Geth rakudo/main: 6801dead6d | (Nick Logan)++ (committed using GitHub Web editor) | 4 files
RakuAST: elide unused implicit setup in re-formed frames (#6579)

A context serialized during a BEGIN-time compilation rebinds its lexicals by name into the unit's frame at load, so a re-formed frame must keep every lexical name that compilation serialized. Previously the re-formation kept the whole emission of the unused implicits, paying container clones, a routine variable bind, and a named slurpy ... (9 more lines)		 00:50
rakudo: ugexe++ created pull request #6580:
Fold the plus reduction over bare values		 00:56
rakudo/main: fd7bd6f573 | (Nick Logan)++ | 2 files
Fold the plus reduction over bare values

The fold's accumulator started as the first slot's container while every later value from the fold arrived bare, and a stored value arrived as whatever its slot held, always a container from an Array and sometimes one from a List, so the operator's callsite kept several argument shapes live and stayed polymorphic. Deconting the ... (9 more lines)		 02:03
rakudo/main: adf522f241 | (Nick Logan)++ | 2 files
RakuAST: fail the static call mark's soft check open

A name form call to a routine of the same unit takes the static call mark through IMPL-RESOLUTION-BOUND-ONCE, whose soft check was the only one of the soft guards that failed closed: a routine that could not answer the soft method was refused the mark. The published method cache of a mixin type in the setting under compilation can ... (8 more lines)
rakudo/main: af214e70eb | (Nick Logan)++ (committed using GitHub Web editor) | 4 files
Merge pull request #6580 from ugexe/ugexe/reduce-monomorphic-accumulator

Fold the plus reduction over bare values
rakudo: ugexe++ created pull request #6581:
RakuAST: re-form regex code with the optimize walk's marks		 02:35
03:28 vrurg_ left 03:30 vrurg joined 03:31 guifa left 03:49 guifa joined 04:47 ShimmerFairy left 05:08 camelia left 05:38 guifa left 05:43 guifa joined 06:22 disbot2 is now known as disbot 07:47 guifa left 07:57 guifa joined 09:33 finanalyst joined 10:22 unicodable6 left 10:24 greppable6 left 10:25 bloatable6 left, bloatable6 joined, bloatable6__ joined 10:27 tellable6 left, tellable6 joined 10:28 notable6 left, notable6 joined 10:29 guifa left 10:30 bloatable6 left, bisectable6 left 10:31 quotable6 left 10:33 sourceable6 left 10:35 releasable6 left 10:36 linkable6 left, linkable6 joined 10:40 guifa joined
Geth ¦ problem-solving: finanalyst assigned to coke Issue Adding type-graphs to documentation on Types github.com/Raku/problem-solving/issues/526 11:27
11:36 finanalyst left 12:17 rakkable left, rakkable joined 12:25 guifa left 12:26 guifa joined 13:14 AlexDaniel joined
AlexDaniel [Coke]: that doesn't look like an issue in whateverable 13:15
c: a0a60ed3c say 42
committable6 AlexDaniel, ¦a0a60ed: «Cannot test this commit (Commit exists, but an executable could not be built for it)»
AlexDaniel what we can do is ask shareable for the archive
shareable6: a0a60ed3c
shareable6 AlexDaniel, whateverable.6lang.org/a0a60ed3c
AlexDaniel and that archive should have the logs inside
13:16 timo left
AlexDaniel fatal: could not get a repository handle for submodule '3rdparty/mimalloc' 13:16
hmm
13:16 timo joined
AlexDaniel c: say 42 13:16
committable6 AlexDaniel, Seems like you forgot to specify a revision (will use “v6.c” instead of “say”)
AlexDaniel c: HEAD say 42
committable6 AlexDaniel, ¦HEAD(a0a60ed): «Cannot test this commit (Commit exists, but an executable could not be built for it)»
AlexDaniel, gist.github.com/331268775e0b67a91f...fa1115453b 13:17
AlexDaniel c: say 42
committable6 AlexDaniel, Seems like you forgot to specify a revision (will use “v6.c” instead of “say”)
AlexDaniel c: HEAD say 42
committable6 AlexDaniel, ¦HEAD(a0a60ed): «No build for this commit»
AlexDaniel okay, deleted it, let's try again and see if that was just a fluke 13:18
committable6 AlexDaniel, gist.github.com/cebbc789b25fd69cd3...a71a223e7f
13:19 guifa left 13:24 guifa joined
AlexDaniel c: HEAD say 42 13:28
committable6 AlexDaniel, ¦HEAD(a0a60ed): «No build for this commit»
[Coke] Lovely. I did make sure that my moarvm copy had no weird 3rdparty changes. 13:50
c: 2026.08 3.say 13:51
committable6 [Coke], ¦2026.08: «3␤»
[Coke] ok, at least the tagged release is working. :)
c: af214e70e 2.say
committable6 [Coke], ¦af214e7: «Cannot find this revision (did you mean “0e10f4e”?)»
[Coke] c: 6801dead6 Inf.say 13:52
committable6 [Coke], ¦6801dea: «Cannot find this revision (did you mean “ffd89e1”?)»
AlexDaniel c: HEAD say 42
committable6 AlexDaniel, ¦HEAD(a0a60ed): «No build for this commit»
AlexDaniel hmm
[Coke] c: c9828b2ae 2.say
committable6 [Coke], ¦c9828b2: «Cannot find this revision (did you mean “a69582e”?)»
[Coke] ok, lots of the revisions after that are just "not found"
(looking at recent commits on main)
14:05 linkable6 left, benchable6 left, shareable6 left, evalable6 left, bloatable6__ left, committable6 left
AlexDaniel I'm not sure why, buildable is kind of just hanging, let's restart all and see what happens 14:05
14:08 notable6 left, tellable6 left, huggable6 left, nativecallable6 left 14:09 guifa left, benchable6 joined, benchable6__ joined
[Coke] thanks, AD 14:10
14:11 quotable6 joined, sourceable6 joined, nativecallable6 joined 14:12 linkable6 joined, greppable6 joined, unicodable6 joined, greppable6__ joined 14:14 benchable6 left 14:16 guifa joined, greppable6 left
AlexDaniel c: HEAD say 42 14:18
mm
14:21 nine left, nine joined
AlexDaniel [Coke]: can you please change the ip to 93.158.237.2 on this line? github.com/Raku/whateverable/blob/...66-L221C81 14:32
for context: there's a limit of connections per IP, so this hack there makes it connect half the bots over ipv4, which dances around the restriction. Once a few years that ip changes 14:35
so I think right now buildable created the builds for everything, but committable cannot connect 14:36
[Coke] done. 14:51
Could probably have that read DNS on restart to pick one. 14:52
14:53 ShimmerFairy joined
AlexDaniel [Coke]: sure. My attempt at this didn't work because it was getting the ipv6 address from DNS, haha, but there's probably a way to force it 14:53
14:53 nativecallable6 left, quotable6 left, unicodable6 left, sourceable6 left, linkable6 left, benchable6__ left, greppable6__ left
[Coke] pushed, but I don't see 93.158.237.2 when I dig for irc.libera.chat 14:53
AlexDaniel I think it's different for everyone, it's just one of
[Coke] FYI, I'm seeing 188.240.145.101 193.57.167.220 108.181.132.149 14:54
14:55 guifa left 14:56 greppable6 joined, coverable6 joined, greppable6__ joined, tellable6 joined, shareable6 joined, sourceable6 joined, unicodable6 joined, bisectable6 joined 14:57 benchable6 joined, committable6 joined
AlexDaniel c: HEAD say 42 14:57
committable6 AlexDaniel, ¦HEAD(a0a60ed): «No build for this commit»
AlexDaniel oh really…
14:57 quotable6 joined
AlexDaniel it's still building, it seems… 14:57
14:57 releasable6 joined 14:58 evalable6 joined, bloatable6 joined, linkable6 joined, nativecallable6 joined, notable6 joined, huggable6 joined 15:00 greppable6 left 15:01 guifa joined
[Coke] huh 15:02
AlexDaniel c: HEAD say 42 15:51
committable6 AlexDaniel, ¦HEAD(af214e7): «42␤»
AlexDaniel [Coke]: OK, it's done, thank you for you help!
[Coke] thank *you* 16:01
Geth rakudo/main: 561c5121db | (Will Coleda)++ | 2 files
New dates are 4th Saturday now.

We've been doing it this way for a while. Also claim next release		 16:20
[Coke] c: 7173ba509f67dc717d639864d090ebcea7261140 2.say 16:40
committable6 [Coke], ¦7173ba5: «Cannot find this revision (did you mean “f20251a”?)»
AlexDaniel “This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.” 16:41
[Coke] weird. github.com/rakudo/rakudo/pull/6572/commits says it has one commit (that one ^^) but I don't see it.
ah. yes, that came in from a branch on ugexe's repo
AlexDaniel it was rebased, I'd guess that there's another commit just like this one but with different sha :) 16:42
whateverable actually builds branches too, but not outside PRs
[Coke] oh, we've been taking advantage of the branches recently, btw, thanks. 16:43
(lets us do a blin run from commit-x to branch-from-commit-x-but-default-to-rakuast)
AlexDaniel :O
[Coke] next merge commit has 2 commits, also from ugexe/ those commits showed up. yah, maybe a rebase.
AlexDaniel: that's where a ton of the rakuast commits in the last 2months have come from (cleaning up the results of those blin runs) 16:44
AlexDaniel nicee 16:45
it's quite satisfying to know that something I've built 10 years ago is getting so much use
a bit more than that now actually github.com/Raku/whateverable/commi...f123cb4b67 16:47
[Coke] \o/ 16:48
releasable6: next
releasable6 [Coke], Next release in ≈33 days and ≈2 hours. There are no known blockers. 16 out of 16 commits logged
17:43 ugexe left, ugexe joined
[Coke] .ask alexDaniel - if I see /tmp/whateverable/rakudo-moar/af214e70ebc4f10a02dc3cd3b9f4bf5de0cae09a is locked. Waiting - when doing a blin run - is a normal state? Or should I be digging into why it's getting a locked? (Often I'll get pages of this in a row) 19:11
tellable6 [Coke], I'll pass your message to AlexDaniel
20:14 finanalyst joined
[Coke] 561c5121dbf8d84c5ed815e7e57831707e58d713 21:47
linkable6 (2026-08-24) github.com/rakudo/rakudo/commit/561c5121db New dates are 4th Saturday now.
[Coke] blin run is showing 2 modules failures on that. wtf. 21:48
ah, one is notorious module AI::NLP
ugexe the source for it is 404 21:51
cpan.metacpan.org/authors/id/H/HO/...1.5.tar.gz
AlexDaniel [Coke]: it is normal, I think. If I remember right, if it's bisecting two modules and during that process they hit the same commits, then one will wait while the other is doing the work 21:52
[Coke]: the only issue if you have a hundred of modules that all bisect to the same commit, and even then they'd need to do it at the same time for that to be a problem
tellable6 hey AlexDaniel, you have a message: gist.github.com/aba5962f69530cb41a...fd59bed62d
AlexDaniel and the reason for this is that they untar the build, do the thing and then remove the build. If locking wasn't there it could wipe the unpacked build while something is using it
[Coke] sehr gut 21:53
AlexDaniel and we can't use another location because rakudo isn't relocatable (at least it wasn't at the time)
[Coke] ugexe: ah, presumably they deleted it with PAUSE?
ugexe yeah
AlexDaniel but you can, in theory, use some kind of atomic counter instead, so that it'd only wipe the build when nothing is using it anymore :) So it can be better, but as is it's not an issue 21:54
[Coke] Let's remove it from the index then?
AlexDaniel: yup, it's fine. (now that I can run this stuff on a machine that takes hours instead of nearly a day, all good)
ugexe it does still exist in REA though
[Coke] Do we have a git repo somewhere for the cpan module list? 21:55
ugexe maybe that isn't the reason then 21:56
[Coke] (not sure how that ecosystem works)
I assume cpan is preferred over REA if present.
but probably no failover if cpan issue. 21:57
ugexe rea basically is cpan since we dont take new cpan distributions
[Coke] ahhh 21:59
ugexe whatever it is has to be flakey. the failure as we see it is zef just isnt finding the name, but if i manually do `zef info AI::NLP` it finds it fine. but the fact it seems to happen so much with that module specifically is odd 22:01
Geth rakudo/main: e84cb29aef | (Nick Logan)++ | 3 files
RakuAST: re-form regex code with the optimize walk's marks

A regex declaration declined the re-formation, since compiling its body stored one entry per alternation on the Regex code object under a name minted from a fresh counter, and a second formation left every alternation NFA in the serialized code object twice. The alternation names now derive from the regex's source origin and the ... (16 more lines)		 22:04
rakudo/main: fd7cbde517 | (Nick Logan)++ | 3 files
RakuAST: run the regex QAST optimizations unconditionally

Previously the optimize-regex flag held the regex QAST optimizations back until the optimize stage began on the unit, so a regex compiled dynamically at BEGIN time ran its plain compilation for the rest of the program. The optimizations do not depend on anything the stage computes, and the NFA is stored before they run either way, so LTM ... (5 more lines)
rakudo/main: c6d72c566a | (Nick Logan)++ (committed using GitHub Web editor) | 5 files
Merge pull request #6581 from ugexe/ugexe/rakuast-regex-reform

RakuAST: re-form regex code with the optimize walk's marks
rakudo: ugexe++ created pull request #6582:
Answer reductions over a Range of Ints from its endpoints
[Coke] github.com/coke/raku-blin-release-...ee/2026.09 - might be some real failures in there. 22:09
(almost done with run)
ADT had the same "no identify found" 22:15
github.com/coke/raku-blin-release-...ee_0.2.3_0 looks like an actual single test failed. 22:16
561c5121dbf8d84c5ed815e7e57831707e58d713 22:17
linkable6 (2026-08-24) github.com/rakudo/rakudo/commit/561c5121db New dates are 4th Saturday now.
[Coke] ... but that's insane - there's no way that test only failed because of that commit.
Geth nqp/main: 73218adde1 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to get code coverage instrumentation filtering

Greg Donald++		 22:21
22:21 AlexDaniel left
Geth rakudo/main: aa293aafa9 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get code coverage instrumentation filtering

Greg Donald++		 22:28
22:57 guifa left 23:02 guifa joined 23:25 djinni` left 23:28 djinni` joined 23:32 japhb left, japhb joined 23:35 guifa left, finanalyst left 23:46 camelia joined 23:52 guifa joined
[Coke] did a redo run with ONLY failing modules. Nearly done, now have 3 failures pointing at aa293aafa96f73658f040447dc68e61bb0f19c72 23:56
linkable6 (2026-08-25) github.com/rakudo/rakudo/commit/aa293aafa9 Bump NQP to get code coverage instrumentation filtering
[Coke] (which wasn't available to break on in the first run) 23:57
ugexe what is the memory limitation for the ooms? 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!