From: Christophe Rhodes
Subject: [Call for Testing] MOP in SBCL 0.9.16
Date:
Message-ID: <sqy7t7wl22.fsf@cam.ac.uk>
I spent some time over the last couple of months modifying SBCL's MOP
implementation to make it closer to the description / specification in
AMOP, and to eradicate known issues. SBCL 0.9.16, released last week,
contains the results of this work: one result is that the
Closer(-to-MOP) project's MOP feature tests (at
<http://common-lisp.net/project/closer/features.html>) now report no
failures, though I should say that I don't think those tests are
intended to be exhaustive or tricky.
That said, I'm sure that there remain bugs, suboptimalities, and other
problems with SBCL's implementation of the MOP (in addition to the
fact that the MOP described in AMOP was never intended as the final
word in MOPs for Lisp...). So I'd like to encourage people with MOP
code, particularly with code using less-explored areas
(e.g. MAKE-METHOD-LAMBDA, customizations of COMPUTE-EFFECTIVE-METHOD,
uses of the dependent-object protocol, non-standard subclasses of
SPECIALIZER...) to give it a try in SBCL 0.9.16. Known deviations
from the MOP of AMOP are documented at
<http://www.sbcl.org/manual/Metaobject-Protocol.html>; reports of
other problems (or just complete successes; who knows?) to the
development mailing list linked to at <http://www.sbcl.org/> would be
appreciated.
I'd also be interested to hear if people have ideas for improving on
the AMOP MOP in the CL context, and if so whether those ideas can be
implemented compatibly with the existing MOP. (That discussion
probably should be had in this forum rather than one implementation's
development mailing list).
Christophe