2009/6/11 Juan Jose Garcia-Ripoll <juanjose.garciaripoll@...>:
> Regarding the issue of maintainers, I want to remark that other people
> had commit rights in the past and that I am more than willing to
> accept regular patches and share maintainance. It is just that there
> are other sexier problems to work with (Clojure, etc) and not too many
> people are interested or need to get involved in such a basic job.
Clojure is nice, but having full featured CL implementation with
friendly license is also very important.
So far each one has its own set of features and problems, and there is
no de facto implementation "to rule them all".
SBCL - Closest to de facto standard. Every lib is tested on it. Non
Linux ports, esp windows are not as good.
CCL - One to watch for. Full feature set supported on all major OS-es,
Nice library support. Unfortunately, requires SSE2 (doesn't work for
me, and other athlon cpus)
ECL - Good support for all major platforms. Clean and has much
potential for improvements. Poor library support.
ABCL - Lot's of momentum lately, access to java libraries. Even poorer
CMUCL - ?? Deprecated in favor of SBCL?
CLISP - Recently resurrected development. Was in major league once.
GCL - ?
> A possible solution to this problem would be to merge implementations,
> but the right way. I.e., not merging ECL into GCL, but rather the
> opposite, or even merging ECL and SBCL. Say, for instance, that you
> take SBCL and build it on top of the ECL runtime using C as a possible
> backend for the generated code... :-)
Merges are always interesting, GCL -> ECL might be interesting, given
common ancestor, but others are not likely because of big differences.
What might be interesting is to merge stdlib for all major
implementations. That way, implementers will only have to take care of
core, and reuse each others work wrt common libraries (network,
threads, Unicode, streams, CLOS, ....).
Also, while we are talking, it might be cool to replace emacs core
with ECL, implement elisp on top of CL, and have a killer lisp
platform, a modern version of Lisp Machines