MasterDuke ugh/arg. still trying to figure out this numeric variable exception. i've kind of reduced the problem to the fact that this gist.github.com/MasterDuke17/f81a0...2e0c450783 cases `my $kinds = 4; say $kinds` to die with `Variable '$kinds' is not declared.  Perhaps you forgot a 'sub' if this 01:19
was intended to be part of a signature?`
both stock and with that patch, `$*VARIABLE-NAME` is '$kinds' for the `$/.add-variable($*VARIABLE-NAME);` call 01:29
nine: do you have any suggestions for debugging rakuast grammar weirdness? i'm finding this considerably harder to make any progress on than the other rakuast stuff i just did, and hopefully i'm just missing something trivial in either the implementation or the debugging 01:41
anyway, likely afk until probably tomorrow evening 01:42
timo is `$<var><sigil>` correct there, does it have to be `$<var-dec><sigil>` instead? 02:14
MasterDuke timo: oops, i had fixed that locally, the error is with the fix 02:28
timo OK
MasterDuke for a little more context, i had added some error reporting to `token variable`, but it wasn't hitting. i realized it's because in src/Raku/Grammar.nqp, `token variable-declarator` has those three alternations, whereas in src/Perl6/Grammar.nqp it has `<variable>` instead (which is where the exception i added is thrown) 02:39
so i changed `token variable-declarator` in rakuast to also call `<variable>` instead and got the error i reported. at first i thought it was because of some other difference between the `token variable`s in rakuast vs legacy, but then was able to minimize it to just moving the existing lines out into a new token broke the same way 02:42
hm, i wonder if it's some inconsistency between `$*VAR`, `$*VARIABLE`, and `$*VARIABLE-NAME`? the legacy grammar doesn't even have `$*VARIABLE-NAME` at all. well, something investigate tomorrow 02:58
nine MasterDuke: stupid question, but does this patch come with accompanying changes in the actions? 07:48
tellable6 nine, I'll pass your message to MasterDuke
Geth rakudo/main: fa95e44195 | (Stefan Seifert)++ | 3 files
RakuAST: support multislice prefix

Fixes: @a[||(1, 2)] # to mean the same as @a[1; 2]		 10:13
rakudo/main: 989a3fc07f | (Stefan Seifert)++ | src/Raku/ast/expressions.rakumod
RakuAST: support multislice for hashes, too

Fixes: %h{||<a 2b} # to mean the same as %h['a'; 'b']		 10:25
nine That's 1308 10:26
lizmat nine: No registered operation handler for 'note' 10:39
setting doesn't build atm
nine Oh darn...must have left in some debug output 10:42
Geth rakudo/main: 01a3362125 | (Stefan Seifert)++ | src/Raku/ast/expressions.rakumod
RakuAST: remove debug output

  lizmat++ for the heads up
lizmat much better :-) 10:43
[Coke] rakudo-star was impacted by one of my issues in the release and needs to be re-generated. 11:59
(it's including the previous version of nqp)
Geth roast: 34d83553bc | (Elizabeth Mattijsen)++ | S26-documentation/block-leading-user-format.t
Fix test for declarator docs

Declarator docs need to be consecutive in RakuDoc: any (empty) lines will break the connection with the declarand		 12:18
rakudo/main: ee87a4bb71 | (Elizabeth Mattijsen)++ | 3 files
RakuAST: tighten up handling of declarator docs

Declarator docs need to be on consecutive lines: no (empty) lines inbetween are allowed, or the declarator doc will be considered to be without a declarand and cause a worry.
The first whitespace character after a #= or #| is ignored, but does not have to exist. So this allows '#=' as content of a whole line, but also '#=foo' to be generating the "foo" string.		 12:28
lizmat nine: I'm at 1305 now, which includes the -4 files I moved to "make test" yesterday 12:29
in any case: t/spec/S26-documentation/block-leading-user-format.t now passes 12:30
nine lizmat: bootstrapped or old frontend? 12:48
lizmat old frontend 12:49
(assuming you're talking about the number) 12:50
nine I am. Weird, my t/spec is clean
Geth roast: 99443c9bf3 | (Elizabeth Mattijsen)++ | S06-operator-overloading/circumfix.t
Remove macro test that was skipped

Because that particular case was a NYI in the first place		 13:04
roast: e72b07a130 | (Elizabeth Mattijsen)++ | S06-signature/sub-ref.t
Remove macro test that was skipped because NYI
 13:07
roast: 645623a4ab | (Elizabeth Mattijsen)++ | 6.c/S14-roles/mixin-6c.t
Remove duplicated macro test
 13:09
rakudo/main: 27e6dc36f9 | (Elizabeth Mattijsen)++ | t/13-experimental/macros.rakutest
Moved all non-specific macros tests here from roast

These tests were collected from test files that were testing non-macro specific properties of macros		 13:10
roast: e5ed02c883 | (Elizabeth Mattijsen)++ | 7 files
Remove all non-specific macros tests

As they have been moved to the rakudo repo		 13:11
lizmat 1307 after ^^ 13:17
Geth rakudo/main: 2604cb5d16 | (Stefan Seifert)++ | 3 files
RakuAST: fix QUIT phaser in supply preventing quit signal

The QUIT phaser needs to return any unhandled exception so it can get forwarded to a quit signal.
Fixes: my class UselessException is Exception { }; my $trigger = Supplier.new; my $s = supply { whenever $trigger { QUIT { when UselessException { } } } };
  $s.tap(quit => { say "quit" });
  $trigger.quit(Exception.new);		 13:25
nine 1309 now for me
Seems like we're 1 test file apart 13:26
Geth roast: 46447c3578 | (Elizabeth Mattijsen)++ | S06-macros/returning-string.t
Remove macros test file that was not part of spectest

Would not compile at all, so assuming it's a NYI		 13:29
roast: 9fb378c6d1 | (Elizabeth Mattijsen)++ | S06-macros/returning-closure.t
Remove macros test file that was not part of spectest

Would compile, but fail with a bytecode validation error		 13:34
roast: e11750d9da | (Stefan Seifert)++ | S05-mass/rx.t
Remove todo'ed Perl5 regex syntax test

There are no plans to implement this and indeed there's a consensus to remove Perl5 regex syntax support altogether. So remove this test that's failing more violently on the RakuAST based compiler.		 13:35
nine Oh, turns out my t/spec was tracking another local checkout instead of the upstream git repo 13:39
So I'm now at 1311. 13:40
Geth roast: 5fda75c47f | (Elizabeth Mattijsen)++ | S06-macros/postfix.t
Remove macros test file that was not part of spectest

Would not compile at all, so assuming it's a NYI
nine I confirmed that the ones we lost were just removed from the repo.
Geth rakudo/main: 570d1f895b | (Elizabeth Mattijsen)++ | 5 files
Move viable macro tests from roast to rakudo

These test files passed on the legacy grammar, but should not have been part of roast as they test experimental features		 13:45
roast: 18cb0ec726 | (Elizabeth Mattijsen)++ | 6 files
Remove valid macro test that now live in the rakudo repo

As they were testing experimental features, and so shouldn't have been in roast in the first place		 13:46
tbrowder has anyone seen the security warning for the nqp dir? says jquery version needs updating 13:51
lizmat nine: 1309 for me now 13:52
nine Seems like we're gonna have to compare lists at some point 13:54
tbrowder never mind, but warning about old github/perl6/std (jquery needs updating... 13:56
lizmat nine: my list: gist.github.com/lizmat/4673679ad5b...d950fa1a1f 14:00
nine t/spec/S12-attributes/augment-and-initialization.rakudo.moar and t/spec/integration/role-composition-vs-attribute.rakudo.moar 14:01
Geth rakudo/main: 76be78354b | (Stefan Seifert)++ | src/Raku/Actions.nqp
RakuAST: remove support for Perl5 regex syntax

Ironically the spectest will skip all and thus pass if all Perl 5 regexes fail. There's a community consenus that this feature should be removed as it's outdated, unmaintained and barely used.		 14:03
nine 1321 in the weirdest way to gain 10 spec tests :D
lizmat hehe 14:04
so, I think I see what the problem is: the grep on the error output: the file names were too long :-) 14:05
nine LOL 14:07
Isn't computing fun?
lizmat sometimes :-)
1321 confirmed 14:11
aka 98% 14:13
[Coke] releasable6: next 14:19
releasable6 [Coke], Next release in ≈24 days and ≈4 hours. There are no known blockers. 50 out of 51 commits logged
lizmat You can't adverb &
at /Users/liz/Github/rakudo/t/spec/integration/advent2013-day10.t:84
------> $z & $y :adverb⏏; # applies to &
releasable6 [Coke], Details: gist.github.com/733e63fa6f5788e308...45aa39005e
lizmat nine: that sorta implies we should be able to adverb infix:<&> ?
nine looks like 14:23
[Coke] As one of the people who added some of the integration tests, i reiterate - that was probably a mistake. the advent article isn't "spec", it was just something we didn't want to break in early days. 14:25
nine I'm fine with that, too :) 14:26
[Coke] Maybe we can consider removing them in 6.e's spec suite.
[Coke] github.com/rakudo/rakudo/issues/5775 - we do want to be testing Inline::Perl5 as part of each release, yes? 14:30
(Because right now we are skipping it)
Is there a stress test make target that *just* runs the delta between stress and spec? or does it always include base spec plus extra stress? 14:32
Geth rakudo/main: c522ddff89 | (Stefan Seifert)++ | src/Raku/ast/variable-declaration.rakumod
RakuAST: support call-assign initializer on shaped arrays

Fixes: my @c[2;2] .= new(:shape(2, 2), <a b>, <c d>);		 14:47
nine 1322
Geth ¦ rakudo: coke self-assigned release guide is in p5 pod github.com/rakudo/rakudo/issues/5834 14:49
[Coke] jdv: any issues with converting the release guide to rakudoc?
jdv will it render on gh? 14:50
i just need to be able to read it. whatever works.
(easily)
Geth roast: 4a284ebc7b | (Stefan Seifert)++ | S32-str/split.t
Remove incidental testing of Perl5 regex syntax

Both Raku and Perl5 regex syntax ought to compile to the exact same regex code. Thus in the context of testing Str split functionality, it doesn't make sense to test both regex styles. The input to split will be exactly the same. As there is a consensus around removing Perl5 regex syntax support, it seems better to remove these superfluous tests.		 14:52
nine 1323
[Coke] No, pretty sure rakudoc won't render on github. :| 14:58
lizmat but we haz nice rakudoc -> markdown translators 15:01
nine But then, why does it matter? If the release guide is readable as it is, why change? I'd bet there are plenty of ways to improve the whole release process that have more effect than translating the same information from one format to another. 15:05
[Coke] Me asking about something here is not "this is my highest priority" 15:07
Geth rakudo/main: ffb3741b69 | (Stefan Seifert)++ | src/Raku/ast/code.rakumod
RakuAST: fix regex signatures missing arity and count

Fixes: rx/b/.count		 15:08
nine 1324
[Coke] documenting issues as I notice them.
nine That's the point: I wouldn't even consider this an issue :) 15:17
Geth rakudo/main: f46e99409e | (Stefan Seifert)++ | src/Raku/ast/call.rakumod
RakuAST: fix arguments missing from metaop hyper calls

Fixes: (-> $a { say $a },)>>.(1)		 15:20
nine 1325
lizmat looks like rakudo.org/community/ is pretty outdated 15:34
only 25 to go! 15:41
Geth rakudo/main: 5c35f2129e | (Stefan Seifert)++ | src/Raku/ast/expressions.rakumod
RakuAST: support adverbing of list infixes

Fixes: sub infix:<&>($a, $b, :$adverb){ }; 1 & 2 :adverb		 15:53
nine 1326
nine lizmat: AFAIK the failing RakuDoc tests are because of macros 16:14
[Coke] ugexe: how can I use zef to list modules installed in, say, my $CLONE/install/share/perl6/site ? (Looks like I can use --to="#inst"... when installing, but what about listing?) 17:03
the Ake release scripts are not installing inline-perl5 to a specific inst, but instead are relying on the right perl6 (yes) and zef being first in path, which I'm not 100% certain is working. 17:05
Geth rakudo/coke/releng1: 267603b867 | (Will Coleda)++ | tools/releasable/Akefile
Fix -I path, use META6.json
 17:21
ugexe [Coke]: zef list --installed site 17:34
you can do `zef list --installed` to see what each individual local repo has 17:37
[Coke] ok, so that has to be the zef in that install, yes?
I can't use system zef to poke around in a local rakudo clone's install dir.
ugexe you don't have to have zef installed to use it 17:39
m: say $*REPO.resolve(CompUnit::DependencySpecification.new(:short-name<Test>)).repo 17:40
camelia inst#/home/camelia/rakudo-m-inst-2/share/perl6/core
ugexe you could s/Test/Inline::Perl5/ though if you want a 1 liner to find where something is resolving (and thus installed) to
[Coke] this ake task is installing zef... but it looks like it's not even installing zef to the ./install dir, so no surprise inline-perl5 isn't there. I'll try hardcoding the -to= here rather than trying to rely on paths. 17:42
ugexe if you want to install to site make sure you use -to=site and not -to=inst#/path/to/site
they both work, but the former doesn't need to double precompile since its referencing the repo by name 17:43
[Coke] (ugh, have 3 rakudo checkouts here. zef *is* getting installed to the ./install dir)
Thanks for the pointers. 17:46
(something screwy here; it says all candidates are installed when I try to install Inline::Perl5, but then that doesn't up in the list --installed from the installed zef. 17:53
ugexe show me the commands and the output for both of those steps 17:54
[Coke] it's in ake, trying to reduce it to shell and getting it working then convert it back. 17:56
ugexe ake is probably adding some paths to e.g. RAKULIB
and Inline::Perl5 is installed to one of those
when you run zef outside of ake it presumably does not have e.g. RAKULIB set the same 17:57
so it does not list anything installed in those paths
[Coke] gist.github.com/coke/1450cc7d1f9e1...fdf42208fe is the custom env passed to the run() commands. 18:00
I don't see anything suspicious there, and do see the updated PATH entries so we are running the just-built-and-installed perl6/zef 18:01
(no RAKULIB)
ugexe hmm im not following entirely, but that kind of suggests it is not the same version of zef. i think what we want to know is 1) what is the full path of zef being invoke in ake to install Inline::Perl5, the arguments passed to zef to install Inline::Perl5, and what the env is at the time of install for Inline::Perl5. And then 2) the same aforementioned steps but for the `zef list --installed` 18:06
command
s/same version of zef/same zef installation/
[Coke] github.com/rakudo/rakudo/blob/main...#L494-L500 18:09
that env is passed to both install zef and install inline-perl5. if I add a shell('which perl6', :%env) and zef, I get: 18:11
/home/azureuser/raku-release/rakudo/tools/releasable/rakudo/install/bin/perl6 18:12
/home/azureuser/raku-release/rakudo/tools/releasable/rakudo/install/share/perl6/site/bin/zef
which look right.
ugexe /home/azureuser/raku-release/rakudo/tools/releasable/rakudo/install/share/perl6/site/bin/zef list --installed 18:14
does that show Inline::Perl5?
[Coke] nope 18:37
[Coke] (it does show zef) 18:37
oh: 18:42
===> Found via inst#/home/azureuser/raku/share/perl6/site
it's not looking in what I would consider "its own" site 18:43
ugexe hmmm what about 19:16
/home/azureuser/raku-release/rakudo/tools/releasable/rakudo/install/bin/perl6 /home/azureuser/raku-release/rakudo/tools/releasable/rakudo/install/share/perl6/site/bin/zef list --installed
[Coke] AHA
So the install is actually working (Inline::Perl5 is there) 19:17
[Coke] interesting, the path complained aboutt in the stress test is different: 19:24
zef --install-to=inst#/home/azureuser/raku-release/rakudo/tools/releasable/rakudo/t/../gen/build_rakudo_home/site install Inline::Perl5 --exclude='perl' 19:25
is the suggested command.
Geth rakudo/main: 6fd413c05f | (Elizabeth Mattijsen)++ | src/core.c/RakuAST/Fixups.rakumod
Add RakuAST::Node.parent method

Inside RakuAST::Node.grep/map/first this will return the parent object of the invocant. Takes an optional integer argument to indicate the number of generations you want to add (default: 0)		 19:31
roast: 0cf20295bc | (Stefan Seifert)++ | integration/rule-in-class-Str.t
Avoid using Perl 5 regex syntax when not actually testing Perl 5 regex support

Usage of Perl5 regex syntax in this test was purely incidental. Since there's a consensus to remove the outdated, incomplete and unmaintained Perl 5 regex syntax support, it's best not to use it unless it's specifically required.		 20:53
nine 1327
ab5tract nine: I’m out of the loop. Is the RakuAST backend now bootstrapped? 20:58
nine ab5tract: that's right 20:59
ab5tract Wow! 😮
nine There's just 23 spectest files still failing. A few are due to advanced generics stuff. The rest is practically all missing compile time checks. 21:00
ab5tract There is indeed an end to this tunnel, and it’s appearing before us now. 21:03
Next comes macros and all sorts of other goodness :D 21:04
I am grateful for your stalwart efforts nine 21:05
nine :)
[Coke] 23? I have 1327/1355 on the changelog draft.
nine [Coke]: I went on lizmat's "only 25 to go" which was 2 spectest files ago 21:06
[Coke] m: say 1355 - 1327
camelia 28
[Coke] I'm just bumping the first number there whenever you say something here. 21:07
nine Files=1350, Tests=119667, 75 wallclock secs (12.79 usr 2.74 sys + 1712.37 cusr 142.92 csys = 1870.82 CPU) 21:08
That's a normal spectest run
lizmat moved a bunch of tests for experimental features to rakudo repo today. So that's why the total number went down
Fix tests by removing tests :D
[Coke] oh, right 21:09
releasable6: next
releasable6 [Coke], Next release in ≈23 days and ≈21 hours. There are no known blockers. 55 out of 56 commits logged
[Coke], Details: gist.github.com/cb2f9340cd74ea5a33...678392f996
[Coke] updated ^^
nine Now this is odd....with RAKUDO_RAKUAST=1 I get: 21:11
Files=1352, Tests=119180, 81 wallclock secs (13.24 usr 2.81 sys + 1925.67 cusr 163.75 csys = 2105.47 CPU)
2 more spectest files with RakuAST?
Geth rakudo/main: 870239b4a2 | (Stefan Seifert)++ | src/Raku/ast/traits.rakumod
RakuAST: report warnings emitted during trait application as worries

This makes us report the proper location in the code rather than some internal function.		 21:23
Geth rakudo/main: 06a7b5daf5 | (Stefan Seifert)++ | src/Raku/ast/term.rakumod
RakuAST: report no such symbol for unknown terms
 21:36
nine 1328 21:37
MasterDuke nine: doh, yes, i had forgotten to add a `method var-dec` to Actions. however, doing so with a body of `self.simple-variable($/);` doesn't change anything. i see it creating something that looks reasonable in `simple-variable()`, but i still get the `Variable '$kinds' is not declared...` error 23:42
tellable6 2025-03-26T07:48:35Z #raku-dev <nine> MasterDuke: stupid question, but does this patch come with accompanying changes in the actions?

Powered by Cro and the Raku® Programming Language.

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

Thank you!