Matt Fowles <[EMAIL PROTECTED]> wrote: > I like the idea of mandating lexicals vars. This would also eliminate > the need for spilling (I think), as the register allocator would only > need to refetch the lexical rather than save it off somewhere to be > restored later.
There are two issues: yes with refetch - no with store (probably). Lexicals and globals are references, so: add Plex, Px, Py stores already x+y in the variable lex. Well, that's a compiler issue and a reason to keep the current "destination exists" semantics of opcodes. (This usage doesn't cope with the generation of new values, though and morping "Undef" isn't a good solution) > I get the feeling that this is equivalent to requiring exception > handlers to be a locally defined closure, which is another way we > could go about this. Yes. That solves it. OTOH going all along with lexicals could be pretty inefficient. > Matt leo
