method definition syntax in defsystem forbids primary methods

[Robert P. Goldman]

You can define some methods on components inline, but these can only be done if you supply a method qualifier. That means that we can't use inline definition for primary methods. It would be a nice enhancement to extend defsystem parsing to make the method qualifier optional.

I haven't thought hard about how difficult it would be to correctly handle this. Possibly we could just allow the use of :primary as a bogus method qualifier.

[Faré (fahree)]

I move we kill this "inline-method" syntax.
It cannot make defsystem more declarative in any meaningful way, and just makes my life harder.

Now to identify everyone who uses it...

[Robert P. Goldman (rpgoldman)]

On 1/19/13 Jan 19 -2:38 PM, Faré wrote:
> I move we kill this "inline-method" syntax.
> It cannot make defsystem more declarative in any meaningful way, and just makes my life harder.
>
> Now to identify everyone who uses it...
>

I would really rather not do that, unless it makes your life
*significantly* harder. Isn't it trivial to parse it in the DEFSYSTEM
macro?

I'm just concerned because I know of a bunch of places it is used
internally to SIFT, and not in Quicklisp (so won't be found by your tests).

I think it does serve a real need, since it's much more terse than the
EQL method dispatch alternative.

[Faré (fahree)]

Had too much time on my hands, so I fixed that in 2.28.4.

I'm not convinced that this inline-method thingie is a good idea, but if it's going to be there, it might as well be done well. Feature still done in backward-internals.lisp.

[Faré (fahree)]

Released in 2.29.