| ab5tract | use v6.*; %*ENV<RAKUDO_RAKUAST = 1; my class A { has $.a where { $_ > 0 } } | 07:23 | |
| m: use v6.*; %*ENV<RAKUDO_RAKUAST = 1; my class A { has $.a where { $_ > 0 } } | |||
| camelia | ===SORRY!=== Error while compiling <tmp> Two terms in a row at <tmp>:1 ------> T = 1; my class A { has $.a where { $_ ><HERE> 0 } } expecting any of: infix infix stopper statement end statement modif… |
||
| ab5tract | m: use v6.*; %*ENV<RAKUDO_RAKUAST> = 1; my class A { has $.a where { $_ > 0 } } | ||
| camelia | ===SORRY!=== Cannot resolve caller infix:«>»(Mu:U, Int:D); none of these signatures matches: ( --> Bool::True) ($ --> Bool::True) (\a, \b) (Real $a, Real $b) (Int:D $a, Int:D $b --> Bool) (int $a, int $b --> Boo… |
||
| ab5tract | m: use v6.*; %*ENV<RAKUDO_RAKUAST> = 1; class A { has $.a where { $_ > 0 } } | 07:24 | |
| camelia | ===SORRY!=== Cannot resolve caller infix:«>»(Mu:U, Int:D); none of these signatures matches: ( --> Bool::True) ($ --> Bool::True) (\a, \b) (Real $a, Real $b) (Int:D $a, Int:D $b --> Bool) (int $a, int $b --> Boo… |
||
| ab5tract | Oh, sorry, thought I was still in a chat with camelia | 07:25 | |
| Is there a roast test that is failing for this? Because it seems really wild that there aren’t at least a dozen test files that use where clauses to some degree | 07:56 | ||
| lizmat | there are more than 1000 matches for the word "where" in roast | 08:07 | |
| ab5tract | Thanks (I am glad I didn’t bring my laptop on vacation but I do feel a bit helpless) | 08:14 | |
| m: use v6.*; subset A where * > 0; my class C { has A $.a }; C.new: :a(7) | 08:15 | ||
| camelia | Use of uninitialized value of type Any in numeric context in any multi_declarator:sym<null> at /home/camelia/rakudo-m-inst-2/share/perl6/lib/Perl6/Grammar.moarvm line 1 ===SORRY!=== Error while compiling <tmp> Attribute definition of type A (i… |
||
| ab5tract | lizmat: I thought that use v6.* would use the new grammar, but this seems to go through the old one? | ||
| Oh, it mentions line 1 of the old grammar. If I check that I imagine it’s a conditional call into the new grammar | 08:17 | ||
| lizmat | use v6.* just activates the existence of the RakuAST classes | ||
| the actual grammar is chosen in src/main.nqp | 08:18 | ||
| ab5tract | Gotcha, thanks | 08:24 | |
| When we release it, though, it will use the new grammar, right? | |||
| m: Q|subset A where * > 0; my class C { has A $.a }; C.new(:a(7))|.AST.EVAL | 08:30 | ||
| camelia | ===SORRY!=== Unknown compilation input 'qast' |
||
| ab5tract | :( | ||
| So I’m not sure what is the point of testing with v6.* if the current semantics of it are not going to match the semantics we are going to ship | 08:33 | ||
| m: use v6.d; subset A where * > 0; my class C { has A $.a }; dd C.new(:a(7)) | 09:00 | ||
| camelia | C.new(a => 7) | ||
| ab5tract | m: use v6.e.PREVIEW; subset A where * > 0; my class C { has A $.a }; dd C.new(:a(7)) | 09:03 | |
| camelia | Use of uninitialized value of type Any in numeric context in any multi_declarator:sym<null> at /home/camelia/rakudo-m-inst-2/share/perl6/lib/Perl6/Grammar.moarvm line 1 ===SORRY!=== Error while compiling <tmp> Attribute definition of type A (i… |
||
| lizmat | yes | ||
| ab5tract | Does the code snippet fail with RAKUDO_RAKUST=1? | 09:04 | |
| Probably it does, as the eval is barfing | 09:05 | ||
|
11:37
[Tux] left
12:05
[Tux] joined
15:19
coleman left
15:21
coleman joined
|
|||
| librasteve_ | rakudoweekly.blog/2025/09/29/2025-...f-control/ | 17:54 | |
| ab5tract | librasteve_++ | 18:11 | |
| Geth | rakudo/presize-slurps: eefa17d3e5 | ab5tract++ (committed using GitHub Web editor) | src/core.c/IO/Path.rakumod Migrate to infix max The infix form is quite a bit faster than the prefix one. |
18:16 | |
| ab5tract | lizmat | 18:30 | |
| lizmat | librasteve_++ | 18:31 | |
| ab5tract | lizmat: so it turns out that those v6.* snippets *also* fail under RAKUDO_RAKUAST=1 .. but if the the use v6.* statement is removed entirely, things work as planned | ||
| github.com/rakudo/rakudo/issues/59...3348463746 | |||
| lizmat | ab5tract: but that's because they're in 6.d, no? | 18:33 | |
| so maybe it's because of the PseudoStash handling in 6.e being different? | 18:35 | ||
| ab5tract | huh, so there have been semantic changes on the way to 6.e that are un-related to RakuAST and are also versioned | ||
| that makes a lot of sense actually | |||
| lizmat | vrurg did a lot of work on 6.e PseudoStashes | ||
| ab5tract | I did notice that while working on that last ticket. The new stuff is much easier to read. | 18:38 | |
| Geth | rakudo/main: 5147b67f7a | ab5tract++ | src/core.e/PseudoStash.rakumod 6.e - Fix overly permissive lookup into CALLER The new PsuedoStash code was a joy to work with, even including a comment that easily led me to the solution: the modes can be combined. This fixes R#5910 (#5910) |
19:22 | |
| linkable6 | R#5910 [open]: github.com/rakudo/rakudo/issues/5910 [6.e][RAKUDO_RAKUAST=1] `$CALLER::foo` allows access to caller's `my $foo` in 6.e | ||
| Geth | ¦ rakudo: ab5tract self-assigned `::('Exception').gist` throws in v6.e.PREVIEW github.com/rakudo/rakudo/issues/5911 | 19:31 | |
| ab5tract | That one appears to be finished :D | 20:06 | |
| so it looks to be like the change in behavior in R#5915 is actually intentional | 20:07 | ||
| linkable6 | R#5915 [open]: github.com/rakudo/rakudo/issues/5915 [6.e][RAKUDO_RAKUAST=1] `my class A { has $.a where * > 0 }` does not compile in 6.e | ||
| Geth | rakudo/presize-slurps: cd08d14644 | ab5tract++ (committed using GitHub Web editor) | src/core.c/IO/Path.rakumod Actually migrate to infix max Last attempt was not quite what we needed. |
21:23 | |