Re: For literals: why I changed my mind

keshlam@us.ibm.com wrote:

> >> I believe that this mechanism was first suggested by Henry Thompson
> >> and used in the dtd for schema. I got it from the HTML WG.
>
> >Do you think it's sufficiently nonobvious to deserve mention in some
> >non-normative place?  (I do.)
>
> I think it's sufficiently nonobvious that I'm tempted to say it should
> quickly be swept back under the rug. Parameter entities are a _disaster_
> from the point of view of trying to develop a data model for the DTD;
> they're almost impossible to do a reasonably intelligent read-edit-write
> cycle on without reparsing after each edit. They're really an embedded
> preprocessing stage -- and are thus both as seductive for the user and as
> annoying for the maintainer as previous preprocessors (cpp, m4, etc).
>
> I've used this trick. I hold my nose every time I do so.

Well, you were making the point that it's possible to reuse DTDs, so that a
single DTD can be used for namespaces with different prefixes.   What do you
think would be a better approach?

Paul Abrahams