| SmokeMachine | it seems the new release is breaking Red... any idea what could be causing this? github.com/FCO/Red/issues/574 | 00:07 | |
| vrurg | SmokeMachine: lizmat will take care of this. AttrX::Mooish is affected too. Seems like something went wrong with BUILDPLAN when a metamodel class gets a mixin via 'does' | 01:20 | |
| Possibly not 'does' alone. | |||
| SmokeMachine | vrurg: thanks for answering! It seems it’s on very good hands! Thanks! | 01:29 | |
|
01:57
vrurg_ joined
01:59
vrurg left
04:04
andinus left,
andinus joined
08:35
sena_kun joined
|
|||
| lizmat | argh... how did Blin not catch this :-( | 08:46 | |
| Geth | nqp/main: a2c05d7f3e | (Elizabeth Mattijsen)++ | src/how/NQPClassHOW.nqp More tweaking fetching MRO Alas, c92ed85ec11e2 didn't prevent the REA from crashing every now and then. Basically revert it and add a little bit of debugging code that will sleep .1 second and then try again: this on the assumption that it is somehow a race condition that makes the $!mro attribute being nqp::null momentarily. If this code still fires, I guess we can rule out a simple race condition, but have to assume something more nefarious. |
10:19 | |
| rakudo/main: 3de8e110ac | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION Bump NQP to get REA crash trigger |
10:42 | ||
| rakudo/main: 7235b599bb | (Elizabeth Mattijsen)++ | lib/BUILDPLAN.rakumod Make BUILDPLAN resilient towards custom HOWs |
|||
|
11:25
patrickb left,
patrickb joined
11:26
patrickb left
11:28
patrickb joined
|
|||
| lizmat | greppable6: :$compiler_services | 11:41 | |
| greppable6 | lizmat, Found nothing! | ||
| lizmat | greppable6: $compiler_services | ||
| greppable6 | lizmat, Found nothing! | ||
| lizmat | which implies AttrX::Mooish is not in the greppable6 set | ||
| Geth | rakudo/main: f329c52d29 | (Elizabeth Mattijsen)++ | src/core.c/Mu.rakumod Use low-level invokability test Instead of testing for HLL Callable alone. This fixes the issue of having low level code blocks in a BUILDPLAN causing havoc, as found with problems in AttrX::Mooish and possibly Red. This is probably caused by the removal of some nqp::hllize() calls in the recent streamlining. Rather than reverting those changes, it feels better to do this change, as it also has a significant effect on performance in some cases (e.g. during "make spectest"). |
11:45 | |
| lizmat | vrurg_ SmokeMachine ^^ | ||
| Geth | nqp/main: abf15c4086 | (Elizabeth Mattijsen)++ | 2 files Remove hint, as per jnthn++ suggestion Also revert a2c05d7f3e6bbe as it did *not* fire on the second nqp::isnull check, but *still* fired on the nqp::elems() check. Which sort of rules out any type of race condition on setting / fetching the attribute. |
12:15 | |
| rakudo/main: 36183f8630 | (Elizabeth Mattijsen)++ | 2 files Bump NQP to get latest hint fix This apparently also required a change in NativeCall. Not quite sure yet whether the hint removal exposed an already existing bug, or that the difference in behaviour is the underlying bug causing the REA harvester crashes. |
12:43 | ||
|
13:13
librasteve_ joined
|
|||
| lizmat | bisectable6: say "2022320".contains(/2022320/) | 13:31 | |
| bisectable6 | lizmat, Will bisect the whole range automagically because no endpoints were provided, hang tight | ||
| lizmat, Output on all releases: gist.github.com/62b1d95284e0e76e0d...5c07daea19 | 13:32 | ||
| lizmat, More than 3 changes to bisect, please try a narrower range like old=2020.01 new=HEAD | |||
| lizmat | bisectable6: old=2023.02 say "9209932022320".contains("2022320") | ||
| bisectable6 | lizmat, On both starting points (old=2023.02 new=36183f8) the exit code is 0 and the output is identical as well | ||
| lizmat, Output on both points: «True» | |||
| lizmat | bisectable6: old=2023.02 say "9209932022320".contains(/2022320/) | ||
| bisectable6 | lizmat, On both starting points (old=2023.02 new=36183f8) the exit code is 0 and the output is identical as well | ||
| lizmat, Output on both points: «False» | |||
| lizmat | bisectable6: say "9209932022320".contains(/2022320/) | 13:33 | |
| bisectable6 | lizmat, Will bisect the whole range automagically because no endpoints were provided, hang tight | ||
| lizmat, Output on all releases: gist.github.com/4139b682cfc739cb22...ab525fa116 | |||
| lizmat, More than 3 changes to bisect, please try a narrower range like old=2020.01 new=HEAD | |||
| Geth | rakudo/main: b88d1d0fab | (Elizabeth Mattijsen)++ | 8 files Revert "Eradicate .publish_method_cache from MoarVM backend" This reverts commit 67ed16b350f737be4835d25f039b4194964a1c5c. Fixes remaining test failures on AttrX::Mooish. |
13:55 | |
|
14:08
nine left,
nine joined
|
|||
| Geth | rakudo/is-item-redux: 2d024c4c58 | ab5tract++ | 2 files Add sigil exception for 'is item' to RakuAST |
15:54 | |
| rakudo/is-item-redux: 5 commits pushed by ab5tract++ | 16:02 | ||
| SmokeMachine | lizmat: thanks! I’ll test that as soon as possible | 17:10 | |
| Geth | nqp/main: fabf296b1b | (Elizabeth Mattijsen)++ | src/vm/moar/QAST/QASTOperationsMAST.nqp Add hint back in This basicallyy reverts abf15c40860c09f as this did not prevent the REA harvester from crashing |
18:04 | |
| rakudo/main: 2a233d52ed | (Elizabeth Mattijsen)++ | src/Perl6/World.nqp Streamline accessor creation - shortcut to the World object, instead of repeated attribute fetches - replace some .push by nqp::push and [] by nqp::atpos |
18:26 | ||
|
18:53
librasteve_ left
22:37
sena_kun left
23:12
MasterDuke joined
|
|||