Many of the languages (if not most) that are planned to be implemented above Parrot are lexically-scoped. The methodology to implement such behaviour on top of an otherwise non-LS VM was described by Abelson and Sussman in the book "Structure and Interpretation of Computer Programs":
http://mitpress.mit.edu/sicp/full-text/book/book-Z-H-21.html#%25_sec_3.2 and http://mitpress.mit.edu/sicp/full-text/book/book-Z-H-26.html#%25_sec_4.1 If I find some time, I'd like to implement a PMC for such Variable Frames for Parrot, so lexical scoping will be more optimized and easier to implement. The way I see it frames should support the following primitives: 1. Frame_New (parent) - creates a new empty frame whose parent frame is parent. 3. Frame_Ref - References a frame again. 2. Frame_DeRef - dereferences a frame (possibly recursively) 3. Frame_Define - Defines a new variable in the frame (akin to (define ... ) in Scheme or "my $var = " in Perl) 4. Frame_Set - Sets an already defined variable. (akin to (set! ) in Scheme or a simple "=" in Perl) 5. Frame_Undef - Perl's undef() operation. 6. Frame_Get - Gets the value of a variable. This gets a bit messier when we assume that variables can be referenced by either their names or IDs and that they can store either INTs, strings or other PMCs. Nevertheless, I'd like to try and write an ad-hoc implementation for Parrot. Does this correspond with the general Parrot philosophy? Any other objections? Regards, Shlomi Fish -- ---------------------------------------------------------------------- Shlomi Fish [EMAIL PROTECTED] Home Page: http://t2.technion.ac.il/~shlomif/ Home E-mail: [EMAIL PROTECTED] He who re-invents the wheel, understands much better how a wheel works.
