Issues.md

1. (OPENED) library enum used in MAIN, import from bin/ script - enum doesn't get parsed
2. (OPENED) built-in names cannot be used as enums in MAIN, otherwise the parsing fails silently
3. %?RESOURCES<>.IO compile-time - why couldn't it be slurped?
4. Slips
   1. X and Z disregards them
   2. they unwrap even if they are in Scalar containers - is this desirable?
5. .& detachment
   1. why is it not detachable in the first place?
   2. why does 'asd' .comb.&dd count as detached? Surely it couldn't be more attached to the .comb part?
6. (OPENED) zef --help writes to stderr
7. (OPENED) smartmatch to S/// is useful and it does work
   • exactly as $foo ~~ /regex/ or $foo ~~ m/regex/ would work
     • sets the $/ match variable
     • returns with the result of substitution
   • given or with is not sufficient replacement because there can be only one of them per statement
   • $foo ~~ S/a/b/ ~~ S/c/d/ is actually a good-looking paradigm
   • alas, the current behavior is documented: https://docs.raku.org/language/regexes#Substitution -> I propose the adoption as well
     • either deleting the note plain and simple
     • or also adding that it works like $foo ~~ /regex/, setting $/
8. (OPENED) sub demo(|c, :$name) { say c; say $name; }; demo(1);
   • "expected 1 positional but got 1" WTF?
   • seems like it actually expects no positionals at all
   • how does this work?
9. (OPENED) ().min == Inf && ().max == -Inf ?
10. ACCEPTS(Match:D:) is a hack
    • confer "smartmatch to S/// is useful and it does work" issue
    • in order to support s/// and m// (perhaps tr/// as well?), smartmatching to a match object always returns the match object
    • this is a more serious and fundamental misuse of smartmatch than any "regex operator"
11. ~~ order of argument evaluation:
    • precedence: like ==
    • when executed:
      1. takes the LHS value
      2. binds $_ to the LHS, inside the RHS expression
      3. evaluates and takes the RHS value
    • pro: very clever
    • contra:
      • hard to describe: specify, document, explain
        • maybe referring to && and || (short-circuiting) can help
      • hard to implement?
        • does it contribute to the Rakudo-as-Raku merger of implementation and specification?
12. my Int &var things:
    • valid, designed concept backed up by Callable[::T] - need to be documented
    • Callable[::T] itself needs to be documented: based on the declared return value of the Callable instance
    • The sole fact that Callable can be parameterized and what that does, needs to be documented.
    • my Int &var = -> --> Int { 6 } fails; this is probably a bug, file a Rakudo issue for this.
    • my Int &var defaults to unparameterized Callable which is bogus typing and my first bet would be that it's also a bug - file a Rakudo issue for it. Callable[Int] should be sufficient for a default value.
13. Nil and min / max:
    • Nil acts
      • as 0 for numbers
      • as the string "Nil" for strings
      • as (0, ) (???) for lists
    • needs consistent semantics, possibly error
14. $<a b> should behave like $[0, 1]
15. (OPENED) Handling of non-breaking spaces when splitting to words

    • https://docs.raku.org/language/traps#___top "using Set subroutines (...)" part

      • it produces "a b" and returns False for both code examples
      • ... for all bisectable6 versions

    • https://docs.raku.org/language/quoting#index-entry-quote_%3C%3C_%3E%3E-quote_%C2%AB_%C2%BB-Word_quoting_with_interpolation_and_quote_protection:_%C2%AB_%C2%BB

      • seems to work the same way as <<>> but that way also doesn't match the output in the docs: ("42 b", "c ")

    • reason: non-breaking space!

    • https://docs.raku.org/language/quoting#Word_quoting:_qw says:

      The :w form, usually written as qw, splits the string into "words". In this context, words are defined as sequences of non-whitespace characters separated by whitespace.

    • words does seem to match this description and produce the supposed output with non-breaking spaces as well

    • both can make sense but which one is correct?

16. Bool(function-returning-native-int) regression:
    • bisects to this: https://github.com/rakudo/rakudo/commit/6bd955e6ea131dee6794256f5bf0da8018c3e095
    • tldr the return value was boxed; it turned unboxed but not quite enough to work
17. assignment of native types:
    • a native int cannot be assigned using a string literal, even if that string literal contains an integer
    • a native int is perfectly content with a native str variable on the right handside: voluntaristic integer parsing will happen; even complete nonsense can give a 0 return value
18. parsing of "unary &"
    • say &5 yields (Any) with RakuAST frontend and Nil with legacy frontend
    • say &(5) yields 5 for both, RakuAST frontend sees it as an item contextualizer
19. utf buffers and Stringy
    • utf8 and the likes are Stringy
    • expectation: their Stringy method will return them
    • reality: their Stringy method upgrades to Str
    • is there a practical reason? can it be changed?
20. support for &infix:<~^> on various Blobs
    • in theory, all Blobs can support this operator
    • in practice, utf buffers upgrade to Str, see 19.
    • if there is one item, Stringy is called which doesn't work for most Blob types
    • if there are no items, the default value is the empty string that won't work well with Blob types (probably with any non-Str types)
21. SignedBlob and UnsignedBlob publicity
    • should they remain implementation details?
      • if yes, can we gist them in a way that points it out?
    • if no: document them, spec them
22. (OPENED) Target(SourceSubset) evaluation order
    • seems that the coercion happens before any checks (for the subset) would be performed?
    • correction: multi dispatch tries the source type object if there isn't a :D constraint - according to vrurg, this is an optimizer bug
23. apparently all roles are Cool
    • tested with smartmatch
    • the underlying NQP op reports the same
    • https://github.com/rakudo/rakudo/commit/0035ddb46de48be17492c4a11be3d4070b30077f deliberately introduced it but why for Cool?
24. dependency management doesn't make a lot of sense at the moment
    • the unit of a dependency is a module
    • the unit of sharing code in a managed manner is a distribution
    • a distribution can contain any number of modules and it's identified with metadata
    • the metadata of a module is not taken into account while installation
    • this leads to the bizarre situation that a module cannot be properly identified without knowing the distribution
    • different motives for creating a distribution:
      • install scripts -> won't be depended upon but needs to be installable via dist id
      • release some modules with tightly coupled code -> perhaps there is no point in pretending that one only depends on the module when the rest of the dist could not be separated
      • bundling modules for more convenient sharing -> it would make sense to actually allow independent metadata for the modules
    • in any case - it doesn't seem reasonable to make dependencies a combo of module metadata (name) and dist metadata (auth, version, etc.)
25. (OPENED) tokens and capture groups
    • my token inQuote { \' .*? \' }; say "'Hello World'" ~~ / <inQuote> / matches
    • my token inQuote { \' <(.*?)> \' }; say "'Hello World'" ~~ / <inQuote> / (mind the match delimeters) also matches
    • my token inQuote { \' (.*?) \' }; say "'Hello World'" ~~ / <inQuote> / (mind the positional capture instead of the match delimeters) doesn't match!
    • why? is this intended or a bug?
26. my Real() $demo = 'almafa' throws, yet 'almafa' ~~ Real() succeeds
    • seems to work with a great variety of type checks, possibly all coercion types where the source type is okay
    • given example bisects to https://github.com/rakudo/rakudo/commit/f2d73287f95ad6fa9eba8856ad15f3e8f9076d6e
27. a Pair value cannot be bound to a positional in a subsignature
    • my $default = 12 => 34; my :($pair) := ($default,) will fail: Too few positionals passed to '<unit>'; expected 1 argument but got 0
    • succeeds with non-Pairs
    • even binding the right handside to a Slip won't help
    • possibly related to rakudo/rakudo#4534
28. say: "d", 5 - why is that a Label?
    • reading both https://docs.raku.org/language/control#LABELs and especially https://docs.raku.org/type/Label, it seems that Labels are only for loops
    • yet the parsing shows that we created a perfect label for an expression
    • either the docs are wrong or this should be an error
    • should be easy to ban this in RakuAST
29. use 6.e.PREVIEW; if 1..0 { say 'This shouldn't be True.' } fires
    • seems like if uses an opcode for coercion and that doesn't respect the setting
    • mind you: so 1..0 would have been False