Front page | perl.perl5.porters |
Postings from November 2021
On Mon, 29 Nov 2021 16:08:54 +0000
"Paul \"LeoNerd\" Evans" <leonerd@leonerd.org.uk> wrote:
> Far less clear is how to do parts 3 and 4. In particular, for part 3 I
> have the following thoughts on possible ways to implement it:
...
Oh I just thought of a 4th way:
3d. Have pp_entersub perform much more of the work; effectively roll
up the behaviours currently performed by OP_ARGCHECK and
OP_ARGELEM into it. It would involve annotating the CV with some
more metadata about its signature params - the counts and slurpy
value (i.e. all the data stored by the OP_ARGCHECK op) and also
the pad indexes of every recognised parameter. This could be used
pp_entersub as well as any other metadata introspection abilities
that might be added later.
This would allow pp_entersub itself to copy elements out from the
stack directly into the pad slots of the new CV, where the
OP_ARGELEM ops would have put them. They'd go directly from stack
into the pad, not via the @_ array then. Once there are no more
argument values on the stack, any optional parameters that
haven't yet got their values can be assigned them by jumping to
the first OP_ARGDEFELEM op for a missing param.
An upside is this would be even more efficient than all of the op
calls down the optree per element. A downside however in that it's
notably less flexible than the current setup. It would also
totally break the way that Object::Pad's `method` keyword
currently works - so I'd be somewhat in favour of not picking
this approach. ;)
--
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