buggable ??? It's time for the monthly Accidental /win Lottery ??? We have 2 ballots submitted by 2 users! DRUM ROLL PLEASE!... 00:00
And the winning number is 42! Congratulations to Zoffix! You win a roll of duck tape!
AlexDaniel squashable6: status 03:35
squashable6 AlexDaniel, ⚠? Next SQUASHathon in 5 days and ≈6 hours (2017-10-07 UTC-12⌁UTC+14). See github.com/rakudo/rakudo/wiki/Mont...Squash-Day
nine .tell sjn IMHO named-anywhere is a hack that works around rather than fixes the issue. PR 688 replaces the hack by convention and still does not fix the whole thing. 04:24
yoleaux nine: I'll pass your message to sjn.
Zoffix goes back to rakudobrew on homebox 04:41
Turns out I do need multiple rakudo versions on occasion to debug stuff :) 04:42
nine Zoffix: well you are one of the elect few in rakudobrew's target audience after all :) 04:44
Zoffix :D
nine OMG it lives! 05:34
This will shave ~ 100 ms off csv-ip5xs' warmup time and even more for more complex Inline::Perl5 users and NativeCall users in general. 05:54
Zoffix \o/
nine That is...if I can fix the issue with the Staging repo this patch uncovers 05:55
Seems like with everything I do I will eventually hit the wall of rouge references breaking precompilation 05:58
Apparently the mere presence of an EXPORT sub triggers it 06:00
Nope, that's not it 06:02
It's that I moved this: my $repr_map := nqp::hash( "CArray", "carray", "CPPStruct", "cppstruct", "CPointer", "cpointer", "CStruct", "cstruct", "CUnion", "cunion", "VMArray", "vmarray",); 06:06
before the use statements in the file.
No wonder I hate this kind of bugs
Yep, just move the use statements to the top and everything's fine again 06:10
Zoffix m: say "xxxxx" ~~ /.**{1..0}/ 07:06
camelia 「xxxxx」
Zoffix m: say "xxxxx" ~~ /.**{2..1}/
camelia Nil
Geth rakudo/nom: 80d6b425ce | (Stefan Seifert)++ | lib/NativeCall.pm6
Compile optimized bodies of native subs at compile time

Previously the optimized bodies of native subs were compiled on the first call of the sub. NativeCall now exports a CHECK phaser to do this at the end of the compilation phase. This means that it happens e.g. during precompilation and can be done once per installation instead of once per run.
Note that loading the native library is still done at runtime. Also since we only know if we can JIT compile the calling code after nativecallbuild, we compile both the JITed and non-JITed body candidates and select the appropriate one in !setup.
nine lizmat: This ^^^ shaves ~ 100 ms off csv-ip5xs' warmup time and even more for more complex Inline::Perl5 users and NativeCall users in general. 07:07
Zoffix digs deeper down to see wtf it thinks 1..0 means inf :/ 07:09
nine lizmat: saves 21 % of Inline::Perl5's test suite's CPU time
Zoffix :o 07:10
m: say "xxxxx" ~~ /.**2..0/ 07:11
camelia ===SORRY!=== Error while compiling <tmp>
Empty range
at <tmp>:1
------> say "xxxxx" ~~ /.**2..0/
Zoffix hm, OK. throwing is easier ^_^
nine Boy am I glad that this work turned out to actually clearly improve things :) 07:13
Zoffix \o/ 07:14
travis-ci Rakudo build failed. Stefan Seifert 'Compile optimized bodies of native subs at compile time 07:56
travis-ci.org/rakudo/rakudo/builds/281845278 github.com/rakudo/rakudo/compare/4...d6b425ce0f
buggable [travis build above] ☠ Did not recognize some failures. Check results manually.
Zoffix 1 job: Stage parse : MoarVM panic: Internal error: invalid thread ID 14475144 in GC work pass 08:06
nine I dare say my commit is innocent, since it touches only NativeCall.pm6 08:07
Thread ID 14475144 looks like memory corruption or a missing MVMROOT to me 08:08
Zoffix ZOFVM: Files=1275, Tests=152192, 149 wallclock secs (22.64 usr 3.88 sys + 3134.07 cusr 258.00 csys = 3418.59 CPU) 08:10
hm 08:14
This dies with "cannot use empty range" say ‘abcefghij’ ~~ / .**{10..5} / but this matches an empty string because endpoints <0 get treated as 0: say ‘abcefghij’ ~~ / .**{-10..-5} /
Wonder if second one should die too
I meant {-5..-10} for the second one 08:22
Geth rakudo/nom: 681d6be974 | (Zoffix Znet)++ | 3 files
Fix and improve `**` regex quantifier

Fixes RT#130911: rt.perl.org/Ticket/Display.html?id=130911 Fixes RT#130127: rt.perl.org/Ticket/Display.html?id=130127 Fixes RT#130125: rt.perl.org/Ticket/Display.html?id=130125 Fixes RT#130124: rt.perl.org/Ticket/Display.html?id=130124
... (11 more lines)
08:28
synopsebot RT#130911 [open]: rt.perl.org/Ticket/Display.html?id=130911 [REGEX] bogus range o ** { 1..0 } succeeds
synopsebot RT#130127 [new]: rt.perl.org/Ticket/Display.html?id=130127 [REGEX] Excluded endpoints in ranged quantifiers do not work at all (/ .**{2^..^2} /)
RT#130125 [new]: rt.perl.org/Ticket/Display.html?id=130125 [REGEX] negative numbers in ranges are understood incorrectly (/ .**{-10..-5} /)
RT#130124 [new]: rt.perl.org/Ticket/Display.html?id=130124 [LTA] NaN in quantifiers is not allowed, message complains about P6opaque (/ .**{NaN..NaN} /)
Geth roast: 99c822abde | (Zoffix Znet)++ | S05-metasyntax/regex.t
Unfudge and add more of ** regex quantifier tests

RT#130911: rt.perl.org/Ticket/Display.html?id=130911 RT#130127: rt.perl.org/Ticket/Display.html?id=130127 RT#130125: rt.perl.org/Ticket/Display.html?id=130125 RT#130124: rt.perl.org/Ticket/Display.html?id=130124 Rakudo fix: github.com/rakudo/rakudo/commit/681d6be974
08:29
Zoffix smiles at the fixed synopsebot not spamming seen tickets anymore o/ 08:30
Geth rakudo/nom: 07c1e4fc89 | (Zoffix Znet)++ | src/core/Match.pm
Add more info to the comment
08:49
lizmat . 08:54
yoleaux 04:07Z <AlexDaniel> lizmat: It's probably a good idea to mention SQUASHathon and Hacktoberfest in the next weekly. github.com/rakudo/rakudo/wiki/Mont...Squash-Day github.com/issues?utf8=%E2%9C%93&a...toberfest+ hacktoberfest.digitalocean.com/
04:09Z <AlexDaniel> lizmat: oh, and raw.githubusercontent.com/perl6/ma...-media.jpg
[Tux] MoarVM oops: Can't find nc_site value on spesh ins <nativeinvoke_o> 0
lizmat .ask timotimo why 4d0ead240aff4510e ? 08:55
yoleaux lizmat: I'll pass your message to timotimo.
[Tux] not in the main test file, but in the comparison tests
$ p6 csv-ip5xs.pl 08:56
MoarVM oops: Can't find nc_site value on spesh ins <nativeinvoke_o> 0
test-t first run was 3.251, and I cannot run the test beyond that currently 08:58
nine [TuxCM]: ah, I already fixed that in MoarVM. Needs a bump 08:59
|Tux|: can you re-test with current MoarVM or should I commit a bump? 09:04
[Tux] i'll retry 09:05
Still get the failure. Should I remove some build folder? 09:13
removed all precomp's. starting afresh 09:15
nine |Tux|: you're running with MoarVM 2017.09.1-62-g89ca8eb08? 09:16
[Tux] CSV6/lib/.precomp was massive :( 09:17
and I still have the nc_site fail 09:18
$ p6 --version 09:19
This is Rakudo version 2017.09-176-g07c1e4fc8 built on MoarVM version 2017.09.1-50-g3059ba28
implementing Perl 6.c.
# git_reference/MoarVM (master) 09:20
HEAD is now at 89ca8eb0 Make "Can't find nc_site value on spesh ins" non-fatal
# moar-blead-nom (nom)
HEAD is now at 07c1e4fc8 Add more info to the comment
# moar-blead-nom/nqp (master)
HEAD is now at 71fc322e5 Map the new nativeinvoke OP
# moar-blead-nom/nqp/MoarVM (master) 09:21
HEAD is now at 89ca8eb0 Make "Can't find nc_site value on spesh ins" non-fatal
travis-ci Rakudo build passed. Zoffix Znet 'Add more info to the comment' 09:56
travis-ci.org/rakudo/rakudo/builds/281861039 github.com/rakudo/rakudo/compare/6...c1e4fc895e
lizmat afk again& 10:13
timotimo lizmat: because it's not polite to output errors to stdout in a user program, i guess? it looks like you can just cause it by putting a BUILDALL method in your class along with an attribute, and it'll complain. 10:37
yoleaux 08:55Z <lizmat> timotimo: why 4d0ead240aff4510e ?
Geth rakudo/nom: f8a74eabae | (Zoffix Znet)++ | src/core/Exception.pm
Fix typo in Exception's message; MasterDuke++
12:13
rakudo/nom: 78c967cbd5 | (Zoffix Znet)++ | lib/NativeCall.pm6
Remove Perlism; MaserDuke++
12:14
nqp: a0618a6256 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
Bump MoarVM
12:39
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...-g89ca8eb
rakudo/nom: 921db910f1 | (Zoffix Znet)++ | tools/build/NQP_REVISION
Bump NQP
¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....-ga0618a6
travis-ci NQP build failed. Zoffix Znet 'Bump MoarVM' 12:57
travis-ci.org/perl6/nqp/builds/281901133 github.com/perl6/nqp/compare/71fc3...618a62560a
Zoffix "t/nqp/059-nqpop.t ...................... 1/333 java.lang.ClassCastException: org.perl6.nqp.sixmodel.reprs.P6strInstance cannot be cast to java.lang.String" 12:58
Geth rakudo/full-make-test: edce8f53bd | (Zoffix Znet)++ | 2 files
Add all Rakudo's tests to `make test`
13:20
Zoffix It's hard to work with my cat two feet away, staring at me... judging me: imgur.com/vhThv1R 13:22
ZofBot: that's why I love you more. You never stare... 13:23
ZofBot Zoffix, replace a with a, and it turns the utf8c8 graphemes into standard text which it should not. will be fixing that too
Zoffix Cool. Thanks for fixing that.
MasterDuke looks like it ran face first into an ink blotter
teatime I was gonna say it looks like Rorschach 13:25
Zoffix :)
timotimo one of our cats likes sitting between my keyboard and the monitors 13:44
Geth rakudo/full-make-test: c6ff787a67 | (Zoffix Znet)++ | 5 files
Don't rely on roast checkout for `make test`
13:53
[Tux] This is Rakudo version 2017.09-179-g921db910f built on MoarVM version 2017.09.1-62-g89ca8eb0 13:58
csv-ip5xs 1.224 - 1.315
test 10.168 - 10.228
test-t 3.367 - 3.527
csv-parser 12.785 - 13.012
Zoffix buggable: speed 4
buggable Zoffix, ▁▁▄█ dates: 2017-10-01–2017-10-01; range: 3.214s–3.527s; speed: 10% slower
Zoffix :(
MasterDuke but it looks like nine++'s recent work sped up csv-ip5xs, a quick look through the logs shows 1.260 as the next fastest 14:03
Zoffix :) 14:04
damn 14:27
We should consider [re]moving all roast files that we aren't running. Just spent time editing S24-testing/1-basic.t just to find out it's not in the spectest.data list -_
Geth roast: c16f1480b6 | (Zoffix Znet)++ | S24-testing/1-basic.t
Add warning about added tests to end of file
14:28
ugexe i've done that too 14:31
Geth 6.d-prep: fa97ad371d | (Zoffix Znet)++ (committed using GitHub Web editor) | TODO/README.md
Mention S24-testing/1-basic.t situation
14:33
rakudo/nom: 8479a1ba03 | (Zoffix Znet)++ | lib/Test.pm6
Make cmp-ok try harder to get useful description

Merely stringifying the $got/$expected sometimes results in warnings/empty descriptions in the failed description output. Make descriptions via .perl//.gist
14:35
roast: 1287ea5ab3 | (Zoffix Znet)++ | S24-testing/13-cmp-ok.t
Test cmp-ok provides useful desc of its args

Rakudo fix: github.com/rakudo/rakudo/commit/8479a1ba03
14:36
rakudo/nom: bbc6570f5d | (Zoffix Znet)++ | t/spectest.data
Add S24-testing/13-cmp-ok.t to list of test files
14:37
Zoffix ZOFFLOP: t/spec/S11-modules/nested.t 14:44
Geth rakudo/nom: 1b9638e2b5 | (Zoffix Znet)++ | src/core/List.pm
Make List.ACCEPTS non-fatal for lazy iterables

Since we tend to never die in our other ACCEPTS
14:59
roast: 62532fb591 | (Zoffix Znet)++ | S03-smartmatch/array-array.t
Spec smartmatch with lazy iterables

Rakudo fix: github.com/rakudo/rakudo/commit/1b9638e2b5
Zoffix m: say (""~~m/^/, 42, 50).ACCEPTS: 42 15:08
camelia (「」 42 50)
Zoffix ಠ_ಠ
I forget, what is this used for?
jnthn Hm, List is a smartmatch terminal for the sake of m:g mebbe? 15:12
Zoffix I guess it's specialcased during codegen, 'cause &infix:<~~> would Boolify it: github.com/rakudo/rakudo/blob/921d...Mu.pm#L862 15:15
m: $_= "abc"; &infix:<~~>(m:g/./, "abc").say
camelia False 15:16
Zoffix m: say "abc" ~~ m:g/./
camelia (「a」 「b」 「c」)
Zoffix OK
travis-ci Rakudo build passed. Zoffix Znet 'Make cmp-ok try harder to get useful description 15:26
travis-ci.org/rakudo/rakudo/builds/281926101 github.com/rakudo/rakudo/compare/9...79a1ba037b
Rakudo build passed. Zoffix Znet 'Add S24-testing/13-cmp-ok.t to list of test files' 16:07
travis-ci.org/rakudo/rakudo/builds/281926519 github.com/rakudo/rakudo/compare/8...c6570f5db7
AlexDaniel “The Unicode support is not "garbage", it's limited to the BMP.” 16:13
AlexDaniel flips the table
teatime heh url? I find bug-thread arguments soothing for some reason. 16:14
AlexDaniel teatime: github.com/perl6/marketing/issues/1 16:15
teatime :/ that was not super satisfying. 16:17
teatime makes you guys some herbal tea and heads over to the systemd bugzilla
Zoffix Turns out my ** fix has a bug. I found it only by writing docs and running code examples in them :P 16:19
jnthn AlexDaniel: Using words like "garbage" to describe something isn't likely to help motivate people trying to maintain/improve the thing in their lack of free time.
I'm sure there's a less loaded word you coulda used.
Geth rakudo/nom: 4ca1fc3c16 | (Zoffix Znet)++ | src/core/Match.pm
Fix regex `**` with non-Int values…

  …leaving behind unhandled Failures
16:21
roast: 98fec08b00 | (Zoffix Znet)++ | S05-metasyntax/regex.t
Test `**` quant does not leave unhandled Failures

Rakudo fix: github.com/rakudo/rakudo/commit/4ca1fc3c16
AlexDaniel there's a very soft alternative: “LTA”. This wouldn't've motivated me a slightest bit. 16:26
I'm not pointing out that there's a problem, everyone knows about it already. I'm pointing out that I can't stand this problem, it makes me angry. 16:27
Zoffix ZofBot: kill all humans and their petty arguments
ZofBot Zoffix, @sorted = sort +*, @unsorted; #ascending numeric @sorted = sort -*, @unsorted; #descending numeric OrderingPair A combination of the two methods above, for when one wishes to take advantage of the internal caching of keys that is e
Zoffix C'mon #perl6-dev is a sanctuary to nourish the hacking spirit. Move your debates to #perl6 :)
AlexDaniel
.oO( #perl6-rant )
16:28
[Tux] This is Rakudo version 2017.09-183-g4ca1fc3c1 built on MoarVM version 2017.09.1-62-g89ca8eb0 16:41
csv-ip5xs 1.214 - 1.215
test 9.761 - 9.841
test-t 3.278 - 3.347
csv-parser 12.733 - 12.922
nine Tomorrow, I'm gonna have ~ 3 hours drive and 10 hours flight worth of coding time :) I guess I'll try extending the NativeCall JIT compiler to more argument types. 16:52
Geth rakudo: skids++ created pull request #1177:
Fix some unspace parsing cases (RT#128462 and degenerate unspace)
16:53
synopsebot RT#128462 [open]: rt.perl.org/Ticket/Display.html?id=128462 [PARSER] Unspace after a sigil-less term causes parser error
travis-ci Rakudo build passed. Zoffix Znet 'Make List.ACCEPTS non-fatal for lazy iterables 16:55
travis-ci.org/rakudo/rakudo/builds/281931896 github.com/rakudo/rakudo/compare/b...9638e2b5dd
Geth rakudo/nom: 11070e0fee | skids++ | src/Perl6/Grammar.nqp
Fix some unspace parsing cases (RT#128462 and degenerate unspace)

see RT for analysis/explanation
17:00
rakudo/nom: f427575432 | (Zoffix Znet)++ (committed using GitHub Web editor) | src/Perl6/Grammar.nqp
Merge pull request #1177 from skids/rt128462

Fix some unspace parsing cases (RT#128462 and degenerate unspace)
roast: skids++ created pull request #340:
Additional unspace tests and unfudge RT#128462 test
17:01
roast: 3f81f39aae | skids++ | S02-lexical-conventions/unspace.t
Additional unspace tests and unfudge RT#128462 test
17:02
roast: c22107d532 | skids++ (committed using GitHub Web editor) | S02-lexical-conventions/unspace.t
Merge pull request #340 from skids/rt128462

Additional unspace tests and unfudge RT#128462 test
Zoffix teatime: "<teatime> does modules.perl6.org/ index CPAN-only modules?" 17:20
teatime: not yet. It was actually somethign I was planning to implement this weekend, but got distracted with compiler :)
teatime cool
Zoffix (though I'm off tomorrow, so... the weekend is still young :))
Yeah, I think I'll work on that next, after pizza lunch and jamming session 17:22
gfldex i wonder if a stacktrace should contain the compiler version 17:28
Geth roast: b9854b796e | (Samantha McVey)++ | 2 files
Update S15-nfg/emoji-text.t for Emoji v5.0
18:49
roast: d8b1178020 | (Samantha McVey)++ | docs/unicode-generated-tests.asciidoc
Add file for documenting Unicode generated test files

Contains information like:
  * List of Perl6 script used to update the tests
  * Roast Test File's generated
  * Version number the test was generated from
  * Date we last generated the file
  * File header of the test file, in case Unicode updates their test files
   (Unicode's files are all dated).
lizmat is back 18:50
hmmm...S32-io/open.t has started flapping for me 18:51
Geth rakudo/nom: 4fae0711ef | (Elizabeth Mattijsen)++ | src/Perl6/World.nqp
We actually don't need an nqp::stmts for accessors

This probably doesn't mean much in the scale of things, but it *does* happen at compilation for every public attribute.
19:04
samcv lizmat, doesn't seem to flop for me on 2017.09-153-g9cb4b167f though i need to update my rakudo 19:17
lizmat well, for me it bailed *once* in test 62... couldn't repeat it :-( 19:21
jnthn I'm sure I've seen it in the past, but don't recall seeing it for a long while.
samcv ah. i ran it like 250 times and it was ok each time
lizmat m: class A {}; role B { has $.a }; dd (A but B).new(a => 42) 19:26
camelia A+{B}.new(a => 42)
lizmat jnthn: ^^^ it appears that this case does *not* get a BUILDPLAN generated. Should it ?
BUILDALL I mean
aah... maybe because the class is already composed? 19:27
jnthn Hmm 19:31
Uh 19:32
It can't
It's too late
It happens at runtime, there's no compiler services object 'cus we're no longer in the compiler
Took me a moment to realize that :)
So yeah, I'd not expect it to
lizmat so that basically means we need to keep the Mu.BUILDALL around :-( 19:34
I found this issue in S02-literals/string-interpolation.t 19:36
it apparently creates X::Worry::P5::LeadingZero objects with X::Comp mixed in 19:37
Geth nqp: 289f79fbd2 | (Samantha McVey)++ | t/nqp/107-index.t
Add nqp::index tests for indexingoptimized on long string

Tests both with and without indexing optimized on a long repeated Haystack, with a concatenated needle.
19:41
lizmat afk for a bit 19:45
travis-ci NQP build failed. Samantha McVey 'Add nqp::index tests for indexingoptimized on long string 19:55
travis-ci.org/perl6/nqp/builds/282000418 github.com/perl6/nqp/compare/a0618...9f79fbd2d9
buggable New CPAN upload: Bailador-0.0.13.tar.gz by UFOBAT www.cpan.org/authors/id/U/UF/UFOBA....13.tar.gz 21:32
timotimo oh, that's neat 21:38
Zoffix :) 21:41
jnthn Nice :) 21:42
timotimo i've got to do a bit of work on json::fast and get it up on cpan, too
jnthn is planning to shift Cro releases to CPAN
timotimo you can trick JSON::Fast into generating faulty json by giving it an IntStr with a broken Str portion 21:44
hm, does json even support anything other than decimal? 21:45
ok, it does not
so i'm not taking a feature away from our users by just using the Int (or Rat or Num) portion in any case
buggable New CPAN upload: Pastebin-Shadowcat-2.001001.tar.gz by ZOFFIX www.cpan.org/authors/id/Z/ZO/ZOFFI...001.tar.gz 22:03
New CPAN upload: Number-Denominate-1.001001.tar.gz by ZOFFIX www.cpan.org/authors/id/Z/ZO/ZOFFI...001.tar.gz 22:26
New CPAN upload: Acme-Anguish-1.001002.tar.gz by ZOFFIX www.cpan.org/authors/id/Z/ZO/ZOFFI...002.tar.gz 22:36