I still see conceptual problems with "names". I can take one source code containing, say a local variable named X that compiles and runs properly, do a text substitution of X to Y (assumming that the variable Y does not aldready occur) to create an entirely new source code that compiles abd which (usually) results in exactly the same machine code. The "name" I give to the variable X (or Y) does not play any logical role in the program.
You are right, but of course all exported names that are in the API matter. Suppose you want to write an interpreter for Aldor in Aldor (something like B-natural perhaps). Ideally I would expect that that can be done in such a way that it does not matter what the underlying libraries are. There might be even libraries that would be written after the interpreter and the interpreter should treat them as the ones he already knows about.
I guess that is a good example of where you would like to be able to ask an available library which types it contains, ask the domains about their exports, and eventually call some of the functions.
Situation is as follows. You have given libaldor.al (no sources). Build an interpreter by just using that functionality. Then I give you libalgebra.al and libmyalg.al and the interpreter should do what it is supposed to do (basically behave like the foam interpreter (aldor -gloop) but a bit faster and maybe with some nice way of guessing types.
Ralf _______________________________________________ Axiom-developer mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/axiom-developer
