> > I'm also not sure if they are really 100% equivalent, since I'm > > throwing in some extra specification of the order of evaluation. This > > makes it possible to play tricks like using objects that only exist at > > compile time without requiring any special notation for it: the eager > > algorithm guarantees that such objects are fully reduced before code > > is compiled (or raise an error). What would be neat is to be able to > > keep this semantics, but use a more elegant way of formulating it as a > > syntactic operation. (If this doesn't make sense I'd be happy to > > elaborate.) > > Yeah, that makes sense, like a kind of abstract interpretation, you're saying?
Yes. I want a simpler specification of what is happening. Currently there's just a bunch of eager pattern matching rules. This works, but feels a bit hacked-together. In addition I guess what I'm looking for is some kind of type system to express this. The current dynamically typed approach will give type errors at macro (= code generator) execution time, but not at macro compilation time. program time = stage 0 program compile time = macro run time = stage 1 macro compilation time = stage 2. There are essentially two cases: one where compile-time objects need to be completely eliminated by properly combining them, and one where they reduce to some 8-bit tag that represents a remnant at run-time. I.e. I want to do "1000000 30000 /" on an 8-bit machine, with the compiler catching errors when the result doesn't fit or the operation can't be reduced when these generators _themselves_ are compiled, not just when they are applied. It seems this will be a simplification of the MetaOCaml type system - one that doesn't need to solve name binding issues. ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Factor-talk mailing list Factor-talk@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/factor-talk
