timotimo m: my $test = Int; my $p = Foo => Bar => $test; $p.value.value = 99; say $test; 01:43
camelia 99
timotimo i wonder what's giving me trouble in my own code
haha, i stripped the container with <>, which did what i wanted, but at the wrong time 01:45
DSLs are fun, i should make them more often 01:47
Geth roast: 7dd5939eda | (Zoffix Znet)++ | S32-io/IO-Socket-Async.t
Remove trailing whitespace
 04:23
rakudo: 7ba50d8647 | (Zoffix Znet)++ | src/core/Buf.pm
Fix explosions when hypering over Buf

Buf.list returns a Seq, which explodes when we rake its guts for List stuff inside &nodemap.
Fix by making it return a List. Part of addressing
  github.com/rakudo/rakudo/issues/1344
Bug find: irclog.perlgeek.de/perl6/2018-01-07#i_15660359		 04:28
roast: 798783b4f9 | (Zoffix Znet)++ | S32-container/buf.t
Test no explosions when hypering Buf

Rakudo fix: github.com/rakudo/rakudo/commit/7ba50d8647 Bug find: irclog.perlgeek.de/perl6/2018-01-07#i_15660359		 04:30
nqp: 5afa8bc715 | (Zoffix Znet)++ | 6 files
Use faster nqp::istype instead of ~~

Simple bench shows it 3x faster		 04:43
travis-ci NQP build passed. Zoffix Znet 'Use faster nqp::istype instead of ~~ 04:56
travis-ci.org/perl6/nqp/builds/325959585 github.com/perl6/nqp/compare/e8dde...fa8bc715b2
Zoffix Does "!inlinable" mean inlinable? QAST::OperationsJAST.add_core_op('callstatic', :!inlinable, $call_codegen); 05:00
I see it's set on call and callstatic but not on, say callmethod or chain
nope, it sets inlineability to 0 05:14
travis-ci Rakudo build failed. Zoffix Znet 'Fix explosions when hypering over Buf 05:24
travis-ci.org/rakudo/rakudo/builds/325957223 github.com/rakudo/rakudo/compare/6...a50d8647ac
buggable [travis build above] ☠ Did not recognize some failures. Check results manually. Across all jobs, 2 unique test files failed.
Geth nqp: 41b724c5dd | (Zoffix Znet)++ | 2 files
Remove trailing whitespace
 05:26
nqp: 4145205863 | (Zoffix Znet)++ | 3 files
Implement nqp::chainstatic op

chainstatic to chain is what callstatic is to call
travis-ci NQP build failed. Zoffix Znet 'Implement nqp::chainstatic op 05:31
travis-ci.org/perl6/nqp/builds/325966893 github.com/perl6/nqp/compare/41b72...45205863b2
Zoffix "Confused at line 457, near "my $chain_"" ... no idea what it's confused about... 05:35
Geth nqp: 41e36ea844 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
[MoarVM Bump] Brings 2 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...2-gc2de234 c2de234 Merge pull request #776 from dogbert17/fix-typo 47769a5 Fix typo		 05:37
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...2-gc2de234
rakudo: a3825a4a17 | (Zoffix Znet)++ | tools/build/NQP_REVISION
[NQP Bump] Brings 4 commits

NQP bump brought: github.com/perl6/nqp/compare/2017....8-g41e36ea 41e36ea [MoarVM Bump] Brings 2 commits 4145205 Implement nqp::chainstatic op 41b724c Remove trailing whitespace 5afa8bc Use faster nqp::istype instead of ~~
MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...2-gc2de234 c2de234 Merge pull request #776 from dogbert17/fix-typo 47769a5 Fix typo
rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2017....8-g41e36ea
b77d87566f | (Zoffix Znet)++ | src/Perl6/Optimizer.nqp

  - Makes 2.4x faster chains like 2 < $a < 1337
  - Makes 2 < $a < 1337 run about the same speed as 2 < $a && $a < 1337
   Phixes github.com/rakudo/rakudo/issues/1368
  - With longer chains, the chain version comes out noticeably faster
   than non-chained version with && ops
Zoffix
.oO( need a command what shows what commits a bump would bring before bumping... )
travis-ci NQP build failed. Zoffix Znet 'Remove trailing whitespace' 05:39
travis-ci.org/perl6/nqp/builds/325966794 github.com/perl6/nqp/compare/5afa8...b724c5dd97
Zoffix .tell bartolin I added chainstatic on JVM github.com/perl6/nqp/commit/4145205863 Travis says it's confused about something but I dunno what it's problem is: travis-ci.org/perl6/nqp/jobs/325966896 FWIW I didn't build anything on JVM or JS 05:40
no bot; greatp
travis-ci NQP build failed. Zoffix Znet '[MoarVM Bump] Brings 2 commits
travis-ci.org/perl6/nqp/builds/325968684 github.com/perl6/nqp/compare/41452...e36ea8444d
Rakudo build failed. Zoffix Znet '[NQP Bump] Brings 4 commits 05:59
travis-ci.org/rakudo/rakudo/builds/325968690 github.com/rakudo/rakudo/compare/7...825a4a17f7
buggable [travis build above] ☠ Did not recognize some failures. Check results manually. Across all jobs, 2 unique test files failed.
Zoffix Man, this was super easy.
Also...
Achievement Unlocked: Create an NQP op 07:24
Geth rakudo: 6632879618 | (Zoffix Znet)++ | src/Perl6/Optimizer.nqp
Move comment closer to code it describes
 07:32
travis-ci Rakudo build failed. Zoffix Znet 'Staticalize chain ops when we can 07:44
travis-ci.org/rakudo/rakudo/builds/325984059 github.com/rakudo/rakudo/compare/a...7d87566f30
buggable [travis build above] ☠ All failures are due to: failed make test (3 failures). Across all jobs, 4 unique test files failed.
Geth rakudo: 3172ef7f5b | (Zoffix Znet)++ | .travis.yml
Add verbose `make test` run

To try to get more info on travis's Linux sickness		 07:46
rakudo: 2bfc882422 | (Zoffix Znet)++ | .travis.yml
Travis verbose make test tweak

I guess it needs make install first, for ./perl6 to exist		 07:57
rakudo: a0634e5fe4 | (Zoffix Znet)++ | .travis.yml
More travis attempt
 08:10
rakudo: 99923747ee | (Zoffix Znet)++ | src/core/Any.pm
Make &dd more resilient to low-level stuff

Stuff like
   dd nqp::getattr(-> Int(Cool) $ {}.signature.params.head,
   Parameter, q|$!coerce_method|)
that crashes because there's no .dispatch:<.?> nor .perl methods on this object.
Fix by wrapping those calls into `try` and marking such things with "low level" in output.		 08:12
rakudo: 880dbfaa13 | (Zoffix Znet)++ | .travis.yml
One more tweak for travis verbose make test
 08:24
Zoffix so make test fails, but verbose make test succeeds -_- 08:25
Geth nqp: bff31bf777 | usev6++ | src/vm/jvm/QAST/Compiler.nqp
[jvm] Remove superfluous parenthesis

Leftover from 4145205863; fixes NQP build.		 08:27
Zoffix bartolin++ 08:28
Totally missed my eye
bartolin also Zoffix++
bartolin will build a r-j with the new chainstatic op later today 08:29
travis-ci Rakudo build errored. Zoffix Znet 'More travis attempt' 08:35
travis-ci.org/rakudo/rakudo/builds/325988320 github.com/rakudo/rakudo/compare/2...634e5fe4b3
buggable [travis build above] ☠ Did not recognize some failures. Check results manually.
travis-ci NQP build passed. usev6 '[jvm] Remove superfluous parenthesis 08:41
travis-ci.org/perl6/nqp/builds/325990363 github.com/perl6/nqp/compare/41e36...f31bf777f3
Zoffix k, so travis now does two make tests; one normal and one verbose. And on linux, the non-verbose is still fails in random files, often with errno 9, but verbose always succeeds. Gonna ping Travis again on Monday to see if they know anything. There was some other dude on twitter also having different issues after the New Years drive issue 08:50
geekosaur errno 9, or error 9 reported by make? the latter could be the Travis infrastructure SIGKILL-ing it 09:00
Zoffix Just says "Non-zero wait status: 9" travis-ci.org/rakudo/rakudo/jobs/3...8321#L1476 09:02
geekosaur that's definitely SIGKILL 09:07
it ran out of memory and the OOM killer got it, or there's something wrong with Travis's containers
Zoffix Ah cool. I'll mention that when I ping Travis 09:08
geekosaur ("wrong" = misconfigured probably; really wrong would probably be illegal instructions or etc.)
travis-ci Rakudo build errored. Zoffix Znet 'One more tweak for travis verbose make test' 09:38
travis-ci.org/rakudo/rakudo/builds/325990041 github.com/rakudo/rakudo/compare/9...0dbfaa134b
buggable [travis build above] ☠ Did not recognize some failures. Check results manually.
Geth rakudo/segv-coercer-qast: 8d34c4dfaa | (Zoffix Znet)++ | src/Perl6/Actions.nqp
Push segv commit
 10:10
nqp: 2666a65cba | (Zoffix Znet)++ (committed using GitHub Web editor) | docs/qast.markdown
Fix markdown error
 10:37
travis-ci NQP build failed. Zoffix Znet 'Fix markdown error' 10:54
travis-ci.org/perl6/nqp/builds/326009408 github.com/perl6/nqp/compare/bff31...66a65cbacb
Zoffix t/concurrency/03-semaphore.t ........... Failed 1/12 subtests
Seems to flop
pmurias Zoffix: thanks for implementing chainstatic on js backend ;) 11:13
Zoffix :0 11:14
*:)
I think I'm gonna put the alternative-coercer thing, where we try Target.new(Source) if Source.Target() fails, on backburner. using nqp::handle op to try .new keeps segfaulting, but even were it to work, nqp::handle, would make the operation 3x slower. And if you just let it explode, like we do with Source.Target() right now, you get these awful errors: $ ./perl6 -e 'class Foo {}; -> Foo() $x { $x.say 11:23
}("42")' 11:24
Default constructor for 'Foo' only takes named arguments
Geth nqp/master: 9 commits pushed by pmurias++ 11:27
travis-ci NQP build failed. pmurias '[js] Add a bunch of NFG using op variants' 11:40
travis-ci.org/perl6/nqp/builds/326017730 github.com/perl6/nqp/compare/2666a...ca7acc8ca1
Geth nqp: 247c7a022d | pmurias++ | t/moar/02-qast-references.t
Fix test plan
 11:44
Zoffix And checking result type of coercion too, not gonna do right now. Just tried this: gist.github.com/zoffixznet/6d0a208...8d15547729 and it made `-> Str(Foo) $x { }(Foo.new)` 8% slower. Just not worth it ATM, as it'd cover 0.0001% of usecases 11:51
7am... time for bed 11:53
travis-ci NQP build passed. pmurias 'Fix test plan' 11:59
travis-ci.org/perl6/nqp/builds/326021182 github.com/perl6/nqp/compare/53ca7...7c7a022d6b
pmurias do we want to read unassigned windows-1251 characters as unicode control characters with that value? 12:22
pmurias doesn't know what's the correct way to handle technology as crappy as windows-1252 12:34
s/windows-1251/windows-1252/
Geth nqp: f6bccf79d8 | usev6++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java
[jvm] Look inside containers for parametric type

Fixes some spectests (cmp. RT #132673).		 12:40
synopsebot RT#132673 [new]: rt.perl.org/Ticket/Display.html?id=132673 [JVM] Container has no information about parametric type
Geth roast: 0dc257e726 | usev6++ | S02-magicals/KERNEL.t
[jvm] Adjust fudging for $*KERNEL.signal
 12:53
[Tux] Rakudo version 2017.12-133-g880dbfaa1 - MoarVM version 2017.12-22-gc2de234c4
csv-ip5xs1.115 - 1.143
csv-ip5xs-2013.256 - 13.873
csv-parser12.870 - 13.750
csv-test-xs-200.452 - 0.500
test11.673 - 11.693
test-t2.855 - 2.920
test-t --race1.224 - 1.249
test-t-2052.690 - 53.676
test-t-20 --race17.670 - 18.762
 13:06
lizmat pythonclock.org # intriguing concept 16:11
pmurias it's interesting how long will v6.c be supported after v6.d is out 16:18
Geth rakudo/js: 4accb64148 | pmurias++ | src/core/Str.pm
Make lines work correctly with \r\n on backends that don't implement NFG
 16:32
pmurias bartolin: 4accb64148ee49580cb15caa276909a20997491b - should also fix Str.lines on the JVM backend 16:33
anyone here has any experience with windows-1252 and opinions how dealing with unassigned characters in it should work? 16:49
timotimo probably should be able to configure a replacement character for the encoder? 16:52
don't think we support that anywhere yet? perhaps? 16:53
pmurias checks 16:56
Geth rakudo: b400707035 | (Elizabeth Mattijsen)++ | src/core/Any-iterable-methods.pm
.head/.tail should take any Callable

This was probably left over from an attempt to assign different semantics for WhateverCode in this place.		 16:57
pmurias timotimo: we do, but moar just assigned control characters with the same value to the unassigned characters 16:59
timotimo: so the real question are they something that is so unassigned that we should throw an exception when it's used or it's just something that can be used in an adhoc manner 17:09
timotimo right, when encoding into such a limited format we don't have anything sensible 17:19
bartolin pmurias++ # 4accb64148 17:20
Geth rakudo/js: 2ce329cada | pmurias++ | 3 files
[js] Use NFG using ops in a bunch of places
 17:35
pmurias m: say "\x[FFFFFF]" 17:36
camelia Error encoding UTF-8 string: could not encode codepoint 16777215 (0xFFFFFF), codepoint out of bounds. Cannot encode higher than 1114111 (0x10FFFF)
in block <unit> at <tmp> line 1
pmurias m: my $foo = "\x[FFFFFF]"; 17:37
camelia ( no output )
pmurias do we want to support \x[FFFFFF] in literals?
Geth rakudo: dbea11ccde | (Elizabeth Mattijsen)++ | 2 files
Simplify the ".skip should get a new iterator" issue

A much better fix than 854c10c27e49041c87841 was.		 17:41
travis-ci Rakudo build errored. Elizabeth Mattijsen '.head/.tail should take any Callable 18:14
travis-ci.org/rakudo/rakudo/builds/326086708 github.com/rakudo/rakudo/compare/8...0070703551
buggable [travis build above] ☠ Did not recognize some failures. Check results manually. Across all jobs, only t/02-rakudo/08-slangs.t test file failed.
Geth rakudo: d034cb582a | (Elizabeth Mattijsen)++ | src/core/Any-iterable-methods.pm
Fix for GH #1349

Also handle .skip(*)		 18:18
synopsebot RAKUDO#1349 [open]: github.com/rakudo/rakudo/issues/1349 Skip using a WhateverCode doesn't work
pmurias jnthn: is rakudo.moar allowing "\x[FFFFFF]" intended or a bug? 18:33
it might be something we want to forbid in 6.d 18:35
timotimo releasable6: status 18:52
releasable6 timotimo, Next release in 13 days and ≈0 hours. No blockers. Unknown changelog format
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Simplify the ".skip should get a new iterator" issue 19:08
travis-ci.org/rakudo/rakudo/builds/326098237 github.com/rakudo/rakudo/compare/b...ea11ccdef7
buggable [travis build above] ☠ All failures are due to: failed make test (3 failures). Across all jobs, 3 unique test files failed.
dogbert17 m: say (1..5).map({$_}).skip(*-3)
camelia (3 4 5)
dogbert17 .oO
lizmat++
lizmat m: say 1..5 .skip(*-3) 19:09
camelia Seq objects are not valid endpoints for Ranges
in block <unit> at <tmp> line 1
lizmat m: say (1..5) .skip(*-3)
camelia (3 4 5)
lizmat no need for .map :-)
dogbert17 yeah, it was just an example from the docs :-) 19:10
so you figured out a nice way to solve this it seems
pmurias jvm-- # it only supports a single character error replacement while Perl 6 allows a multi character one :( 19:57
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Fix for GH #1349 20:00
travis-ci.org/rakudo/rakudo/builds/326106684 github.com/rakudo/rakudo/compare/d...34cb582a8d
buggable [travis build above] ☠ Did not recognize some failures. Check results manually. Across all jobs, only t/04-nativecall/04-pointers.t test file failed.
Geth rakudo: b4df753dfc | (Elizabeth Mattijsen)++ | 5 files
Introducing EmptySeq constant

  - The Empty Sequence (we only need one of them)
  - use EmptySeq in setting where possible
Empty sequences in Any-iterable-methods and Str where too soon in the setting. Please revert if having an EmptySeq constant is a bad idea.		 20:07
nqp: 5b80052cf4 | usev6++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java
[jvm] Remove dead code

All of those methods have been replaced with a version that honors ThreadContext.		 20:17
pmurias bartolin: I'm working on implementing nqp::encoderep on the JVM (and JS) 20:27
bartolin pmurias++ # sounds great (I had to lookup first, what that op does :-) 20:35
bartolin will be short of tuits for the next weeks, since he's moving house 20:36
lizmat wishes bartolin a lot of strength in the chaos that will develop 20:39
timotimo bartolin: i feel you :S 20:40
still haven't got a target for the move
bartolin at least, my target is only a few hundred meters away. but still ... :-) 20:41
thanks, lizmat 20:42
pmurias bartolin: I hope the new house ends up awesome :) 20:57
jnthn lizmat: Will EmptySeq not introduce a discontinuity on the "can only iterate it once" rule? 21:05
lizmat no, because it has its own "iterator" method that doesn't check
jnthn m: my \s = Seq.new; for s { }; for s { } 21:07
camelia Too few positionals passed; expected 2 arguments but got 1
in block <unit> at <tmp> line 1
jnthn m: my \s = ().Seq; for s { }; for s { } 21:08
camelia The iterator of this Seq is already in use/consumed by another Seq
(you might solve this by adding .cache on usages of the Seq, or
by assigning the Seq into an array)
in block <unit> at <tmp> line 1
jnthn m: my \s = EmptySeq; for s { }; for s { }
camelia ( no output )
jnthn That's what I mean
lizmat ah, so the lack of error is what is the problem ?
ok, I'll revert then 21:09
jnthn: you seemed to expect that Seq.new() would give you an empty Seq 21:10
shall I implement it like that then ?
as a shortcut for Seq.new(Rakudo::Iterator.Empty)
jnthn No, I don't really think it should give one, I'm just a bit too full of curry and beer to really think :) 21:14
Well, only the one beer, so I think food coma is the main issue here ;)
I'm not outright sure it should be reverted
Just pointing out that it results in a Seq that behaves unlike others
Which is at least worth pausing to think about a little 21:15
(My main worry being, will this edge case ever cause somebody's tests to pass by not exploding, when they would not do so on some real data?)
lizmat yeah, see your point 21:16
I think your point is valid, so I will rework this to having Seq.new return an empty Seq
jnthn bbiab 21:17
lizmat so we can at least get rid of a lot Seq.new(Rakudo::Iterator.Empty)
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Introducing EmptySeq constant
travis-ci.org/rakudo/rakudo/builds/326132991 github.com/rakudo/rakudo/compare/d...df753dfcd4
buggable [travis build above] ☠ Did not recognize some failures. Check results manually. Across all jobs, only t/04-nativecall/01-argless.t test file failed.
pmurias bartolin: I have started to pass some NFG tests on the js backend by having separate (more expensive) NFG aware ops, I'll experiment with that approach more on the JS backend and if it works out well we might port it to the JVM one 21:48
bartolin \o/ 21:50
timotimo hm, try compiling moar's string implementation and ops to js with emscripten for the time being? :D 21:56
pmurias timotimo: making nqp ops NFG aware isn't that hard 22:13
timotimo: doing it fast is
timotimo right
that's where asm.js comes into play :P 22:14
pmurias asm.js/webasm aren't really built for freely mixing it up with normal js 22:22
timotimo: nqp.js code performance is what will need to be addressed first (as that slows development and rakudo.js is written in NQP) so I'm super hestitant to slow that with NFG stuff 22:28
timotimo right, that's fair 23:07

Powered by Cro and the Raku® Programming Language.

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

Thank you!