Geth nqp: MasterDuke17++ created pull request #841:
Use locals instead of class variables
03:21
timo can you try "pop a value from @arg_kinds and @arg_mast until empty" instead of increasing $i from 0 to length of @arg_mast for that last bit? 03:38
jdv releasable6: we'll see about that 04:14
releasable6 jdv, I cannot recognize this command. See wiki for some examples: github.com/Raku/whateverable/wiki/Releasable
nine m: my class InterpolationTest { has $!a = "a"; method m() { "a" ~~ /$!a/ } }; dd InterpolationTest.new.m 08:24
camelia ===SORRY!=== Error while compiling <tmp>
Attribute '$!a' not available inside of a regex, since regexes are
methods on the Cursor class. Consider storing the attribute in a
lexical, and using that in the regex.
at <tmp>:1
------> has $!a =…
nine m: 'my class InterpolationTest { has $!a = "a"; method m() { "a" ~~ /$!a/ } }; dd InterpolationTest.new.m'.AST.EVAL
camelia Match.new(:orig("a"), :from(0), :pos(1))
nine Funny...we forbid interpolating attributes in regexes because back then we couldn't figure out a way for it to DWIM and now spec tests fail on RakuAST because it just DWIMs instead of failing... 08:25
Geth rakudo/main: fc3fca455e | (Stefan Seifert)++ | src/Raku/Actions.nqp
RakuAST: throw appropriate exceptions on unsupported name assertions in regexes
08:46
rakudo/main: 62b8a649f3 | (Stefan Seifert)++ | src/Raku/ast/regex.rakumod
RakuAST: forbid interpolating attributes in regexes

We forbade interpolating attributes in regexes, because regexes are really methods and have their own self, thus the attribute would refer to a (likely non-existing one) on the Cursor. Since there seemed to be no way to make this do what the user probably meant, it was deemed better to forbid this usage altogether. ... (6 more lines)
nine 1186
m: my $counter = 0; my $subrule = '{$counter++; \'<alpha>\'}'; 'abc' ~~ /<$subrule>/; 09:27
camelia ===SORRY!=== Error while compiling /home/camelia/EVAL_0
Prohibited regex interpolation (use the MONKEY-SEE-NO-EVAL pragma to
override this error but only if you're VERY sure your data contains no
injection attacks).
at /home/camelia/EVAL_0:1…
nine m: use Test; my $counter = 0; my $subrule = '{$counter++; \'<alpha>\'}'; 'abc' ~~ /<$subrule>/;
camelia ( no output )
nine WTF?
timo does "use Test" accidentally export monkey somehow? 10:08
lizmat yes 10:33
nine Actually it seems to be on purpose: sub MONKEY-SEE-NO-EVAL() is export { 1 } 10:37
lizmat yup 10:38
a lot of spectests break without, is my recollection
Geth rakudo/main: 1665ac0f83 | (Stefan Seifert)++ | src/Raku/ast/regex.rakumod
RakuAST: support indirect lookups in regex interpolations

Fixes: /<::($name)>/
11:13
rakudo/main: cd0fd56db9 | (Stefan Seifert)++ | 2 files
RakuAST: Allow for interpolating code via strings in regexes

Fixes: my $c = 0; my $r = '{$c++; \'<alpha>\'}'; say 'abc' ~~ /<$r>/;
rakudo/main: 83694c0548 | (Stefan Seifert)++ | src/Raku/Actions.nqp
RakuAST: support imported MONKEY-SEE-NO-EVAL sub
nine 1187
Guest93 lizmat: re stackoverflow.com/questions/794113...g-grammar, it's very fast by setting RAKUDO_RAKUAST=1 12:05
lizmat and does it do the right thing ?
I mean, backing up to /dev/null is also very fast :-) 12:06
Guest93 i see it outputs qast code
at leas Stage parse is fast 12:07
least
without RAKUDO_RAKUAST, looks like moarvm gc eats most cpu, so it's slow 12:09
lizmat well, the new Raku grammar does not do the right thing in a lot of cases yet: this may be one of them 12:10
so unless it completes compilation *and* it produces the expected results, I'm not going to declare victory just yet
nine Well according to spec tests, our grammar stuff is actually in pretty good shape 12:13
JimmyZ lizmat: github.com/rakudo/rakudo/issues/5789 ,  recently I found nqp::getrusage(@rusage) eats about 15%CPU on my computer. 12:18
lizmat yeah, if the supervisor runs, it takes a bit of a single core
and apparently the repl starts the supervisor thread
(or probably rather the underlying editor logic does) 12:20
Geth rakudo/main: 1f8f25f04b | (Stefan Seifert)++ | 2 files
RakuAST: fix overzealous synthetic code point in charranges prohibition

Looks like code points generated from octal or hex sequences should be exempt from the check.
14:38
rakudo/main: b64475789f | (Stefan Seifert)++ | src/Raku/ast/regex.rakumod
RakuAST: fix LTM on lexical regex assertions

The CALL_SUBRULE indirection made it impossible to find the NFA which made LTM impossible.
rakudo/main: 4e811d550b | (Stefan Seifert)++ | src/Raku/ast/regex.rakumod
RakuAST: fix LTM with interpolated constants in regexes
nine 1188
Geth rakudo/main: 1a6fe31ccb | (Stefan Seifert)++ | src/Raku/ast/regex.rakumod
RakuAST: flip conditions of <?after ...> regex assertions

Method after expects the QAST of the condition to have been flipped during compilation.
15:23
nine 1189
Huh, tr is totally unimplemented 15:36
jdv gonna start the release in a bit. generating changelogs now. 15:38
Geth nqp/main: f86b874c9a | (Justin DeVuyst)++ | tools/templates/MOAR_REVISION
Bump MoarVM for release testing
15:41
rakudo/main: 67d722ca75 | (Justin DeVuyst)++ | tools/templates/NQP_REVISION
Bump NQP for release testing
15:42
lizmat jdv: am online to do any changelog vetting, if that still makes sense 17:58
jdv how did ccccfb7 break spectests but make it into the last release? 17:59
lizmat: thanks. almost done with rakudo changelog. moarvm one's been done for a bit. 18:00
nine jdv: it broke them when running with RakuAST
jdv ah 18:02
lizmat: rakudo one is up now 18:06
lizmat jdv: done 18:13
jdv ah, cool. thanks! 18:24
was just grabbing a bite to eat. i'll start the releases now then. 18:25
lizmat ++jdv
Geth rakudo/release-2025.02: cbe20d8f2b | (Justin DeVuyst)++ | 3 files
Update changelog + announcement

Deliberately not logged:
  [01339daf][e123587f][b891c808][bbd27707][e4e56554][e413a6e2]
  [3b064798][0cad3547][4350d5ca][c250768b][2be9b080][af3b5a65]
  [736208db][3cfb4f17][22367bb9][6d8e7474][62c83711][d82d91e8]
  [ac06359c][a8fe6d8a][81d59d64][61fa0c85][67d722ca]
18:49
jdv looks like errata spectest fallout, handling... 21:01