[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: portability of .fas files is questionable

Joerg Hoehle wrote:
> I agree that having portable .fas files would be a nice thing to have,
> but what scheme could be used to avoid problems with machine-specific
> macros that get compiled into one's code? As an example,
> USER1.LSP:with-keyboard is extremely different between the different
> machines.
> [...]
> Was the thread about portability only between UNIX version?

No, not really. As always, `portability' is a term denoting a
continuum, not a few (two) discrete points. The question was ``how
much portability could we achieve with .FAS files'' with resonable
effort and reasonable price (in terms of performance and storage, for

When we access the operating system below, in particular, when we do
it with forms that are not in any way standardized, we must expect to
take a deep look if we can transport the respective source to another
platform. That's one region of the continuum above. But another
region are programs that use only CLtL2 code and run on a (roughly)
POSIX.1 compliant system. (Note, that this term shall include DOS.)
As an example, I've written an SGML backend in Lisp, and that backend
does not use any system-dependent stuff. So my question was `what are
the differences in FAS files for such kind of sources'.

Actually, I got the answer, and it satisfied my curiousity. Now I
know that I must not distribute FAS files for convenience, but the
installation process must compile the Lisp files. It makes packing
distributions a bit harder for me; but that is just an annoyance, not
a problem.


Joachim Schrod			Email: schrod@iti.informatik.th-darmstadt.de
Computer Science Department
Technical University of Darmstadt, Germany