[Coke] c: d56246d410 e 02:03
committable6 [Coke], ¦d56246d: «»
Geth rakudo/main: 4 commits pushed by (Nick Logan)++ 02:10
ugexe dunno if i'll finish this last one tonight
[Coke] ok, I'll do a run before I heead to bed 02:17
c: 99ca13d635 e
committable6 [Coke], ¦99ca13d: «Cannot find this revision (did you mean “c1fd673”?)»
[Coke] c: 99ca13d635 e 02:35
committable6 [Coke], ¦99ca13d: «Cannot find this revision (did you mean “c1fd673”?)»
releasable6 Next release in ≈2 days and ≈11 hours. 1 blocker. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 07:00
Geth nqp: patrickbkr++ created pull request #878:
fix empty args in Windows build dir nqp binary		 07:55
08:42 finanalyst joined
Geth nqp/main: 2251e3d897 | (Patrick Böker)++ (committed using GitHub Web editor) | tools/templates/windows/nqp-m-build.c
fix empty args in Windows build dir nqp binary

The respective fix in rakudo dd116c27a applies to nqp here as well.		 08:59
rakudo/main: bcb3ef7a18 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get empty args in Windows build fix

  patrickb++		 09:17
10:11 finanalyst left
lizmat done all the outstanding community modules PRs now afaik 11:53
an interesting breakdown of zef's performance: gitee.com/skyter10086/raku-pm/blob...ture.en.md 12:08
still one blocker left :-( github.com/rakudo/rakudo/issues/6714 12:17
adapted MoarVM::Profile, removed Blocker label from the issue 12:49
assuming it was the only module suffering from the issue 12:50
afk for a few hours& 12:52
[Coke] releasable6: next 13:09
releasable6 [Coke], Next release in ≈2 days and ≈5 hours. There are no known blockers. 479 out of 518 commits logged
[Coke], Details: gist.github.com/f1dc54799554a548ea...2c4e1fdc24
[Coke] kicking off a blin run at bcb3ef7a18 13:18
linkable6 (2026-09-24) github.com/rakudo/rakudo/commit/bcb3ef7a18 Bump NQP to get empty args in Windows build fix
[Coke] If more come in early, I'll just restart.
Meant to go last night, but fell asleep before the builds finished, sorry. :)
Geth rakudo: ugexe++ created pull request #6725:
RakuAST: give a FIRST or POST statement's declarations to its attach block		 14:09
ugexe i think we can make the BEGIN A.a thing work, but in general that pattern seems potentially fraught 14:16
for instance your example breaks if make `method a ...` into `multi method a ...` 14:17
so i think we can make A.a work, and the other instances give an error like "not yet composed" or some such
[Coke] f0a24041462ebd6979a86 14:54
linkable6 (2026-09-22) github.com/rakudo/rakudo/commit/f0a2404146 Make RakUAST default for core and programs
[Coke] so far, these have failed in latest run: Control DEBUG FindBin French List Mux P5__DATA__ Pod Polyglot Proc RakupodObject SelectiveImporting 14:55
all on the conversion to AST. 14:56
up to 43%
sorry, Control::Bail DEBUG FindBin French List::Allmax Mux P5__DATA__ Pod::Perl5 Polyglot::Regexen Proc::Feed RakupodObject Regex::FuzzyToken SQL::Lexer SelectiveImporting Slang::Forgiven Slang::Predicate 15:13
(my awk-fu was bad) 15:14
ugexe yeah all of those either have PRs open or won't be fixed
or in other words: so far so good 15:15
[Coke] |o/ 15:16
lizmat fwiw P5__DATA__ should be considered dead.... 15:44
and I have half a mind to remove it from the REA, as that's the only place where that lives 15:45
15:48 Pixi` joined 15:49 Pixi left 15:59 Pixi` is now known as Pixi
lizmat I still think the Slang::Predicate issue is legit 16:03
see gist.github.com/lizmat/2b5dc79f3b4...1829eb549e 16:07
ugexe the machinery for adding aliases is wrong 16:20
i think it was you that added the => and one other binding alias to the grammar
but all the other aliases work because there are multi subs for them
lizmat hmm... that's a good point, lemme check that 16:21
ugexe i think a better solution wouldnt add anythign ot the grammar at all 16:23
the := variant is particularly problematic though 16:25
lizmat well, how would we handle precedence then?
ugexe i don't quite know what a better solution looks like overall. personally i'd remove the := and => aliases that were added until a better solution is actually designed / thought out. they can be made to work though -- see github.com/rakudo/rakudo/compare/m...y?expand=1 -- but i don't really care for that solution 16:27
(hence why i never PRd it)
[Coke] 78% done, 29 at f0a24041, 1 at bcb3ef7a, 1 at aab07787 16:48
lizmat bcb3ef7a, that's interesting, as you're not doing this blin run on Windows, right? 17:08
[Coke] linux, as always 17:30
Mux failed on that one 17:31
no failure output, though? Weird.
We'll see where we end up when the run is done.
(94%) 18:15
99.36, nearly done 18:51
The last 5 modules include a bunch of big, slow ones with lots of bisects. :| 19:39
Geth rakudo/main: 5 commits pushed by (Nick Logan)++ 19:40
20:01 finanalyst joined
ugexe m: my $a; for 1..10000 { $a *= 1.1 }; say $a 20:06
camelia Inf
ugexe m: my $a; for 1..10000 { $a *= 2 }; say $a
camelia 19950631168807583848837421626835850838234968318861924548520089498529438830221946631919961684036194597899331129423209124271556491349413781117593785932096323957855730046793794526765246551266059895520550086918193311542508608460618104685509074866089624888…
ugexe thats not new, just surprising to me
[Coke] ooof. 10000 Fatat multiplications are slow 20:15
Geth rakudo: ugexe++ created pull request #6726:
RakuAST: add a method to its package as soon as it is declared		 20:41
[Coke] github.com/coke/raku-blin-release-results is up to date - missing the data.json file 20:56
47 failures. 20:57
ugexe i just skimmed it and i think those are all either 1) have a PR open to fix it 2) have merged such a PR but not made a release 3) we won't fix. the MCP distribution was the only one left unfixed, but that is fixed in a PR i merged in rakudo today 21:01
[Coke] c: c023c4a531 e 21:43
committable6 [Coke], ¦c023c4a: «»
[Coke] I can do a run through that whenever - any more commits coming today? 21:45
lizmat not from me 22:20
ugexe there are a couple of other problems with the unicode variants. one is that it requires adding checks everywhere we check for e.g. $op eq '*', so for instance most optimizations probably are ignored for unicode variants. another problem (somewhat related to that) is that every time a variation is added we have to add it in a bunch of different spots all over the code base. and a third issue is no 22:34
one has decided what happens when a user overrides such a variant or a variant's parent (i.e. does overriding × also automatically override *? and vice versa)
if we treat × and * as exactly the same it is probably cleanest, but it would break any distributions that declare a multi already with both of those (since dispatch would then be ambiguous) 22:35
[Coke]: github.com/rakudo/rakudo/pull/6726 for sure, just waiting on ci
Geth rakudo/main: 30ebd72ce0 | (Nick Logan)++ | 4 files
RakuAST: add a method to its package as soon as it is declared

Previously the methods of a package were collected while its body was parsed and only added to the meta-object when the package was composed. As such BEGIN time code later in the body, e.g. `constant X = A.a` or
  `BEGIN A.a`, could not find a method declared right above it, and
  `A.^methods` came back empty. The legacy frontend adds each method as it
... (24 more lines)		 22:51
rakudo/main: 3fefa055fb | (Nick Logan)++ | 2 files
Say a class is not composed yet when a method is not found on it

Previously calling a missing method on a class that declares no parent, before the class was composed, gave `Method A.a not found`. Building the usual message died, since the invocant accessor calls `.item` on the class and it does not get its default parent until it is composed.
This reads the invocant attribute directly, and for any class that is not composed yet adds a tip that its multi methods and the methods from its roles or its default parent are not available yet. A class that declares a parent keeps its suggestions alongside the tip.
rakudo/main: eb6d4be9ff | (Nick Logan)++ (committed using GitHub Web editor) | 5 files
Merge pull request #6726 from ugexe/ugexe/rakuast-eager-method-attach

RakuAST: add a method to its package as soon as it is declared
22:52 finanalyst left
22:52 nine left, nine joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!