Front page | perl.perl5.porters |
Postings from January 2022
On Wed, 26 Jan 2022 at 00:40, Ovid via perl5-porters <perl5-porters@perl.org>
wrote:
> Add a module keyword with basic, modern features to support common Perl
> use cases.
>
First off, the idea is interesting but I disagree with the overall concept:
this seems to be conflating "we should improve the way libraries of non-OO
functions are provided" and "let's apply lots of new defaults".
Implementing just the `:export()` attribute seems a much better - and
simpler! - option.
The lack of versioning means you're locked in to a particular snapshot of
Perl features.
We end up with something that applies to these library files but isn't
available in a script... why make that distinction? There's mention of MAIN
and extending this to cover scripts, but...
Wouldn't we just put all that `indirect/multi_dimensional/signatures/etc.`
into `use 5.038` so that it's available in *all* code? We've already
reduced boilerplate to a single line, while making it clear to the Perl
interpreter what behaviour we expect.
There are a lot of key technical details missing here, for example:
- what do these exports actually do?
- is there any introspection?
Is it implementing the same mechanism that people would expect from
Exporter, i.e. populating the symbol table of the target package? If so, is
that guaranteed - can this be used to implement mixins, for example? Or
would it allow for subs that are only visible to the caller?
Does @EXPORT get populated? How about @EXPORT_OK, %EXPORT_TAGS?
Can all subs be marked as exported without having to put the `:export` tag
on each one? If that's the case, what happens to `our $VERSION`, would that
overwrite the caller's copy? Is $VERSION automatically populated by the
`:version()` attribute?
What happens if you have `module X { sub import { ... } }`? Can you push
imports up a level, e.g. something like `module X { use List::Util qw(min);
sub min : export }`?
Why does it have to be a block - can't we have `module X; ...` to avoid
that extra indentation level?
Two minor things:
- you mention "Currently, Amores' syntax is almost entirely
backwards-compatible because the code does not parse on older Perls that
use strict"... then the first example shows that it *does* indeed parse
perfectly happily, those are *runtime* errors
- getting way offtopic, but... the RFC is in violation of the "civility"
core P5P policy: "stick to the facts while avoiding demeaning remarks,
belittling other individuals", even if the person is famous, perhaps not
appropriate to criticise their ego in a core Perl document
>
Thread Previous
|
Thread Next