Could we add a low-level PMC interface for subroutine-like PMCs?

[Tyler Curtis]

Oops. I used [0] and [1] and forgot to put the footnotes:

[0] should have been http://www.parrot.org/content/past-optimization
[1] should have been http://www.parrot.org/content/past-pattern-matching

On Sun, May 16, 2010 at 2:55 AM, Tyler Curtis <tyler.l.curtis at gmail.com> wrote:
> Currently, there are low-level PMC interfaces(is there a name for
> them) that PMCs can provide for "array"s, "boolean"s, "event"s,
> "float"s, "Handle"s, "integer"s, "library"s, "scalar"s, "socket"s, and
> "string"s. There is no "sub" low-level interface for PMCs that support
> being invoked, even though there are 11 different core PMCs that have
> custom invoke methods. I suppose no one has needed one so far. I'll
> soon be implementing PAST traversal[0] and pattern matching[1]
> libraries for use in writing optimizations. Something that would be
> very useful in PAST::Pattern would be a way to detect that a PMC can
> be invoked. A "sub" low-level interface for PMCs that can be invoked
> would greatly simplify some aspects of its implementation.
>
> Should we add such an interface? Is there any reason not to?
>
> --
> Tyler Curtis
>



-- 
Tyler Curtis