jdk8 12:47
Zoffix lsb_release tells me my bodhi is based on buntu trusty 12:48
tbrowder .tell AlexDaniel My nqp PR #414 is now passing all checks after a second commit. Ready for merge if it satisfies you. 13:48
yoleaux tbrowder: I'll pass your message to AlexDaniel.
timotimo psch: yeah, string-oriented output has gone from moar, we only do byte-oriented output now. there's a "note" sub, though, that will take care of encoding a string and outputting it for you 13:49
Zoffix Working on JVM backend makes me feel like a buddhist monk. 14:04
ZofBot: "We cannot learn real patience and tolerance from a guru or a friend. They can be practiced only when we come in contact with the JVM backend. According to Shantideva, JVM backend bugs are really good for us as we can learn a lot from them and build our inner strength."—Dalai Lama 14:05
ZofBot Zoffix, moar/install/share/nqp/lib/QRegex
Zoffix Tho I'm fairly sure I got I fix for R#1526
synopsebot R#1526 [open]: github.com/rakudo/rakudo/issues/1526 [JVM] [JVM] build broken after re-design of Whatever currier
Geth rakudo: 30b82b9823 | (Timo Paulssen)++ | src/core/Exception.pm
LiteralType: make sure natives are lowercased

while at the same time not suggesting "num" or "int" as methods to call for the coercion.
14:15
Zoffix ZOFFLOP: t/spec/S10-packages/basic.t 14:36
weird. `make some/test/file` works fine (on JVM backend), but `make spectest` all files die with "Parse errors: No plan found.." 14:41
psch Zoffix: something about parallelization, TEST_JOBS=1 does help
timotimo: thanks for the info. carrying that change over to nqp-j probably wants some more consideration wrt IO i assume..? 14:43
Geth rakudo: f2188e45fb | (Zoffix Znet)++ | src/Perl6/Actions.nqp
Fix whatever currier on JVM

Fixes github.com/rakudo/rakudo/issues/1526 R#1526
When we gut a WhateverCode to relocate its bits into a chained curry, don't copy over the p6bindsig op.
On MoarVM we use the VM binder and the curry doesn't go through
  "need full binder" path that pushes this op into declarations,
which is why the crash in the ticket was never observed on MoarVM.
synopsebot R#1526 [open]: github.com/rakudo/rakudo/issues/1526 [JVM] [JVM] build broken after re-design of Whatever currier
Zoffix psch: ahh... yeah, I got TEST_JOBS=30
Geth rakudo: 7a15d9de27 | (Zoffix Znet)++ | tools/build/Makefile-JVM.in
Remove trailing whitespace
14:56
rakudo: 0d0fe79341 | (Zoffix Znet)++ | 3 files
Test whatever curries with 3+ args do not duplicate p6bindsig op

  github.com/rakudo/rakudo/issues/1526
Rakudo fix: github.com/rakudo/rakudo/commit/f2188e45fb
14:57
lizmat commute to NR-PM meeting& 15:04
decommute& 20:59
Geth nqp: 18fc370a0f | (Jeremy Studer)++ | src/vm/js/Operations.nqp
[js] Allow 'chain' op to use child as callee

Modify the 'chain' op to allow the option to use the first child as the callee. Before the name of the op served as the operator sub and the children the operands. This modification makes it so that, if there is no name provided to the chain, child 0 serves as the operator and children 1 and 2 the operands. ... (5 more lines)
21:15
nqp: 641fafd8b6 | (Paweł Murias)++ (committed using GitHub Web editor) | src/vm/js/Operations.nqp
Merge pull request #390 from jstuder-gh/js_neg_chaining

Rakudo Issue 1304: [js] Allow 'chain' op to use child as callee
[Coke] boggles to see a non-pmurias commit to nqp-js. woot! 21:23
that's awesome!
Geth nqp/master: 13 commits pushed by pmurias++
review: github.com/perl6/nqp/compare/641fa...f27d17ca75
21:27
pmurias [Coke]: I had some in the past ;) , I'm not sure how to encourage them more 21:28
japhb pmurias: It might help to give people an idea of where the current js implementation is still lacking, and what you are currently working on, so they can either choose to help you on your current task, or stay out of your way by working somewhere else that scratches their particular itch.. 21:31
[Coke] pmurias: any thoughts on webassembly? does it make sense to go through JS to get there, or should we consider another nqp backend?
japhb pmurias: For a while it looked like you were concentrating on encodings and string manipulation / regex -- but the last set of commits look like general cleanups. 21:33
pmurias japhb: I'll try to write something up tommorow when I'm fresher and more eloquent 21:35
[Coke]: IMHO webassembly needs to evolve a fair bit before it becomes a sensible Perl 6 target
[Coke]: the last time I checked it was super C/C++/Rust focused offering just an abstracted subset of machine instruction sets 21:37
Geth nqp: d3794d7a15 | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/build/MOAR_REVISION
[MoarVM Bump] Brings 5 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...gc7774e37d c7774e37d [Spesh] Always split if_o into istrue and if_i 3a100179c Merge pull request #802 from MasterDuke17/patch-1 0db0645f5 Fix typos in comments 1c79ca35d set test jobs back to 1 c512dce8e travis: use 2 cores for compilation and tests
22:10
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...gc7774e37d
rakudo: beaad7b979 | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/build/NQP_REVISION
[NQP Bump] Brings 28 commits

NQP bump brought: github.com/perl6/nqp/compare/2018....gd3794d7a1 d3794d7a1 [MoarVM Bump] Brings 5 commits f27d17ca7 [js] Minor cleanup 9a53d1d6f Test 3 argument nqp::chain ff9d008d2 Test flattening of natively typed arrays ... (24 more lines)
22:11
¦ rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....gd3794d7a1
pmurias japhb: the last set of commits was refactoring for a calling convention change I need to do to get some test to pass 23:33