Leopold Toetsch <[EMAIL PROTECTED]> wrote: > Jeff Clites <[EMAIL PROTECTED]> wrote:
>> But that loops back to a previous proposal of mine: If they're not >> being preserved, and in fact need to be "synced" between caller and >> callee, then having these registers physically located in the >> interpreter structure, rather than in the bp-referenced frame, saves >> all the copying, and makes it more obvious what's going on. > Well I answered that already. Having two distinct addressing schemes for > volatile and non-volatile registers has a serious overhead for > non-prederefed run cores. Err, for all but unrolled run-cores (i.e only JIT could cope with it). For prederefed cores all OUT arguments would need duplication, IN arguments, which have usually a constant addressing too would use the addressing of the constants for the volatiles. CGoto and plain function core would grow towards insanity. > ... OTOH in the light of a recent discussion this > approach could be an alternative. So not really, sorry. >> JEff leo
