On Wed, Jun 5, 2013 at 12:50 AM, <[email protected]> wrote: > The following proposal combines all the advantages of both the > R5RS and R6RS semantics, while avoiding all the disadvantages > mentioned by John Cowan for either R5RS or R6RS semantics. > This proposal would also enable boxing/unboxing optimizations > that John Cowan didn't mention because they aren't feasible > with either the R5RS or R6RS semantics. > [...]
Although this breaks a lot of my (and existing IEEE Scheme) code, I'm not opposed to this compromise in principle, if it actually improves the situation. However, we had discussed this before and dismissed it. Perhaps we were missing something. As I see it, once a procedure escapes, the existence of any semantics in the language which can discriminate the procedure location requires it to be boxed. This is true whether the discriminator is eq? or eqv?. So it doesn't seem this allows the desired unboxing optimization. -- Alex
_______________________________________________ Scheme-reports mailing list [email protected] http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports
