Re: text/html for xml extensions of XHTML

On Sat, 16 Jun 2001, William F. Hammond wrote:
>>>
>>> The specific issue in this sub-thread: What is the reason for a
>>> user agent's policy-level refusal to parse as xml, rather than as
>>> tag soup, an http object served as text/html upon finding an xml
>>> declaration at the body origin.
>>
>> http://www.damowmow.com/mozilla/html-not-xml.html
>
> I did mean *XML-aware* user agent. Yes, an SGML parser that does not
> look inside PI's will buy it. But for an XML-aware user agent a PI
> named "xml" has non-optional special meaning.

That file has ABSOLUTELY NOTHING WHATSOEVER to do with XML (from a
technical perspective, I mean).

It's a completely valid HTML instance document sent as text/html and
rendered completely correctly by conforming implementations such as
Netscape 6.1 beta 1 or Mozilla 0.9.1.


> Here's what Amaya thinks about this example:
>
> *** Errors/warnings in http://www.damowmow.com/mozilla/html-not-xml.html
>   No encoding specified, assuming UTF-8
>   line 1, char 11: syntax error
>
> Is Amaya wrong?

If Amaya claims to support HTML, then yes, since that document is a
valid HTML document sent as text/html, and thus should be rendered as
such. If it does not claim to be an HTML UA, then I have no idea.

In my opinion there is no valid reason for any HTML user agent to
decide to treat that document as XML.


>> . . .
>>>> "text/xml" is simply too general to be sensible for internal
>>>> handling by unified http/html user agents.
>>> I don't understand what that sentence means.
> Under RFC 3023 "text/xml" coverage includes any UTF-8 encoded
> instance that conforms to the W3C XML Specification
>    http://www.w3.org/TR/REC-xml
> and degrades gracefully to "text/plain". Most of that territory is
> well beyond CSS-based rendering and Mosaic descendants as renderers
> except for (nearly useless) tree layouts. Some, but not all, of
> these will have more specialized MIME classification. As a user, I
> do not want those that are outside the XHTML family to be handled
> internally in Mosaic descendents.

(Does the mozilla.org family of browsers classify as a Mosaic
descendent? How about Opera? Konqueror?)

Mozilla has a native MathML implementation (in fact this thread
originated, and is still being cross-posted to, the mozilla-mathml
mailing list). If you send a text/xml instance to a suitably compiled
version of Mozilla 0.9.1, for example, it will render it using the
MathML rendering rules. There need not be any XHTML involved.

Ditto SVG, although Mozilla's SVG implementation is significantly less
mature than the MathML version.

Opera also natively supports WML, another XML language.

In addition, I have many documents in the "http://www.example.com/"
namespace that are rendered (almost) correctly by Mozilla and Opera.

For example:
   http://damowmow.com/mozilla/xml-not-xhtml.xml

How would you suggest a UA should tell the difference between these
perfectly valid XML documents that are renderable by UAs and the
documents you refer to that are beyond stylesheets and native support?

-- 
Ian Hickson                                            )\     _. - ._.)   fL
Invited Expert, CSS Working Group                     /. `- '  (  `--'
The views expressed in this message are strictly      `- , ) - > ) \
personal and not those of Netscape or Mozilla. ________ (.' \) (.' -' ______

Received on Saturday, 16 June 2001 20:55:54 UTC