develooper Front page | perl.perl5.porters | Postings from April 2022

Re: Pre-RFC: a `library` keyword

Thread Previous | Thread Next
From:
Paul "LeoNerd" Evans
Date:
April 1, 2022 14:16
Subject:
Re: Pre-RFC: a `library` keyword
Message ID:
20220401151612.5f7fb905@shy.leonerd.org.uk
On Fri, 01 Apr 2022 09:54:37 -0400
"Ricardo Signes" <perl.p5p@rjbs.manxome.org> wrote:

> Item 5 is good.  We should try to eliminate the need for "magic true
> value" in v5.38, generally.

Indeed; I think last time when we discussed the previous iteration of
the idea ("module"), we said that overall everything can be already
solved by the `use VERSION` and `package NAME VERSION {BLOCK}` features
that already exist in Perl and have done for a very long time, apart
from two small additions of:

 * A mechanism (maybe attribute?) to request per-function lexical
   exporting.

 * A named feature that causes the .pm file as a whole to yield a true
   value at the end.

I forget what we called them but those could be spelled:

  use feature qw(export_attr yield_true);

  package Some::Module v1.23 {
    sub this_is_exported :export { ... }

    sub this_method_is_not { ... }

    my sub this_is_lexically_private { ... }
  }

  # no need for "1;" here because of the yield_true feature

That can all work *right now* with the `package` keyword. Furthermore,
the two new features, once implemented, could become part of a
`use VERSION` bundle at some future version of perl, so you could
simply write

  use v5.100;  # has already enabled strict, warnings, signatures and
               #   yield_true; has disabled indirect, etc...

  package Some::Module v1.23 {
    sub f :export ($x, $y) { ... }
  } 


Looking further over the (closed) PR from last time:

  https://github.com/Perl/RFCs/pull/11

What I said was:

  @Ovid I think the :export attribute could stand alone well in its own
  RFC. Do you want to write one up, or should I?

  A couple of further thoughts on it:

   * Perl core already reserves all the lowercase attribute names for
     its own future expansion. There's no need to use feature-guard it.
     We can add `:export` right now.

   * I think the named tags should take a leading colon; e.g.
     `:export(:strings)`. That way they visually match what the
     `use Module ':strings';` import line looks like, and also it gives
     us some space to put other options in there one day.

I would definitely be in favour of seeing (two?) RFCs to specifically
add these ideas:

  1. An :export attribute for lexical exports (see quoted PR comment
     above)

  2. A `use feature 'yield_true'` to avoid having to put "1;" at the end
     of the file; with a future plan to include that by default in the
     versioned feature bundles.

I believe that combination, plus a proper understanding of how
`use VERSION` and `package NAME VERSION {BLOCK}` already work in Perl,
will fully cover the use-cases and benefits of your proposed "library"
keyword, without the need for new keywords or a significant departure
of existing Perl syntax.

-- 
Paul "LeoNerd" Evans

leonerd@leonerd.org.uk      |  https://metacpan.org/author/PEVANS
http://www.leonerd.org.uk/  |  https://www.tindie.com/stores/leonerd/

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About