[the current draft] defers too much to the SGML spec
the stuff about "tokenization" needs to be expanded
I'm very glad that you've agreed to do this, even with a delay in the
resulting document. It is very much needed by folks who don't have
a copy of the SGML Handbook or of ISO 8879. However, I would urge
you to take care not to reinterpret the text of 8879 in your expansion.
I would also like to see you elaborate on the following legal constructs
which may appear in an HTML document:
- formal public identifier as found in document type declaration;
e.g., the fact that "-//ietf//dtd html 2.0//en" is not a legal FPI
while "-//ietf/DTD html 2.0//EN" is.
- internal declaration subset
- marked section declarations
- processing instructions
- minimized tags (e.g., unclosed, empty, and null tags)
I would also like to see the spec:
(1) emphasize the importance of the presence of a document type declaration
rather than describing it as essentially extraneous; this is becoming
much more important as multiple DTDs are utilized.
(2) delineate to what extent an internal declaration subset may be used; e.g.,
allowing the presence of internal general entity declarations, disallowing
(semantically speaking) all other declarations which might otherwise appear in
the internal declaration subset, but, which, nevertheless should be parsed
(at least for the purpose of ignoring them).
Some of these may be more a subject for a future version rather than
the current version; e.g., use/restrictions on the internal declaration subset.
However, some mention need be made of them for the purpose of parsing them
should they appear in a document.
Regards,
Glenn Adams