> I agree (obviously) that documents should be validated. However, you
> are suggesting that this should be done via the browser. This seems to
> be more related to the authoring task. A browser user can do nothing
> about a non-conforming document unless they are also the owner.
Until everyone is using full wysiwyg authoring tools, people are going
to continue using text editors. They then run the browser to check the
results - at this point the browser should inform them that they have
screwed the syntax! So the practical approach is for browsers to flag
whether documents conform or not to the DTD!
> Additionally, it would be a waste of effort for an HTML "lint" to be
> developed. There are a several public domain SGML parsers available and
> these should be used to perform document validation.
You seem to forget that the objective is to help users as much as possible.
A custom lint program for HTML would incorporate models of common misunder-
standings and can offer dramatically better feedback than a vanilla SGML
parser. Check out books on compiler technology, e.g. "Understanding and
Writing Compilers" by Bornat. The tough part of writing a compiler or lint
program is providing really useful feedback when the source file is wrong.
Dave