JJMerelo Within the esLibre (sort-of Spanish FOSDEM) we will be having a Raku+Perl devroom, and I just published the CFP jj.github.io/raku-perl-scripting-d...ME-en.html 10:30
tellable6 2020-08-19T09:56:27Z #raku <tbrowder> jjmerelo thnx, amigo
JJMerelo Proposals welcome!
Geth rakudo/rakuast: 02408f28dd | (Jonathan Worthington)++ | 5 files
Basic RakuAST handling of `start`

However, the AST side of this needs some work; the implicit $/ and $! fresh declarations are not yet handled. Generally, thunking things that come with declarations want a unified approach, as there are a good number of them.
rakudo/rakuast: 7c0f575bda | (Jonathan Worthington)++ | src/Raku/Actions.nqp
Compile simple multi-part names to RakuAST
rakudo/rakuast: 2b78b35a8e | (Jonathan Worthington)++ | 3 files
Implement type/call disambiguation

Meaning that we can now correctly compile `say Int` in the RakuAST-based compiler.
releasable6 Next release in ≈3 days and ≈7 hours. 2 blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 11:00
Geth rakudo/rakuast: f335d556a0 | (Jonathan Worthington)++ | 3 files
Make package's name a RakuAST::Name

So we'll (with some more work) be able to support declaring things like
  `class Foo::Bar` (plus any version adverbs).		 13:16
rakudo/rakuast: 9e3afe4ac4 | (Jonathan Worthington)++ | src/Raku/ast/resolver.rakumod
Fix a typo and thinko in the resolver

We only need to deal with type/value differentiation in the compile resolver; an AST cannot have such ambiguities.
[Tux] Rakudo version 2020.07-66-g2db92e04f - MoarVM version 2020.07-19-g9a5203395
csv-ip5xs0.827 - 0.829
csv-ip5xs-208.112 - 8.234
csv-parser26.367 - 28.657
csv-test-xs-200.392 - 0.399
test7.809 - 7.854
test-t1.935 - 1.975
test-t --race0.817 - 0.841
test-t-2032.001 - 33.411
test-t-20 --race8.785 - 8.871
 13:30
Geth rakudo/rakuast: c47f1bfdb5 | (Jonathan Worthington)++ | 3 files
Stub in RakuAST compilation of packages

This gets us creating the RakuAST::Package element with the resolved meta-object. It doesn't yet make anything aware of the declared package, however, so we cannot yet resolve the declared type name.		 13:43
Geth rakudo/sorting-with-junctions: 18cb45afdc | (Elizabeth Mattijsen)++ | src/core.c/Rakudo/Sorting.pm6
Another go at addressing AlexDaniel++ remarks

  - Removed the explicit allomorphic candidates
  - Int/Num candidates check for stringiness if same, then compare strings
  - Str candidate checks for numberness, uses default cmp if so
  - Added Date candidate because Date's $!daycount is a native int
This keeps the basic performance improvements in place with only a slight overhead for the Str case (still 4x faster than before), and a slight overhead for the allomorphic cases other than IntStr and NumStr.		 14:25
lizmat ah, and there I thought I'd run spectests, when I did, but on master :-( 14:33
back to the drawingboard
AlexDaniel` Which spectest fails? I can't see any issue yet 14:38
lizmat ambiguous dispatch on IntStr 14:40
AlexDaniel` Aaah
lizmat brb
Geth rakudo/sorting-with-junctions: 42f1bd58fc | (Elizabeth Mattijsen)++ | 8 files
Introduce "allomorph" role

  - this allows for a single type to accept *all* allomorph types
  - add 3 candidates to handle all allomorph types
  - remove special allomorph handling in Int/Num/Str candidates
  - back to 4.4x faster
  - should allow for more optimizations / maintainability for allomorphs as well
  - add tests for adding additional type to the core		 16:13
[Coke] .tell jjmerelo changing the last paragraph in the article for tomorrow. I don't like what i wrote. 16:38
tellable6 [Coke], I'll pass your message to JJMerelo
[Coke] .tell jjmerelo thanks for getting it *into* wp and setting up the publish 16:39
tellable6 [Coke], I'll pass your message to JJMerelo
[Coke] done. 16:50
vrurg m: my $a = "a" but 13; say "Int: ", $a.Int; say +$a 20:11
camelia Int: 13
Cannot convert string to number: base-10 number must begin with valid digits or '.' in '3⏏5a' (indicated by ⏏)
in block <unit> at <tmp> line 1
vrurg Smells like a bug, isn't it? 20:12
lizmat why do you expect prefix + to call Int? 20:14
prefix + calls .Numeric 20:15
not Int, so it won't use the role mixin and thus fail 20:17
m: my $a = "a" but role { method Numeric() { 13 } }; say +$a' 20:18
camelia 5===SORRY!5=== Error while compiling <tmp>
Two terms in a row
at <tmp>:1
------> 3ole { method Numeric() { 13 } }; say +$a7⏏5'
expecting any of:
infix
infix stopper
postfix
statement end
…
lizmat m: my $a = "a" but role { method Numeric() { 13 } }; say +$a
camelia 13
vrurg lizmat: reasonable. Thanks! 20:20
Geth rakudo/suggest-capitalization-errors: b7ff282fc5 | (Elizabeth Mattijsen)++ | src/core.c/Exception.pm6
Suggest method names that differ only in capitalization

Currently, an expression like `say 42.which` will fail with:
   No such method 'which' for invocant of type 'Int'
  *without* suggesting .WHICH. This patch changes that to:
   No such method 'which' for invocant of type 'Int'. Did you mean 'WHICH'?		 20:30
rakudo: lizmat++ created pull request #3853:
Suggest method names that differ only in capitalization		 20:31
Geth rakudo/introspect-WHAT: 04a1fc6202 | (Elizabeth Mattijsen)++ | src/core.c/Mu.pm6
Add Mu.WHAT for introspection

Yes, foo.WHAT will be codegenned without actually calling any method. But to not have a method Mu.WHAT also means it won't show up in any introspection or error messages.
Together with b7ff282fc5, this will suggest WHAT for "say 42.what".		 20:52
Geth rakudo: lizmat++ created pull request #3854:
Add Mu.WHAT for introspection		 20:52
Geth roast: 89a82220f6 | (Elizabeth Mattijsen)++ | S03-metaops/hyper.t
Make test slightly less specific

It will break if there is suddenly a suggestion when one wasn't before.		 21:46
rakudo/suggest-capitalization-errors: 30c4b6e599 | (Elizabeth Mattijsen)++ | src/core.c/Exception.pm6
Perform Levenshtein on .fc of strings

To more easily cover capitalization errors		 21:47
Geth rakudo: 08f81f2474 | (Stoned Elipot)++ | src/core.c/Baggy.pm6
Fix Baggy.pickpairs with a Callable

Baggy.pickpairs' Callable should be invoked with the Baggy number of elements as the pairs are picked independently of the elements' weight in the Baggy object		 21:57
rakudo: 88aa6db7f6 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/core.c/Baggy.pm6
Merge pull request #3852 from stoned/baggy-pickpairs-callable

Fix Baggy.pickpairs with a Callable
rakudo/suggest-capitalization-errors: bb7c617aac | (Elizabeth Mattijsen)++ | t/05-messages/01-errors.t
Make test less specific

It was breaking because the number of suggestions changed.		 22:13
lizmat and that concludes my hacking for toda& 22:14
sena_kun lizmat, rest well! 22:40
AlexDaniel hmm Could not find symbol '&Junction' in 'X::Cannot' 22:41
anyway mentioned that in the ticket 22:42
timotimo X::Cannot find symbol Cannot in X 23:16

Powered by Cro and the Raku® Programming Language.

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

Thank you!