mb> I actually go further and claim that call/cc is too difficult mb> even for your run-of-the-mill XEmacs implementer, like me, to mb> understand. So we'll let Michael understand it. (Then we put him into protective custody and don't let him out until we have an alternative maintainer. :-) This is what happens with all complicated subsystems, like it or not. I don't claim to understand call/cc, and I don't think I _will_ understand it soon. But I don't think full understanding is necessary; only those aspects that get _used_ in core code. Package maintainers will have to understand their own code. However, somebody who does understand it will need to document its uses in core XEmacs code carefully. Eg, who really understands Mule? Hrvoje doesn't claim to AFAIK---but he knows how to write code around it, and even wrote the tutorial for that. The questions are "How useful might call/cc be? Do we have implementor(s, preferably) on board who can write code to take advantage of it? Once that code is written, how useful will it be to other implementors, package writers, and users?" Then, of course, we aggregate up all such features (and disfeatures) for Scheme and features (and disfeatures) for CLisp, and features (and disfeatures) of sticking with the present engine, and compare. mb> Maybe I'm in need of some 180 degree neck-twisting. Please, no! Not before you overhaul ./configure. (Yet again! Many thanks for past efforts.) -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +1 (298) 53-5091