On 17 Jul 1998 at 09:11:43 +0200, sperber(a)informatik.uni-tuebingen.de (Michael Sperber [Mr. Preprocessor]) wrote: I did, in fact, lose track of the context of your remarks, for which I apologize. I knew you had never said that, but that wasn't the point I wanted to make anyway. To make that point, I'll have to digress for a moment. That is absolutely correct. But, are we talking about replacing the elisp substrate with *any* [Common Lisp/Scheme] engine, or with *some particular* engine? If the former, then you have to program against the lowest common denominator which, in the case of Common Lisp, does not include tail recursion elimination. My impression, though, is that the latter is intended. In that case, if the intention is to provide a Common Lisp substrate and a Scheme emulation on top, then it would be wise to pick a Common Lisp implementation that does tail recursion elimination. In that case, I see no reason why you could not provide a Scheme emulation that also provides tail recursion elimination. Note that this is independent of whether extension programmers take advantage of the TRE capabilities of the Common Lisp substrate or not (i.e., whether they program for general Common Lisp implemenations, or optimize for the specific engine chosen). So, the point I was trying to make with my previous message is that Common Lisp implementations that do TRE actually exist. If Common Lisp is judged to be a desirable replacement for elisp, then one could identify TRE as a necessary (or at least highly desirable) property of the specific implementation chosen. Two more things, in random order: (1) I am still interested in a further clarification of call/cc vis-a-vis lexical closures. (2) Do not consider me, by any means, a Common Lisp proponent. I am merely trying to be helpful by providing some bits of information that I happen to have in my head. I share the concerns of some others, namely that: (a) Free CL implementations seem to be having a hard time moving from CLtL1 to CLtL2, much less to full ANSI compliance (and this is true of CLISP in particular); and (b) CL may be too heavyweight for an XEmacs implementation. Regards, -- Jerry James Email: jerry(a)cs.ucsb.edu WWW: http://www.cs.ucsb.edu/~jerry/