00:17 kjp left 00:20 kjp joined 00:34 MasterDuke joined 00:48 MasterDuke left
Geth roast: 00c182ec4f | (Nick Logan)++ (committed using GitHub Web editor) | S12-construction/destruction.t
Stabilize destruction.t plan against GC missing the outer Child

The test plans 6 and the for-loop emits one `pass "Parent after child"` per `/Parent/` match in @destructor_order. To get the second pass test the GC has to collect both the outer $child (line 31, set to Nil immediately) and the loop-scoped $chld. Under spectest pressure with Test.rakumod loaded from source via RAKULIB=./lib ... (10 more lines)		 01:23
rakudo: ugexe++ created pull request #6172:
Fix Test.rakumod failure-attribution across frontend / Repository combos		 01:30
roast/ugexe-patch-1: 837b304ae3 | (Nick Logan)++ (committed using GitHub Web editor) | S02-names/SETTING-6e.t
Stop hardcoding the SETTING::OUTER depth in SETTING-6e.t

The original test baked in SETTING::OUTER::OUTER::OUTER::OUTER, which broke once the 6.e PseudoStash adjusted how far past the unit marker SETTING:: lands. Discover the depth at runtime and assert the lookup forms (top-level indirect, indirect inside EVAL, CALLER indirect) rather than a specific OUTER count.		 03:06
roast: ugexe++ created pull request #887:
Stop hardcoding the SETTING::OUTER depth in SETTING-6e.t		 03:07
roast: a14a90c630 | (Nick Logan)++ (committed using GitHub Web editor) | S02-names/SETTING-6e.t
Stop hardcoding the SETTING::OUTER depth in SETTING-6e.t (#887)

The original test baked in SETTING::OUTER::OUTER::OUTER::OUTER, which broke once the 6.e PseudoStash adjusted how far past the unit marker SETTING:: lands. Discover the depth at runtime and assert the lookup forms (top-level indirect, indirect inside EVAL, CALLER indirect) rather than a specific OUTER count.		 04:04
linkable6 ROAST#887 [closed]: github.com/Raku/roast/pull/887 Stop hardcoding the SETTING::OUTER depth in SETTING-6e.t
rakudo: ugexe++ created pull request #6173:
RakuAST: report stub grammars in X::Package::Stubbed alongside other stubs		 04:36
rakudo/main: aa5abe4056 | (Nick Logan)++ | lib/Test.rakumod
Fix Test.rakumod failure-attribution across frontend / Repository combos

  `proclaim` walks the call stack to find the user's source location for
the "Failed test at <file> line <N>" line. The walker skips frames whose file matches Test.rakumod's own `$?FILE` via `ends-with`. The one-way `$?FILE.ends-with($caller.file)` worked when both strings were equal but breaks when one is shorter than the other, which ... (27 more lines)		 05:06
rakudo/main: 57b9943f31 | (Nick Logan)++ (committed using GitHub Web editor) | lib/Test.rakumod
Merge pull request #6172 from ugexe/ugexe/fix-test-frame-walker

Fix Test.rakumod failure-attribution across frontend / Repository combos
rakudo/main: 585e93fe6f | (Nick Logan)++ | src/Raku/ast/package.rakumod
RakuAST: report stub grammars in X::Package::Stubbed alongside other stubs

Grammar.PERFORM-CHECK overrode Package.PERFORM-CHECK without delegating, so the parent's stub-not-defined check never ran for grammars and `my class A {...}; my grammar B { ... }` reported only A in X::Package::Stubbed.
... (11 more lines)
rakudo/main: 6654591f80 | (Nick Logan)++ (committed using GitHub Web editor) | src/Raku/ast/package.rakumod
Merge pull request #6173 from ugexe/ugexe/rakuast-grammar-stub-package-stubbed

RakuAST: report stub grammars in X::Package::Stubbed alongside other stubs
rakudo: ugexe++ created pull request #6174:
Add CI job for RAKUDO_RAKUAST=1 built rakudo to run RAKUDO_RAKUAST=1 make spectest		 05:08
ugexe i think RAKUDO_RAKUAST=1 make spectest passes now 05:10
lizmat_ ugexe++ 09:25
09:25 lizmat_ left 09:26 lizmat joined 10:10 ilogger2 left, ilogger2_ joined 10:57 librasteve_ joined
Geth rakudo/main: b49217d2d7 | (Elizabeth Mattijsen)++ | IN-MEMORIAM.md
Add obituary for Clifton Wood
 11:55
[Coke] :( 14:22
timo what? no!! :o 14:23
Xliff was not on IRC constantly, showing up at random-ish intervals if i remember correctly, so it's easy to not think anything's wrong if you don't see someone for longer than usual 14:30
coleman Appreciate the time we have. <3 17:31
18:53 ds7832 joined 19:21 ds7832 left
disbot2 <librasteve> just to share that I have released the refactored version of Physics::Measure (and friends) on a clean build of moar-2026.04 - all tests now pass - I wasnt sure if the recent blin test failure was 2026.04 or HEAD - feel free to point me to any new test errors 20:14
20:44 ds7832 joined
lizmat the blin run was on HEAD, 2026.04 already had its blin run before the release :-( 21:45
disbot2 <librasteve> well a lot of the tests are new so maybe will pass on new P::M release … happy to try to build HEAD and test if that helps 21:56
lizmat Physics::Measure installs fine on HEAD 22:00
22:08 ds7832 left
librasteve_ =b 22:10
I ended up making FatRatStr.round in my module of that name (otherwise the $scale eg Num can infect the precision of a FatRat invocant) … I’ll go in and make an issue for FatRats.round now the tactical alternative is done. 22:16
22:57 MasterDuke joined
MasterDuke ugexe: fantastic work with rakuast. now that it's getting even closer to done-as-in-we-consider-it-passes-enough-of-a-blin-run, have you thought about performance at all? on my m2 mba it's about 50% slower, on my zen 2 desktop it's about 30% slower. i've only roughly measured building rakudo, i assume it's probably less worse at running more 23:06
"normal" raku programs. i chatted with nine_ about this a little, and istr he was of the opinion that looking at profiles wasn't going to be much help, at least as far as pointing out particularly slow functions. instead slightly higher level concerns needed to be found and fixed (e.g., `foo` is being called way too many times, making it faster is
good, but really we need to do the work so it just gets called less frequently)
but identifying those sorts of things is going to be harder that just working down a profile sorted by exclusive time 23:08
timo, lizmat: i'm sure you have some thoughts also 23:10
timo i like turtles 23:11
MasterDuke but do they go all the way down? 23:12
timo i sure hope they do 23:13
the shape of the call graph can maybe still give some interesting insight into where we're doing too much, though since I'm mostly unfamiliar with the entirety of what goes on between RakuAST and compiler, I'm not sure where to even start looking 23:17
MasterDuke m: BEGIN my int $w; 23:18
evalable6 (exit code 1) ===SORRY!=== Error while compilin…
MasterDuke, Full output: gist.github.com/caa0de22804fb2833d...53c8e0efeb
MasterDuke bisectable6: BEGIN my int $w; 23:19
bisectable6 MasterDuke, Will bisect the whole range automagically because no endpoints were provided, hang tight
MasterDuke, More than 3 changes to bisect, please try a narrower range like old=2026.01 new=HEAD
MasterDuke, Output on all releases: gist.github.com/aae4e95e4c9f4e52ce...5d6fbe0133
MasterDuke huh 23:21
m: .say for $*OUT 23:25
evalable6
MasterDuke m: "CREDITS".IO.words.say 23:28
evalable6 (exit code 1) Failed to open file /srv/CREDITS: No such file or directory
in block <unit> at /tmp/CVECN0Ycf0 line 1
MasterDuke well, it's a little hard to show here, but there's something wrong about BEGIN with `RAKUDO_RAKUAST=1` 23:29
`raku -ne 'BEGIN my $w = 0; $w += .words.elems; END say $w' <path/to/file>` gives the correct value, but `RAKUDO_RAKUAST=1 raku -ne 'BEGIN my $w = 0; $w += .words.elems; END say $w' <path/to/file>` just prints the value from the last line of the file 23:31
timo interesting. have you checked the --target=ast and --target=optimize yet? 23:32
MasterDuke not yet. afk for a bit, got to put at least one kid to bed 23:34
timo ah yes, when you're not quite sure how many kids you have, but you can tell that it's gotta be at least one! 23:37
23:53 MasterDuke left

Powered by Cro and the Raku® Programming Language.

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

Thank you!