Re: Magic-v2, Clone, and a possible `sv_clone` API function

On Thu, Jun 4, 2026 at 3:54 PM Paul "LeoNerd" Evans <[email protected]>
wrote:

> I've been looking around CPAN for possible modules that might break
> with some of the new "Magic v2" things I am lining up for next
> development cycle. Originally I was looking for things that might get
> upset if we start removing any of the internal v1 magic types, and I
> encountered the Clone module - because it has a comment mentioning
> PERL_MAGIC_arylen. While this itself is just a comment, on a further
> reading of the code in that module it looks like it's actually
> concerned quite closely with a lot of the details of how Magic (v1)
> currently works. This code wouldn't cope well with v2 Magic structures.
>
> My first thought for thinking about Magic-v2 is that a lot of the code
> would need quite a bit of editing and adjusting.
>
> Looking around CPAN more I also see that the code has basically been
> copy-pasted into at least two other dists; Clone-AsUTF8Bytes and
> Class-XSConstructor. This starts to suggest that actually, rather than
> living in a CPAN module, this sort of "sv_clone" API function ought to
> exist at the core perl level itself. That way, it would be able to do
> the right thing with a new version of Magic, or any other newer
> behaviours that get added.
>
> (While I'm there I notice it also doesn't handle SVt_PVOBJ, the new
> "object" SV type that was added in Perl 5.38 as part of the `class`
> feature, so that too would want to be looked at).
>
> Perhaps between us we can manage to find a good solution here.
>
> --
> Paul "LeoNerd" Evans


I can't help noticing that this sv_clone is almost the same thing as sv_dup
(or probably sv_dup_inc), except that the new value doesn't live in a
different interpreter.

Leon