Geth rakudo/main: 13 commits pushed by (Nick Logan)++
review: github.com/rakudo/rakudo/compare/c...b0d563742e
01:33
ugexe [Coke]: that one is worth of a new run itself^ 01:34
worthy rather
[Tux] 🐧 zef update Slang::Tuxic 06:41
🐧 zef install Slang::Tuxic
All candidates are currently installed
No reason to proceed. Use --force-install to continue anyway
🐧 raku --version
Welcome to Rakudo™ v2026.09-21-g69f2b375f4.
Implementing the Raku® Programming Language v6.d.
Built on MoarVM version 2026.09-2-g6e5144cb6.
`zef upgrade Slang::Tuxic` did the job. Still I see new warnings, but I will have to look at those later 06:45
Rakudo v2026.09-21-g69f2b375f4 (v6.d) on MoarVM 2026.09-2-g6e5144cb6
csv-ip5xs0.299 - 0.306
csv-ip5xs-201.163 - 1.191
csv-parser0.989 - 1.018
csv-test-xs-200.115 - 0.116
test2.088 - 2.126
test-t0.571 - 0.580
test-t --race0.345 - 0.351
test-t-207.345 - 7.420
test-t-20 --race1.682 - 1.704
csv-test-xs 0.016 - 0.017
tux.nl/Talks/CSV6/speed4-20.html / tux.nl/Talks/CSV6/speed4.html tux.nl/Talks/CSV6/speed.log
something slowed down
11:17 finanalyst joined
[Coke] so, it looks like tux had a new rakudo but an older module installed. If I install a new rakudo (esp. with rakubrew), none of the previously installed modules are available. Which one of those is the expected case? 12:22
ugexe:running 12:23
[Tux] The laslines of my raku update/install script are 12:30
zef update || true
zef install --exclude=perl \
Slang::Tuxic File::Temp CSV::Parser \
Data::Dump Inline::Perl5 p6doc Bench || true
zef upgrade --exclude=perl \
Slang::Tuxic File::Temp CSV::Parser \
Data::Dump Inline::Perl5 p6doc Bench || true
Does that need more?
laslines → last lines
[Coke] For me, those would be installing into a new repo tied to the new install, so I'd get whatever the latest was. if you're using the same repo somehow, I think the upgrade should have handled it. 12:34
(I have something similar, a script that invoke zef for all my "always use" modules) 12:37
I only have 'zef install' though, as the zef install is new at that point as well, so nothing to update.
Just wanted to make sure you weren't doing something like "export list of old modules" -> "install this exact set of modules" - your script there looks reasonable to me! 12:38
suggestion: do blin runs a branches that turn deprecation warnings into failures so we can see what all the ecosystem fallout will be for, e.g. "Please use %?RESOURCES<key> directly instead." 12:48
ugexe: For the Cyclic Dependency errors in blin - it's at least partially an issue with how blin is processing things (but it seems to make sense) - e.g. Geo::Hash depends on Geo::Hash::CustomBuilder which in turn depends on Geo::Hash::CustomBuilder - but zef doesn't care, and installed Geo::Hash with no complaint. blin says "heyyyyy" and puts it on the naughty list. 12:50
I think maybe blin needs to realize that Geo::Hash::CustomBuilder is IN Geo::Hash and therefore that's not a cycle. 12:51
ugexe i'm honestly not sure if it should be listing that in the build-depends if its included in the distribution 14:23
its slightly different than typical modules in that its listed in the "builder" field so i might just be forgetting some detail
[Coke] it's also under build-depends 14:24
which is where blin is seeing it, I think. 14:25
I agree that it shouldn't be there, but since zef allows it, blin probably should too
57% done with blin, no failures yet compared to 2026.09 14:26
I also want to eventually go through the AlwaysFail list and see if any of these can be resurrected or skipped. 14:48
lizmat afk for a bit for an update 15:05
15:06 lizmat left 15:11 finanalyst_ joined, finanalyst left 15:18 lizmat joined 15:31 finanalyst_ left
Geth rakudo: ugexe++ created pull request #6764:
RakuAST: instantiate a generic array or hash where it is declared
15:37
15:53 finanalyst joined
Geth rakudo: ugexe++ created pull request #6765:
RakuAST: only leave the scope for a heredoc body after a closing brace
16:18
[Coke] ugexe: IO::Directory::Watcher and Stomp (run still in progress) bisecting to fdb0d563742e1671bf84fe840e161ab89c72e662 16:31
linkable6 (2026-09-29) github.com/rakudo/rakudo/commit/fdb0d56374 Merge pull request #6754 from ugexe/ugexe/rakuast-short-circuit-no-curry
[Coke] 87% doe
done
Geth rakudo/main: a0433344aa | (Nick Logan)++ (committed using GitHub Web editor) | 3 files
RakuAST: only leave the scope for a heredoc body after a closing brace

The `cheat-heredoc` token parses the body of a heredoc right after the line that queued it, and it left the current scope while doing so. Leaving is correct after the closing brace of a block, since the body then lies outside that block. However the same token also runs after the semicolon that ends the statement of a statement prefix or phaser, a `constant` initializer, or a ... (12 more lines)
17:13
17:43 hankache joined 18:03 ShimmerFairy left, ShimmerFairy joined, Pixi` joined 18:06 Pixi left 18:07 Pixi` is now known as Pixi
[Coke] managed to kill blin with 3 modules remaining. rerunning with those 3 plus the failing 2... 18:29
ugexe: fdb0d56374 is clean on redo 18:43
Geth ¦ problem-solving: hankache assigned to codesections Issue Decoupling Raku module requirements from Rakudo versions github.com/Raku/problem-solving/issues/531 18:44
¦ problem-solving: lizmat unassigned from codesections Issue Decoupling Raku module requirements from Rakudo versions github.com/Raku/problem-solving/issues/531 21:38