RFC: OO system - CLOS or other? about jscl HOT 3 CLOSED

We should try to support CLOS eventually, because JSCL is CL, but other system can go in a contrib or extensions if we consider it appropiate.

from jscl.

If it helps, I've been working on a library for doing generic functions that work a lot like CL's, but it's a work in progress. They're more meant to mix with the way JavaScript works, too (flexible function arity, dispatch across all arguments at call time instead of a set of predefined primary arguments). I also implemented Sheeple in CL itself, back in the day. That might be loadable/editable as a potential object system, but it won't map transparently to JS objects, since, for example, Sheeple supports multiple inheritance.

There's also the possibility of loading up Closette, which seems to have a fairly permissive license. This is a good chunk of CLOS, and it was implemented as part of AMOP, so it has some MOP support iirc. I assume this is the best bet for having an actual CLOS implementation, and I think some existing Lisps start off with Closette.

Another alternative is to try and yank the PCL (Portable Common Loops) code out of somewhere and hack it until it runs, but the PCL implementations I know of are fairly sizable and honestly pretty scary codebases to work with. This is the "portable" CLOS implementation that SBCL uses.

I guess the question is whether you want an OO system that can integrate tightly with JavaScript objects, or a system that hides the fact that JS exists, or something in between.

from jscl.

Implemented in the closette part, as the most common version of OOP for Lisp

from jscl.