On Wed, 4 Mar 2009, Timothy S. Nelson wrote:

# Method descriptor objects: .name, .signature, .as, .multi, .do.

       In Callable.pod

       .name is on Routine
       .signature is on Code
       .do is on Code (I'm guessing here)

       I'm not sure where the others go:
.multi must be on something that gets it into Routine
.as must be on something that gets it into Method (higher?)

See S12:1987 for this.

Are these line numbers you're using? POD line numbers? I don't see a line 1987 in this (the file isn't big enough).

I still believe that must be interpreted as
there being a specific MethodDescriptor object on which all of .name,
..signature, .as, .multi, .do are attached.

       I'm guessing that .^methods should really return a bunch of Method
objects.

No, MethodDescriptor objects. See S12:1987.

Hmm. I don't like it :). You may be right, but I was assuming that we were returning *Method* objects.

I asked $Larry about this on IRC; specifically I asked whether we were talking about a MethodDescriptor object or an Array of Method. His answer:

that's one of the things I was gonna let the implementors flesh out

feel free to speculate and label it so :)

just keep it as simple as possible, but no simpler...

In light of this, I'd argue that we should do it as an Array of Method, until we have a reason to think otherwise. I'm expecting, of course, that if the implementors think this is a bad idea, they'll chime in and say so :).

Furthermore, I'd argue that if the specs don't name a particular object type (ie. "method descriptor" isn't, "MethodDescriptor" would be) then we're free to try to figure it out ourselves.

# Attribute descriptor objects: .name, .type, .scope, .rw, .private,
.accessor, .build, .readonly.

       I wonder what type would be appropriate for an attribute descriptor
object.  I haven't done anything with these either.

See S12:1999.

        Only has 1764 lines in my editor.  And wc says 1763.

I presume you're implying there should be an AttributeDescriptor role, yes?

...I'm not sure I agree that there should be an AttributeDescriptor role.

        :)


---------------------------------------------------------------------
| Name: Tim Nelson                 | Because the Creator is,        |
| E-mail: wayl...@wayland.id.au    | I am                           |
---------------------------------------------------------------------

----BEGIN GEEK CODE BLOCK----
Version 3.12
GCS d+++ s+: a- C++$ U+++$ P+++$ L+++ E- W+ N+ w--- V- PE(+) Y+>++ PGP->+++ R(+) !tv b++ DI++++ D G+ e++>++++ h! y-
-----END GEEK CODE BLOCK-----

Reply via email to