I thought this was taken for granted - that there would be a single RE in
all XML-l[ai]n[gk] specifications. I also assumed (naively?) that POSIX
defined such an RE, and we merely needed an implementation. There might well
be subsidiary questions such as 'do we want to implement a subset', 'are there
any clashes between RE syntax and XML syntax', 'are PEs expanded before
evaluating the RE :-)', etc.
> is such a variance between the RE processors in perl, Tcl, sed, awk, vi, etc.
> that having RE inconsistencies among XML applications would be worse than
> having no RE support at all.
Fully agreed.
>
> If we choose a code base that contains more RE features than the minimal
> set supported by all RE processors, we need to be clear which features
> are part of and required by XML. (This sounds like a mess to me.)
Since XML-LINK-TEI has shrunk since its first airing, I suspect that there
is a desire not to overreach. Certainly we should not force implementors
to have to work hard to comply to unnecessary features. (For all I know
some REs would be sufficiently powerful to act as XML parsers :-)
P.
-- Peter Murray-Rust, domestic net connection Virtual School of Molecular Sciences http://www.vsms.nottingham.ac.uk/