[Coke] 18% 00:13
(I said 5 hours, but it's more like 8, oops) 00:24
ugexe probably worth trying to figure out which ones take the most time to test at some point, although i imagine they are just things with a whole lot of tests 00:27
[Coke] can probably get that to be part of the status report at the end. 00:30
30% 00:41
60% 01:59
(can't fudge on OS) guessing it just needs a fudge tweak 02:01
02:21 spimmo joined 03:04 spimmo left
[Coke] 85%. heading to sleep 03:42
ugexe as my going to bed thought i do wonder if we should wait another week given today seems to be one of the first times anyone has outside of the core has tried to test their stuff with rakuast and some of those things already have fixes available. on the other handle it'll probably just be the same situation next week, just different problems and fixes :) 03:50
03:51 spimmo joined 03:52 spimmo left 03:53 spimmo joined
ShimmerFairy I don't think there's getting around some level of breakage as people get swapped over to rakuast for the first time, no matter how long you wait. About the only thing I could think of doing about it, if it was *truly* necessary, is a temporary modification to compiler errors to mention rakuast as a potential cause. 03:53
03:53 spimmo left
ShimmerFairy But I think if the ecosystem tests say it's mostly fine, then people will have a mostly seamless experience themselves. 03:54
03:54 spimmo joined 03:55 spimmo left 06:01 kjp left 06:20 kjp joined 08:58 finanalyst joined 09:13 finanalyst left 10:09 finanalyst joined 10:18 finanalyst left
[Coke] sorry, getting up early was not in the cards. 13:03
41 failures on latest blin run, putting them in the repo.. 13:04
github.com/coke/raku-blin-release-...ailures.md has the latest. 13:05
looks like some new, some cleared. 13:24
Geth rakudo/main: 50fffd091b | (Nick Logan)++ (committed using GitHub Web editor) | 2 files
RakuAST: sink the code of a sunk start (#6758)

Previously a start always compiled its code as wanted. A wanted while, until, repeat or loop is a lazy Seq.from-loop, and so is a statement with a while or until modifier. So `start until $c.closed { ... }` kept that Seq as the Promise result and the loop body never ran. S17-channel/stress.t hung on its bogosort block as a result. The same ... (8 more lines)		 14:07
ugexe i looked at the results, anything new was flakey
14:13 disbot left, disbot joined
[Coke] need that in the release branch. 14:19
a cherry-pick shouldn't be a problem with the merge back, should it?
guess actually I'm the only one on the release branch, I can rebase on master and force push 14:20
main, i mean 14:24
Geth rakudo/release-2026.09: 0487c338ad | (Will Coleda)++ | 3 files
Update changelog + announcement

Deliberately not logged:
  [561c5121][a0a60ed3][682fbc6b][79c849db][d1abb79b][12b5be61][b83057b1][ff413403][a107b0ce][a456c3a3]		 14:25
[Coke] restarting the nqp/rakudo release... 14:27
ah, crap. I need to also add that item to the changelog. 14:28
Geth rakudo/release-2026.09: 4e723f7d97 | (Will Coleda)++ | 3 files
Update changelog + announcement

Deliberately not logged:
  [561c5121][a0a60ed3][682fbc6b][79c849db][d1abb79b][12b5be61][b83057b1][ff413403][a107b0ce][a456c3a3]		 14:48
[Coke] todo passes (still) in rakudo make test. 15:22
Guessing with the cutover we can untodo those also
ab5tract makes sense to me 15:24
[Coke] ran the master stress test - failure on t/spec/6.c/MISC/bug-coverage-stress.rakudo.moar (Wstat: 256 (exited 1) Tests: 14 Failed: 1) 15:30
Failed test: 9
trying 6.c, 6.d and will retry master
6c is clear 15:33
6d fails t/spec/MISC/bug-coverage-stress.rakudo.moar (Wstat: 256 (exited 1) Tests: 12 Failed: 1) Failed test: 7 15:39
rerunning d,master 15:41
d ok on rerun - need to check this test to avoid flappiness 15:44
master passes on rerun 15:47
github.com/Raku/roast/issues/922 15:51
OK - release looks good here. I am paused at the step where I push the commits back to the repos. 16:02
(for nqp/rakudo)
Anyone have any concerns/questions?
I am on a call right now, will pull the trigger in an hour or so unless someone squawks.
anyone? 16:42
ok. pushing to github in about 8 16:53
ugexe 🤷 16:54
[Coke] m: say "🤷".uninames 16:58
camelia (SHRUG)
[Coke] (just making sure I wasn't missing some deep meaning there. :) 17:09
Geth nqp/main: 60175fc914 | (Will Coleda)++ | tools/templates/MOAR_REVISION
[release] Bump MoarVM revision to 2026.09
 17:18
nqp/main: 35344955ad | (Will Coleda)++ | VERSION
[release] Bump VERSION to 2026.09
rakudo/release-2026.09: 96fd3f7bbc | (Will Coleda)++ | tools/templates/NQP_REVISION
[release] Bump NQP revision to 2026.09
rakudo/release-2026.09: 9a94801c15 | (Will Coleda)++ | VERSION
[release] Bump VERSION to 2026.09
[Coke] can't scp the release files to trinity.rakudo.org 17:24
Geth rakudo: coke++ created pull request #6759:
Release 2026.09		 17:34
rakudo/main: 4 commits pushed by (Will Coleda)++
[Coke] ah, that was user error, all good, release proceeding... 17:37
apogee_ntv [Coke]++
japhb [Coke]: So is 2026.09 officially out now? 17:41
[Coke] japhb: typically the release manager will annoounce that. 17:46
.tell patrickb - source release is done for 2026.09 17:48
tellable6 [Coke], I'll pass your message to patrickb
[Coke] .tell el_che - source release is done for 2026.09
tellable6 [Coke], I'll pass your message to El_Che
[Coke] moving on to the rakudo.org updates... 17:59
japhb [Coke]: Apologies, wasn't trying to be a dick. Just was confused about whether the user error was the scp failure, or the merge to main.
18:07 Pixi` joined 18:10 Pixi left
[Coke] ok. everything's done, including finally fixing the wikipedia logo image to not say perl. 18:36
apogee_ntv++ for that last bit! 18:37
So, I think we're good. please try to build with rakubrew, test things out, especially on non-ubuntu
japhb [Coke]++
[Coke] japhb: sorry if I sounded disgruntled. (I am but was trying to not sound it. :)
japhb I figured you were just annoyed at *me*. 18:38
Geth marketing: m-doughty++ created pull request #29:
Add notext variant of Rakudo logo + thumbnail		 18:39
[Coke] no, you're good.
Geth marketing/master: 4 commits pushed by (Matt Doughty)++, (Will Coleda)++ 18:41
japhb Commented on github.com/Raku/marketing/pull/29 18:43
[Coke] whoops, I merged too quick, I guess.
apogee_ntv Weird, will fix and PR 18:45
japhb Local bespoke build-all has gotten through MoarVM, nqp, rakudo, and zef, and started on my module stack. Looking good so far.
apogee_ntv Oh, my white fill didnt actually fill the layer
Will fix 18:48
japhb Module stack failed on Test::Async:ver<0.1.16>:auth<zef:vrurg>:api<0.1.2> ... I think that had been pinned by another module. I'll see if I can break that pin. 18:49
ugexe test::async needs a new release. you can install from the repo head and it should work 18:51
japhb Ah, OK, will do.
apogee_ntv github.com/Raku/marketing/pull/30 There, when I filled the bg white I missed a spot so the transparent rendered grey 18:55
[Coke] github.com/rakudo/rakudo/wiki/post...%902026.09 18:59
Any notes for users who are having issues post-2026.09 (we can list problem modules, things that need to be updated and haven't been yet, how to workaround until they are, etc.) - and let's remove items once they are fixed. (e.g. once Test::Async is released) 19:00
apogee_ntv There's a bunch of stuff thats fixed by an install latest IIRC. DBIish was one. 19:02
[Coke] Feel free to add notes to the wiki page.
apogee_ntv Will take a gander through all the pins I updated after dinner 19:03
Is the new binary release out btw or is that another stage? 19:04
I think thats what gets it into setup-raku github action
[Coke] binaries are done by patrickb and el_che 19:17
(we should also get backups for all the binaries as well as the regular releases) 19:22
19:42 disbot left, disbot joined
apogee_ntv Sweet, I have a bunch of updates to do when the binaries are up. 19:45
japhb Hmmm, Test::Async from git HEAD didn't work either; in this case it was a failed test rather than just blowing up completely like before. 20:21
Hmmm, looks like a small smattering of failed tests, not just a single one. (Because it's nested testing a test module, it took a minute to determine which failures were actually top-level "module not working right" failures.) 20:26
20:40 finanalyst joined
ugexe if you're talking about your Terminal:: modules they did pass their tests with the fixed Test::Async at one point 20:45
but that would have been like 2 months ago at this point 20:46
ah yeah i see it does fail now again. sad, but i guess thats just a unavoidable side effect of fixes not getting released promptly 20:49
japhb nodnod 20:50
ugexe japhb: github.com/vrurg/raku-Test-Async/pull/5 21:27
zef install github.com/ugexe/raku-Test-Async.g...99cdea6b65 21:32
you can install it via that for now
Geth rakudo/main: 6 commits pushed by (Nick Logan)++ 21:41
[Coke] c: 66ba2944e1 pi 22:00
committable6 [Coke], ¦66ba294: «Cannot find this revision (did you mean “1942243”?)»
[Coke] we have our first ast bug report... from an rsc member... who didn't test their stuff pre-release. 22:03
[Coke] grumbles very loudly about that. 22:04
github.com/rakudo/rakudo/issues/6760
[Coke] thinks it's ENOBUG 22:09
Geth rakudo/main: 5 commits pushed by (Nick Logan)++ 22:27
rakudo/main: 6eb744de7a | (Nick Logan)++ | 7 files
RakuAST: make a bracketed infix behave as the infix it brackets

Previously `RakuAST::BracketedInfix` passed only its properties and compilation through to the infix it wraps and took everything else from the `RakuAST::Infixish` defaults. As such `[op]` was always considered pure, sank all of its operands, and never thunked or primed them. `@a[1] [=] 9` in sink context warned that both the assignment and ... (18 more lines)		 23:15
rakudo/main: d6107d0502 | (Nick Logan)++ | 2 files
RakuAST: let a bracketed list infix chain with itself

Previously the list associativity check compared the operator text of the infix on the operator stack against the full text of the incoming infix. The full text of a bracketed infix includes its brackets while its operator text does not, so `1 [,] 2 [,] 3` and `1, 2 [,] 3` died claiming that `,` and `[,]` differ. This compares the operator text of both.
rakudo/main: 3334454bb3 | (Nick Logan)++ (committed using GitHub Web editor) | 8 files
Merge pull request #6747 from ugexe/ugexe/rakuast-bracketed-infix

RakuAST: make a bracketed infix behave as the infix it brackets (fixes Useless use warning)
rakudo/main: 357988eaba | (Nick Logan)++ | 2 files
Decide a literal smartmatch the way the literal's ACCEPTS does

Previously the optimizer reduced `$x ~~ 42` inside a CATCH that answered False, even for `!~~`, so `$obj !~~ 42` was False for a topic without a Numeric. The undefined topic path ignored `!~~` the same way. The CATCH also swallowed exceptions from a topic's `.Str`, which
  `Str.ACCEPTS` lets through. A string literal compared any topic over
... (12 more lines)		 23:25
rakudo/main: bc3d643019 | (Nick Logan)++ | 2 files
RakuAST: decide a literal smartmatch the way the literal's ACCEPTS does

Previously the reduced `$x ~~ 42` called `$x.Numeric(:fail-or-nil)` unguarded, so a Failure topic threw, as did any object whose Numeric died or did not exist, where the literal's ACCEPTS answers False. A Numeric that returned a type object threw or warned, and a Failure it returned was never marked handled. An Int topic was compared through ... (9 more lines)
rakudo/main: 33d0242de9 | (Nick Logan)++ (committed using GitHub Web editor) | 4 files
Merge pull request #6751 from ugexe/ugexe/rakuast-literal-smartmatch-failure

Decide a literal smartmatch the way the literal's ACCEPTS does
[Coke] OOF 23:31
Geth rakudo/main: 1f48abb1d6 | (Nick Logan)++ (committed using GitHub Web editor) | 12 files
RakuAST: set declarator docs at BEGIN time (#6753)

Previously the RakuAST frontend set a declarator doc as the .WHY of what it documents only at CHECK time, when it also made the doc's $=pod entry. A trait of the declaration and any BEGIN time code after it saw no .WHY. A class that did a role got the role's attributes and methods without their docs. A role punned at BEGIN time gave its instances no doc. Both ... (16 more lines)		 23:39
[Coke] ugexe: let me know when you're done for the day, will kick off a run 23:43
ugexe 👍
[Coke] github.com/rakudo/rakudo/issues?q=...RakuAST%22 - 72 issues tagged as "fix in rakuast" 23:52
Geth rakudo/main: a541fb6b7b | (Nick Logan)++ (committed using GitHub Web editor) | 2 files
RakuAST: parse Int(Str:D) and Int(Array[Int]) as coercion types (#6752)

Previously a type name followed by a type in parentheses was only taken as a coercion type when the inner type's own name was a type. `Str:D` is named `Str:D` and `Array[Int]` has no name at all, so `Int(Str:D)` parsed as a call that died with "Impossible coercion", `Int(Str:U)` died with "Cannot create an Int from a 'Str:U' type object", and ... (7 more lines)		 23:57

Powered by Cro and the Raku® Programming Language.

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

Thank you!