Re: those predeclared entity refs

In message <libSDtMail.9703261102.27801.altheim@mehitabel/jurassic> altheim writes:
> I really do think it's time we began discussing *conformance levels* so
> that people can build lightweight apps, intermediate apps, and the big
> monster processing apps that do TEI-like linking, DSSSL-o stylesheets, etc.

I agree completely.  The variety of discussion and approaches on xml-dev 
suggests that we cannot have one-suits-all tools.  The mythical CS grad
student has now spent a mythical month discussing parsing.  Of the two 
parsers, one has &lt; hardwired and the other tells me that 'lt'
is undefined :-)  I can work happily with either, but not easily with
both.  And I'd prefer not to have to edit all my old files to add
ENTITY at the start.

The simple minded approach that I would take is for a parser to have 
command-line switches that activate different levels of conformance.  
-e 'pretend that <!Entity amp '&'>  (etc.) are included in the declaration
-isolat1 'pretend that <!Entity isolat1 PUBLIC ... > is included'

There has been discussion about RFCs firming up existing practice.  If we do
not mandate practice in some areas, then we should have a mechanism for 
encouraging it (perhaps with a view to firming it up later).  There are
probably standard methods for this, but how about 
'Generally Accepted Conventions'
as something that is published alongside XML material but without the specific
accreditation of the WG/ERB.  For example, if the draft says nothing about
&lt; etc., then at least there should be something stating that almost
everyone thinks it's a good idea to use them in a standard manner.  

Other areas for GAC could be tag sets (I see this as a growing problem for XML
where people want to mix and match from (say) the Math DTD, HTML3.2, etc.).
It's probably not soluble, but we want to avoid some of the worst excesses
that have happened with HTML.  We must remember that if/when XML is in
widespread use, most of the authors will never have heard of SGML.  Yet another
area is URL specifications and the choice of XML-link names.  If the lead
is set with good examples, then a lot of people will work from those.



Peter Murray-Rust, domestic net connection
Virtual School of Molecular Sciences