[slim-vim] Slim vim design review

[Larry Clapp]

On Mon, May 22, 2006 at 01:58:24PM -0700, Brad Beveridge wrote:
> Up until now I've just been hacking on the slime-vim code with
> little regard for consistency, design or big picture stuff.
> Hopefully it isn't too horrible, but now is definitely the time to
> get some design goals down on paper.  I know you're all dying to
> start helping, but you'll just have to hold off for a bit until we
> do some refactoring :)
> 
> So here are some thoughts I have on our goals:
>  - Primarily we are here to develop Slime support in Vim
>  - Secondarily our code ought to be easy to pick up and put into
>  other like-minded projects (ie, they want to use Swank for whatever
>  reason)
> 
> With these goals in mind I think we ought to logically partition the
> code into 3 segments
>  - A Vim specific part
>  - An ECL specific part (sockets, etc)
>  - The majority in plain Common Lisp
> 
> Any Vim script parts fall under the Vim specific part.  I think that
> CLOS is the obvious answer, but IMHO that is really a runtime
> solution.  Swank's definterface/defimplementation (see
> swank-backend.lisp) appeals to me, it also uses generic functions,
> but really all you do is specify an interface that the pure Lisp
> code can rely on.  I'm leaning toward this approach.
> 
[snip]
> All suggestions are welcome.

I don't have any suggestions, but this approach sounds good.  I like
and agree with the secondary goal.  The J/ABCL and Climacs folks seem
to've done a bare minimum of work on subsets of Slime-for-CL (and
different subsets, too), so it'd be nice if we could a) help them out,
and b) leverage their userbases when we get more working code.

Keep up the good work, and keep the status reports coming!

-- Larry