Re: Revised language on: ISO/IEC 10646 as Document Character Set

Dan Connolly (
Sat, 6 May 95 19:15:54 EDT

Terry Allen writes:
> Dan:
> | Terry Allen writes:
> | > >I actually make ISO10646 a binding constraint without putting it
> | > in the public text (the SGML declaration). See what you think:
> | > That's bogus. We need the SGML declaration that goes with your language.
> | This is argument by assertion.
> | Why do we "need the ..."
> So as to construct a valid SGML entity for parsing, a goal you have
> yourself enunciated for MIMESGML. Yes, I assert that I need an SGML
> decl in order to parse HTML.

You've got one. It has ISO-8859-1 as the document character
set. That's all that's required. Parse away!

The conformance clause just says you can use a different document
character set, as long as it agrees with ISO10646. How you represent
such a document as a MIME/HTTP message entity isn't (yet)
specified. The restriction about iso10646 is just so that folks don't
go and use other document character sets, leading to confusion
regarding numeric character references.

An HTML 2.0 user agent is under no obligation to deal with anything
beyond ÿ. However, if it does deal with ਗ਼, it must do so in
a way that's consistent with ISO10646.

The whole picture of how to use ISO10646 effectively with HTML is not
specified in the 2.0 document, and it won't be unless we agree that we
need to tackle the whole internationalization issue in this document.