Geth rakudo/full-make-test: 6256ec8272 | (Zoffix Znet)++ | 2 files
Make repl tests force interactive repl mode
00:08
gfldex m: my @a = 1..500; @a.=hyper.map(* * 2); say [+] @a; 00:09
camelia 0
gfldex looks like .hyper has stopped working
Zoffix It never did. 00:10
It's not yet implemented. 6guts.wordpress.com/2017/03/16/con...semantics/ 00:11
gfldex it did about a week ago, as long as there was no recursion
Geth rakudo/full-make-test: e72dedf5cb | (Zoffix Znet)++ | t/02-rakudo/repl.t
Simplify routine call
00:13
gfldex it still works in some spots it seams 00:14
Geth rakudo/full-make-test: 4358871873 | (Zoffix Znet)++ | 2 files
Add scrubber routine to is-run-repl
00:22
rakudo/full-make-test: 92bc011ab0 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Follow convention of core multi sub declarations
00:24
rakudo/full-make-test: 63643ad042 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Friendlify is-run-repl()

  - If code is a Positional, assume its separate lines to feed REPL with
  - If out/err tests are Positionals, assume they're a bunch of lines
00:33
rakudo/full-make-test: 9c4025f33f | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Improve is-run()

Default to empty Str test for err/out and 0 exit code
00:34
BenGoldberg just noticed a very silly limitation in ecosystem module Inline. If a thousand and one instances of perl6 attempt to create a C function named, say, 'foo', then the thousand-and-first will hang forever (or at least, until one of the earlier thousand perl6s exits). 00:47
Geth roast: bitrauser++ created pull request #344:
Add tests for RT#125466
00:49
synopsebot RT#125466 [open]: rt.perl.org/Ticket/Display.html?id=125466 [MATH] bitwise shift is inconsistent on int
BenGoldberg The problem line is: $name = $basename ~ 1000.rand.Int while $name.IO.e || "$name$o".IO.e || "$name.c".IO.e;
Zoffix heh :) 00:51
BenGoldberg $basename is roughly "$*TMPDIR/inline_{$subroutine.name}". The sensible solution would be to incorporate into the filename $*PID and a SHA1 hash of the C code being compiled. 00:59
Zoffix
.oO( or just use Temp::Path )
01:01
BenGoldberg (A) that would be cheating ;), (B), by ommiting the $subroutine.name, it makes the code work with anonymous subroutines, (C) if you've got two or more functions with the *exact same* C code body, it doesn't make sense to compile it more than once -- just reuse the compiled code. 01:05
Geth rakudo/full-make-test: 71a0ad5528 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Rewrite Positional $code handling

The multi setup isn't working because non-Positional one takes precedence when desc is given
rakudo/full-make-test: 2f0bb20e82 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
diag() received output when Callable test fails

Otherwise we can't see why exactly it failed
01:07
rakudo/full-make-test: 153f133bc2 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Implement Map test to test specific lines

  - Keys are line numbers and values are expected lines
  - Key "t" is total expected line numbers
02:00
rakudo/full-make-test: be4d57deb5 | (Zoffix Znet)++ | t/02-rakudo/repl.t
Start rewriting REPL tests using new test routine

Tossed sanity tests (what's the point?) and two TODOed tests for postfix if/for on next lines. Such behaviour is undesirable, since we'll have to force users to use a semicolon at the end, having unhelpful behaviour of expecting further lines of input when it's missing.
02:15
travis-ci NQP build canceled. Samantha McVey 'Add a removal message if ornate parentheses are used 02:56
travis-ci.org/perl6/nqp/builds/284315409 github.com/perl6/nqp/compare/02a42...6d78eef877
Geth rakudo/full-make-test: c45d0cd1b5 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Disable error colouring
03:04
rakudo/full-make-test: 338a097246 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Allow :out test given as third positional
03:31
rakudo/full-make-test: 7c8a2739a5 | (Zoffix Znet)++ | t/02-rakudo/repl.t
Rewrite remaining REPL tests using new routine
03:32
rakudo/full-make-test: d82c09b351 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Toss `Map` test option

Too convoluted and turned out to be rather useless afterall.
03:33
samcv . 03:46
yoleaux 7 Oct 2017 16:33Z <Zoffix> samcv: Do tests need to be adjusted to accomodate ornate stuff? I'm getting failures in t/spec/S02-lexical-conventions/comments.t
7 Oct 2017 16:44Z <Zoffix> samcv: nevermind. Found the commit with reasonings and tossed the incorrect tests from master and 6.c-errata in github.com/perl6/roast/commit/85b9380acd and github.com/perl6/roast/commit/d78ab87a65
samcv Zoffix++ i must have missed that. thanks! 03:47
Geth rakudo/full-make-test: c9913277c9 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Consistify interface

By allowing Seq :out/:err tests, since we allow Seqs for $code
rakudo/full-make-test: f8edb82957 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Document test helpers
rakudo: zoffixznet++ created pull request #1187:
Full make test
03:48
roast: 447ceef0d2 | Joel++ (committed by Zoffix Znet) | S03-operators/bit.t
Add tests for RT#125466 (#344)

  * Add tests for RT#125466
  * Bump `plan`
  * Add missing types and "todo fudge" unclear cases
04:56
synopsebot RT#125466 [open]: rt.perl.org/Ticket/Display.html?id=125466 [MATH] bitwise shift is inconsistent on int
roast: 6aa3422b63 | (Zoffix Znet)++ | S03-operators/bit.t
Change todo fudges to skip

So they don't get reported on stresstests as passing todos
04:58
rakudo/nom: 456358e3c3 | (Zoffix Znet)++ | src/core/array_slice.pm
Make @a[42..*] 4.2x faster; AlexDaniel++ for persistence

  - Faster with any Inf .max Range
  - Faster only with non-lazy @a (we can't .elems lazies)
  - Fixes RT#125344: rt.perl.org/Ticket/Display.html?id=125344
06:06
synopsebot RT#125344 [open]: rt.perl.org/Ticket/Display.html?id=125344 [PERF] Int..Whatever ranges are slow (~20 times slower than Int..Int)
Geth roast: bitrauser++ created pull request #345:
Add tests for RT#132083 | [REGRESSION] Broken math (-2147483648 != -2147483648)
06:30
synopsebot RT#132083 [new]: rt.perl.org/Ticket/Display.html?id=132083 [REGRESSION] Broken math (-2147483648 != -2147483648)
Geth roast: b2a148d684 | bitrauser++ | S03-operators/arith.t
Add tests for RT#132083
06:54
roast: 64227ec7a2 | Joel++ (committed using GitHub Web editor) | S03-operators/arith.t
Add tests for RT#132083 (#345)

Add tests for RT#132083 | [REGRESSION] Broken math (-2147483648 != -2147483648)
synopsebot RT#132083 [resolved]: rt.perl.org/Ticket/Display.html?id=132083 [REGRESSION] Broken math (-2147483648 != -2147483648)
Geth rakudo/awesome-malformed-loops: a757328bdd | (Aleks-Daniel Jakimenko-Aleksejev)++ | src/Perl6/Grammar.nqp
Awesomify “malformed loop spec” errors

This is a questionable fix for RT #127857.
Makes each parse of a loop construct 0.04 ms slower. So if you have one hundred 「loop」-s in your code, that's already about four milliseconds slower.
Also breaks one 6.c-errata test that checks for “loop spec” text.
07:48
synopsebot RT#127857 [new]: rt.perl.org/Ticket/Display.html?id=127857 [RFC] [LTA] error message when using commas in c-style “loop” (loop (my $x = 0, $x < 10, $x++) {})
rakudo: AlexDaniel++ created pull request #1188:
Awesomify “malformed loop spec” errors
07:49
roast: bitrauser++ created pull request #346:
Add tests for RT#125215 | [BUG] 10 ** -1 is a Num, but should probably be a Rat since it can in Rakudo
07:56
synopsebot RT#125215 [open]: rt.perl.org/Ticket/Display.html?id=125215 [BUG] 10 ** -1 is a Num, but should probably be a Rat since it can in Rakudo
lizmat Files=1227, Tests=75630, 356 wallclock secs (14.69 usr 5.14 sys + 2510.32 cusr 218.01 csys = 2748.16 CPU) 07:59
Geth roast: a55c0f4ae5 | Joel++ (committed using GitHub Web editor) | S32-num/rat.t
Add tests for RT#125215 | [BUG] 10 ** -1 is a Num, but should probably be a Rat since it can in Rakudo (#346)

  * Add tests for RT#125215
  * Bump `plan`
08:12
synopsebot RT#125215 [resolved]: rt.perl.org/Ticket/Display.html?id=125215 [BUG] 10 ** -1 is a Num, but should probably be a Rat since it can in Rakudo
Geth rakudo/nom: 8300844303 | (Zoffix Znet)++ | 2 files
Do not explode in IO::CatHandle.nl-out

Doing so breaks .Capture, which is more important than throwing in a feature clearly documented as NYI
08:20
buggable New CPAN upload: App-MPD-AutoQueue-0.1.0.tar.gz by TYIL cpan.metacpan.org/authors/id/T/TY/...1.0.tar.gz 08:42
New CPAN upload: App-MPD-AutoQueue-0.1.1.tar.gz by TYIL cpan.metacpan.org/authors/id/T/TY/...1.1.tar.gz 09:12
New CPAN upload: perl6-App-MPD-Notify.tar.gz by TYIL cpan.metacpan.org/authors/id/T/TY/...ify.tar.gz 09:22
lizmat Zoffix: seeing two fails in t/02-rakudo/repl.t : Failed tests: 6, 28 10:57
is that to be expected ?
in test 6, stdout gives '' instead of '>' 10:59
on test 28, stdout is also not ok, but not clear what it was or what it should be
afk for a while& 11:30
buggable New CPAN upload: IO-Prompt-0.0.2.tar.gz by WBIKER cpan.metacpan.org/authors/id/W/WB/...0.2.tar.gz 12:42
stmuk PASS for me on linux (although it didn't yesterday!) 13:06
buggable New CPAN upload: Math-Model-0.1.tar.gz by MORITZ cpan.metacpan.org/authors/id/M/MO/...0.1.tar.gz 14:32
New CPAN upload: Math-RungeKutta-0.1.tar.gz by MORITZ cpan.metacpan.org/authors/id/M/MO/...0.1.tar.gz
Geth rakudo/nom: 775c367f34 | (Zoffix Znet)++ | t/02-rakudo/repl.t
Fudge OSX-failing tests

REPL output differred slightly on OSX for ages and we already had fudged tests for that reason. Fudging a couple more of these new ones.
15:27
Zoffix lizmat: thanks. Fudged them. 15:28
Geth rakudo/nom: 12fcece494 | (Elizabeth Mattijsen)++ | src/core/Hash.pm
Fix for RT #132246
16:22
synopsebot RT#132246 [new]: rt.perl.org/Ticket/Display.html?id=132246 .pick and .roll return incorrect results for object hashes 16:23
Zoffix I'm nailing down what .Capture does for core types. Stuff marked with "new" (🆕) is behaviour that's different from current one. Here's the full list gist.github.com/zoffixznet/548850d...04bc033113
Gonna impl later today
lizmat afk again& 16:29
dogbert2 bisect: use NativeCall; await start { my $a = "dubba"; explicitly-manage($a); } xx 15 17:01
bisectable6 dogbert2, Bisecting by exit code (old=2015.12 new=12fcece). Old exit code: 1
dogbert2, bisect log: gist.github.com/df465fe767df61ec6a...890d2631c7
dogbert2, There are 13 candidates for the first “new” revision. See the log for more details
[Coke] buggable's new cpan upload has duplicated info - maybe just show the full URL? 17:03
Zoffix It's name-ver => url. It's hard to see the name of the dist from URL only. 17:11
ZofBot: we don't pay character, so what's it matter 17:12
ZofBot Zoffix, pick" instead
buggable New CPAN upload: Math-Model-0.2.tar.gz by MORITZ cpan.metacpan.org/authors/id/M/MO/...0.2.tar.gz 19:22
[Tux] This is Rakudo version 2017.09-350-g12fcece49 built on MoarVM version 2017.09.1-576-g64b5dc03 19:47
csv-ip5xs 1.219 - 1.240
test 9.951 - 10.055
test-t 3.318 - 3.336
csv-parser 0.875 - 0.992
CSV::Parser still br0ken
japhb [Tux]: Broken how? 19:54
buggable: speed :4 19:55
buggable japhb, ▂▆ ▅ ▆↑↑█ ▄ ▇ dates: 2017-09-22–2017-10-08
japhb, ▄▆▁▂▃▄▄▆██▅█▅ ████ █ █ range: 3.154s–4.010s
japhb, █████████████▄████▄ ██▂▆ ▃█▂ ▁ ▁▂▅ ▆▄ speed: 3% faster
japhb, ███████████████████▇▃▄▇▄▇▄▅▆████████▂▁▃▅█▅███▇▆▅██
japhb Is there any way to tell that a particular unicode codepoint is preferred to render as full-width emoji instead of a narrow monochrome glyph? (Recognizing that due to the vagaries of fonts, this may not be the *actual* rendering) 20:01
It's the preference or default that I'm looking for. 20:02
buggable New CPAN upload: Math-Model-0.3.tar.gz by MORITZ cpan.metacpan.org/authors/id/M/MO/...0.3.tar.gz
ugexe windows doesn't pass make test anymore 20:22
jdv79 there's no real free windows vm, right?
i found one that expires in december which is ok i guess
jnthn Zoffix: Don't break Rat (and friends) and Complex destructuring; those are totally reasonable (and useful) behaviors that fall straight out of the default "objects destructure to their properties". I've used them multiple times in talks as a basic example of signature unpacking. 20:24
Zoffix: I'm not sure your proposed change to Range is an improvement either, though could argue that one either way. 20:25
Zoffix: Rest look OK to me
Wait, why doesn't Complex destructure already to im/re? :S 20:26
m: (2 + 3i).Capture.perl.say
camelia \(:im(3e0), :re(2e0))
jnthn oh, it does, the gist is just wrong :)
Was gonna say, I'm sure I've used that as an example too 20:27
cschwenz is the following a known issue? «make: *** No rule to make target `/Users/cal/local/bin/../share/nqp/lib/MAST/Ops.nqp', needed by `gen/moar/stage1/MASTOps.moarvm'. Stop.» (this issue was encountered while doing a build of rakudo from a fresh git clone.) 20:29
jdv79 doesn't look terribly familiar 20:31
cschwenz :-/
jdv79 what platform and how did you configure? 20:32
cschwenz one sec, making gist showing more detail…
jdv79 wonder if my puny rpi will compile rakudo these days 20:35
its only got 512M ram and 100M swap. i'm gonna predict its gonna OOM. 20:36
cschwenz jdv79: gist.github.com/cschwenz/e1df3cb05...3dbeb7d93b
the above is what i see in my shell when i attempt to build rakudo 20:37
:-(
jdv79 but on a single 700M cpu it might take a while
ugexe i suspect its related to --prefix somehow 20:39
jdv79 you could try the simplest configure to whittle it down - something like "perl Configure.pl --gen-moar" 20:40
cschwenz okay, trying that...
ugexe i wonder if --prefix isn't meant to be used with --gen-moar or --gen-nqp 20:41
Zoffix m: say <42+i>.^name 20:42
camelia Str
cschwenz so far, «perl Configure.pl --backends=moar --gen-nqp --gen-moar» appears to work
Zoffix Ah, I thought I had a complex there
cschwenz ugexe, —prefix used to work :-/
jdv79 i think various people pkg for distros and stuff - maybe look at how those do it? 20:43
Zoffix The current Range.Capture behaviour is useless because it doesn't include .excludes-min/.excludes-max attributes. Maybe they could be made public
ugexe cschwenz: maybe it doesn't work *if* you already have an installation at the --prefix location then 20:44
(assuming you even had one)
cschwenz i had one, but removed it trying to track down an unrelated issue :-( 20:45
Zoffix hmm
m: < 1/2 >.Capture
camelia ( no output )
Zoffix m: < 1/2 >.Capture.say
camelia \(:denominator(2), :numerator(1))
Zoffix ok, that works out
jdv79 looks like fedora does rakudo, nqp, and moar seperately 20:46
cschwenz jdv79, so the build order to do them seperately is: 1) moar, 2) nqp, 3) rakudo? 20:47
jdv79 should be, right? 20:48
cschwenz i think so, just never done it that way, and asking for confirmation. :-) 20:49
jdv79 i've not tried. i just build all three at once with no supplied prefix
Zoffix OK, here's updated gist with the suggestions incorporated into it: gist.github.com/zoffixznet/548850d...04bc033113