Re: AW: Parameter entities vs. GI name groups

Weichel Bernhard (K3/EES4) wrote:

> >David G. Durand wrote:
> >>
> >> the harders thing about learning PEs is that the string
> >> substitution model is _not_ correct.
> >
> >Please explain this.  String substitution is
> >precisely what they appear to be.  So, I am missing
> >something and wish tutoring.
> 
> Try this small document with nsgmls. It should work
> if PEs are pure string substitution. But it does
> not work...

Right.  PEs have to respect boundaries, so they are 
not in fact, pure string substitution in terms of a 
search and replace command.  I wouldn't want to use 
PEs that didn't respect the boundaries because it is all too probable 
someone (me) will trash the DTD.  

But neither are they classifying mechanisms semantically.
They *group* in the same way a well-behaved string 
substitution does.  Again, if PEs can be added and keep 
to the goals if not requirements for XML developers, that 
is good.  If that is done to get TEI/DocBook on the 
Web, good for those who use it.  My experience is very 
large and very complex DTDs are the principal barrier 
to the acceptance of SGML.  I think those that promote them 
in XML should be very aware of that and be prepared 
to put a lot of effort into overcoming the 
resistance.   Many shops will be better served by 
subsets of these that are simpler to understand.

len

Received on Saturday, 21 June 1997 11:15:11 UTC