Re: A simple guy with a simple problem

[Tim Bray <tbray@textuality.com>]

At 04:02 PM 14/03/01 +0000, Sean McGrath wrote:
>Hello, my name is Bob and I'm a programmer.

Some responders have missed Sean's point; I highlight
it for reference:

>Here is the DTD:
>         <!ATTLIST lit text CDATA "STUFF">

Sean is worried that a nonvalidating processor won't
fetch the DTD & therefore won't apply the default 
"STUFF" value of the "text" attribute to the 2nd
<lit> element in his instance, and thus won't
make the required substitution.

I.e., validating and non-validating processors
will produce different results. 

First of all, in this scenario, you really should
have an XML declaration with standalone="no" as a 
hint that the DTD's going to change the document, so
the example already contains de jure bad practice.

The deeper and de facto bad practice is in having a 
production system rely on defaults out of an external 
DTD subset at all.  Sean gives a good example of why 
this is the case.  Avoid this bad practice and you're
fine.  -Tim