Re: Local entities in XHTML

On 11/12/2007, Yves MARCOUX <Yves.MARCOUX@umontreal.ca> wrote:
> 1- Don't you think that, if a file is served with a "close enough"
> type (or extension, for a local file), the browsers should infer from
> the namespace & DOCTYPE that the file is actually XHTML? My two cents
> is that it would be a safe inference to make, but maybe I am
> overseeing something.

Yes, you're overseeing something: There's a vast number of documents
containing an XHTML doctype and XHTML namespace that are not well
formed XML. A significant portion are also invalid. These documents
are in general built for HTML tag soup parsers, not SGML parsers.

If browsers made the inference and changed over to an XML parser also
for non-XML Content-Type headers (or equivalents such as type metadata
in the file system or type associated with the file name extension if
the document is not served over HTTP) then the majority of those
documents would yellow-screen.

Since browsers usually don't contain SGML parsers, internal subsets
would not work in HTML mode. So, browser makers would have to add
internal subset support for their HTML tag soup parsers, which is
nontrivial and opens an increased possibility for bugs and
interoperability problems.

> 2- What do you think of ISO-HTML, which explicitly forbids a local
> declaration subset?
>
> "The document type declaration shall not include a document type
> declaration subset" (from ISO/IEC 15445:2000(E) ISO-HTML, at least as
> it can be read at <http://www.cs.tcd.ie/15445/15445.html>).

Hardly matters since browsers don't use SGML parsers and don't support
internal subsets in their HTML tag soup parsers.
-- 
David "liorean" Andersson

Received on Tuesday, 11 December 2007 16:06:35 UTC