>>>>> ">" == Rogers, Raymond writes:
>> Could I suggest working on a import function instead of
>> individual imports. With a stretch; Maxim should allow an
>> abstraction layer above individual expressions or functions.
>> The idea would be to: examine the source file, convert it, then
>> write a linkage to it. I realize that full automation is probably not
>> possible, but I would think that would be the goal and organizing
>> principle.
This is possible. In general, it's kind of hard to know what the
Fortran routine F(A,B,C,D,E) wants as inputs and outputs, and which
are both inputs and outputs. f2cl has some smarts to figure out some
of these, and tell you that D is an output, and E is an input. It's
not smart enough to know that C is both an input and an output.
And once you've figured it out, then what? What do you want exposed
to maxima? Does it have meaning to maxima?
This idea only works if all parameters are inputs, and F returns
exactly one value. Anything more complicated means doing it by hand.
I don't think there were too many slatec functions where the calls are
so simple. It certainly wasn't true for the Bessel functions.
Most of the work, really, is figuring out what to expose to maxima,
and then making it happen. Everything else is easy. :-)
Ray