On Tue, 5 Sep 2000, Jeffrey Goldberg wrote:
> > Can I make a suggestion ? Allow !( ... ) constructs, or some such, in
> > lists.
>
> I haven't looked at the configure parsing code, but my intuition[1] is
> that this what change something that can be done with a finite state
> parser to something requiring a context-free parser. While such a
> construct would be useful, I think that it massively complicates the
> parsing of the configure file.
This issue has come up before and is number 124 on the Wish List. You
are correct in believing that it would complicate things. At present
there is a simple function for splitting things up, based on ':' (or
other) separators, and each item is then handled independently. Having
to cope with continuing state adds complication.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
