SLATEC and FFI



>>>>> "Ray" == Raymond Toy <toy@rtp.ericsson.se> writes:

    Ray> I've been hacking a bit on f2cl, and it's now capable of keeping track
    Ray> of enough information about each converted function to be able to
    Ray> generate a foreign function interface.  Thus, f2cl could potentially
    Ray> automatically generate the appropriate FFI magic and also a Lisp
    Ray> wrapper around the function for use with maxima.  Still need to write
    Ray> the code to actually do it, but it's conceptually straightforward.
    Ray> I'd need help in writing the FFI for other Lisps, since I'm really
    Ray> only familiar with CMUCL's FFI.

    Ray> This is rather nice.  Platforms with an FFI get fast routines, and
    Ray> those that don't at least still have access to them.

    Ray> What are the plans for the release?  Is this something we would want
    Ray> for that or should we wait until after release?  Perhaps we should
    Ray> wait until after the release since this has the potential of
    Ray> destabilizing maxima enough to crash it.

Have you looked into UFFI?  I know nothing about the details, but some
people have attempted to provide a "Universal Foreign Function
Interface" that is portable across (most) implementations that support
FFI.  From what I can tell, it's a good start, but doesn't attempt to
solve the hard problem of foreign interface (interpreting C header
files, which wouldn't be a problem for your intended use).

Here's the URL: http://uffi.med-info.com/.

Liam