Re: [Sbcl-devel] little introspection functions

[Christophe R. <cs...@ca...>]

Luke Gorrie <lu...@me...> writes:

> Hi guys,
>
> David Lichteblau <da...@li...> writes:
>
>> Quoting Daniel Barlow (da...@te...):
>> > What are you expecting it to do with a generic function anyway?  The
>> > defgeneric form?  First method?  Nth method?
>> 
>> Currently, nothing in particular.  What about
>>   - for a generic function object: locate the defgeneric form
>>   - for a MOP method object: locate the defmethod in question
>
>>From the SLIME perspective, we would like a way to enumerate every
> method of the generic function. We'll then jump you to the first one
> (or to the generic function itself first perhaps), and pop up a little
> buffer listing the others -- and let you cycle through with
> `C-M-.'. We don't do this for generic functions today, but we do for
> xref results and plan to use the same interface for generic functions.

Well, this bit sounds easy enough to do with basic MOP functionality;
the accessor <mop-package>:GENERIC-FUNCTION-METHODS gets you a list of
the methods of a generic function, and
<mop-package>:METHOD-SPECIALIZERS the specializers of the particular
method.  (<mop-package> is SB-MOP for sbcl).  Of course, if you plan
to support non-MOP implementations, you'll want to encapsulate this in
your own interface.

> For each method we'll want to be able to get (off hand):
>   Filename it's defined in.
>   Character position of the definition in that file.
>   Argument types for that method, for the summary buffer.

This indicates the need for some kind of two-way protocol on the
source location information.  If you evaluate one form from a file
buffer, the lisp implementation needs to be told where the form is
from, right?  Otherwise it can't possibly guess where it should claim
the form is defined.

Cheers,

Christophe
-- 
http://www-jcsu.jesus.cam.ac.uk/~csr21/       +44 1223 510 299/+44 7729 383 757
(set-pprint-dispatch 'number (lambda (s o) (declare (special b)) (format s b)))
(defvar b "~&Just another Lisp hacker~%")    (pprint #36rJesusCollegeCambridge)