04:45 hurufu joined 04:47 hurufu left 04:48 hurufu joined 07:38 hurufu left 07:53 hurufu joined 09:03 hurufu left 10:07 rnddim is now known as ShimmerFairy 12:45 hurufu joined 12:53 donaldh left 14:44 hurufu left 17:13 El_Che joined
El_Che 19:11 < El_Che> is redhat 8 still a target for redhat 8 (supported until 2029)? The new moar breaks there, was ok on last release: ./libmoar.so: undefined reference to `gettid' 17:14
Geth nqp/main: ca2e391a07 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to get gettid fix
 17:15
rakudo/main: 5fae814f6f | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get gettid fix
 17:22
lizmat El_Che: ^ should be ok after this
El_Che does it bump a vesion? 17:24
version
lizmat you mean, will we do a Rakudo point release for this ? 17:28
El_Che or a moarvm dot release, I can set the version for each component separately
[Coke] I can probably do a moarvm dot release to get that one thing. 17:29
probably not until saturday
El_Che it's cleaning than patching for 1 OS
I'll release the packages saturday then 17:30
[Coke] ok. will mention here if I can get to it sooner. 17:32
El_Che [Coke]++
[Coke] is there any way we can catch this stuff sooner?
our process should have caught this on the original PR. 17:33
El_Che I had a dev build on rakudo-pkg where you could build any commit on the matrix of distros+versions but I removed it because no one used it 17:34
I guess you get get an idea by forking rakudo-pkg, and change config/setup.sh with your commits, push it and the action would build the matrix 17:36
[Coke] where is rakudo-pkg? 17:40
ah, under nxadm 17:41
El_Che github.com/nxadm/rakudo-pkg
Here is what's built: github.com/nxadm/rakudo-pkg/blob/b...ge.yml#L28
[Coke] can that test just that moarvm doesn't build? Does it need to have a full rakudo? 17:42
El_Che config/setup.sh has the version it checks out and builds
as is, it is not granular, no 17:43
[Coke] moarvm is doing 22 compilation checks - why doesn't it include all the OSes on your list? 17:49
added MoarVM/issue/2011 17:55
github.com/MoarVM/MoarVM/pull/2010/checks 18:04
that's the fix PR that was just merged. it is failing more than hald the CI checks. 18:05
half
nearly 2/3
Shouldn't that be 100% green, especially if we're trying to fix a compiler issue on one OS without breaking others? 18:07
I'm not going to do the point release until I'm sure we're not making things even worse. :) 18:12
lizmat [Coke]++ 18:13
[Coke] guessing part of the failure was that the merge happened before they finished? some of them are git failures that it couldn't fetch that branch 18:17
ok, looking better on main, down to 5ish failures, first one I checked is actually failing in rakudo native call? 18:19
dev.azure.com/MoarVM/MoarVM/_build...f1d64735f6 18:20
Moarvm builds are being tested against Rakudo main 18:22
I guess that's better than rakudo-latest-release, but i imagine we could get some false negatives there. 18:23
the moarvm failures are with MVM_clang, 18:24
El_Che flipping tests are pretty common
(flapping?)
[Coke] the moarvm failures are with MVM_clang, MVM_clang_malloc, MVM_clang_njit, MVM_clang_no_c11_atomics
also: if we have any flappers, fix them. we shouldn't have tests written in a way that flap at this point. 18:25
(just found one this release that was depending on hash key order in roast) 18:26
oof. just did "rerun failed checks" on the last commit in moarvm and it's rerunning *everything*? 18:27
El_Che the fails I saw, but disappeared after 1 retry: t/04-nativecall/09-nativecast.t, t/09-moar/Line_Break__LineBreak.t, t/12-rakuast/xx-fixed-in-rakuast.rakutest, t/02-rakudo/rakuast-code-assuming.t, t/04-nativecall/19-function-pointers.t, t/04-nativecall/00-misc.t , t/12-rakuast/origins.rakutest 18:34
[Coke] I'm rerunning the CI tests. 18:38
looks like the original 4 that died still died.
failing on t/04-nativecall/02-simple-args.t consistently 18:39
timo yeah, that's the thing with clang and gcc disagreeing and we're getting caught in the middle 18:42
the simple-args one
20:39 hurufu joined
Geth rakudo/lizmat-8: ca0d8d6e95 | (Elizabeth Mattijsen)++ | src/Raku/ast/package.rakumod
RakuAST: step one for codegenning POPULATE in every class

This actually compiles and works as expected (as far as I have been able to test in a simulated environment.
The next step will be to actually convert the created AST for the POPULATE method into an actual Method object, and add it to the ... (5 more lines)		 21:44
rakudo: lizmat++ created pull request #6211:
RakuAST: step one for codegenning POPULATE in every class
21:53 hurufu left
lizmat ugexe: re POPULATE, I think PRODUCE-META-ATTACHABLES is the wrong place for it, as it is also being called for roles 22:12
I think that the POPULATE codegenning logic needs to live in RakuAST::Class.PRODUCE-META-OBJECT 22:13
or be called from there
ugexe yeah sounds good to me. only thing i might add is having a PRODUCE-POPULATE method or some such that is called from PRODUCE-META-OBJECT 22:14
tellable6 2026-05-28T22:11:50Z #moarvm <lizmat> ugexe: re POPULATE, I think PRODUCE-META-ATTACHABLES is the wrong place for it, as it is also being called for roles
lizmat ok, I'll go that way tomorrow... :-) 22:15
sleep&
22:59 apogee_ntv left 23:00 apogee_ntv 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!