I guess that I should not have read the draft while trying
to fall asleep. (I did much better at the latter.)
OK, So F2K probably has everything that we need, but 5 years
too late for practical purposes. If we go the route of the
preprocessor, I'll definitely push for making the client code
look as much like F2K code as possible. I'll leave it up to
others around here as to whether the class specifications should
look more like fortran or C++. As you say, the preprocessor
(or its cousin) should be able to spit out pure F2K code
for later on. I feel that such an extension will be relatively
trivial by the time one extracts and parses the information
contained is a class declaration.
Cheers,
- Tom
>>>>> "Van" == Van Snyder <[log in to unmask]> writes:
Tom.Clune> ... my guess is that it [Perl] would be a more
Tom.Clune> suitable language for the preprocessor than fortran per
Tom.Clune> se. Another responder suggested using m4.
Van> I give students in my compiler classes a
Van> compiler-construction framework written in Pascal, Modula-2,
Van> Ada 95 or Fortran 90. The ones in Ada and Fortran 90 are
Van> tied, in my opinion, for cleanliness. If you write the
Van> preprocessor using ad hoc search-and-replace methods instead
Van> of using the usual systematic compiler construction methods,
Van> Perl or m4 might be more appropriate. William Clodius has a
Van> parser for Fortran 90 at
Van> http://hometown.aol.com/wclodius/index.html. There's also a
Van> more substantial front-end at
Van> http://ftp.digital.com/pub/plan/eli/Examples/.
Tom.Clune> ... Do "extended" objects inherit methods from the
Tom.Clune> parent object, or just the components?
Van> The current F2K design inherits "methods." They're called
Van> "type-bound procedures" in the draft.
Tom.Clune> ... by the time I implement a preprocessor of this
Tom.Clune> power, converting it to accept a different syntax (F2K
Tom.Clune> <-> C++) will be relatively easy....
Van> I hope you're also planning to make a preprocessor that will
Van> accept the old syntax as input, and emit the new syntax
Van> (perhaps pure F2k) as output. Or would you rather ask your
Van> clients either to throw away their "legacy" code or re-write
Van> it?
Van> -- What fraction of Americans believe | Van Snyder Wrestling
Van> is real and NASA is fake? | [log in to unmask] Any
Van> alleged opinions are my own and have not been approved or
Van> disapproved by JPL, CalTech, NASA, Dan Goldin, Bill Clinton,
Van> the Pope, or anybody else.
--
--
Thomas Clune, Ph.D. Parallel Applications Consultant
SGI [log in to unmask]
Code 931 NASA GSFC 301-286-4635 (work)
Greenbelt, MD 20771 301-286-1634 (fax)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|