D. Tweed writes:
> Unfortunately the last time I looked at CWEB it still had the same idea
> about identifiers as WEB itself, namely that if they match textually
> they are the same, which makes the automatic indexes produced less useful
> for C++: I've got twelve classes all with a `read()' method plus any in
> the standard libraries and lumping them all together is less than
> optimal. I'd imagine the same problem would occur for haskell with type
> class members. Of course, solving this would require at least
> type-deduction info to be passed to the literate programming system, which
> is why I'm a little sceptical of the language independent implementations
> of literate programming. (Of course, the indexes only matter if like me
> you prefer reading hardcopy or work on trains, etc.)
I see the problem for a language like C++, which has true overloading, but not
for Haskell. In Haskell it makes sense to identify type class methods with the
same name, since they are semantically related. You don't need to do any type
inference to index them; that only works when the instance is statically
decidable anyway.
The problem that remains is handling scope correctly, but that must be a
problem for C/C++ as well. Here you could make a good argument for
language-dependent literate programming systems.
--
Frank Atanassow, Dept. of Computer Science, Utrecht University
Padualaan 14, PO Box 80.089, 3508 TB Utrecht, Netherlands
Tel +31 (030) 253-1012, Fax +31 (030) 251-3791
