Re: XHTML2 MIME type

Henri Sivonen wrote:

> Non-validating parsers don't have to process external entities.

This is what <http://www.w3.org/TR/REC-xml#include-if-valid> says, but
does it apply to XHTML user agents? I expect a browser that accepts
"application/xhtml+xml" to support XHTML, not just generic XML. Of
course, a document type declaration is mandatory if entity references
are used, but there's one in <http://schneegans.de/temp/entities.xml>,
and Opera 6 and 7 fail nevertheless.

In <http://www.w3.org/TR/xhtml1/#uaconf>, the XHTML specification says
about conforming user agents:

  "If it encounters an entity reference (other than one of the entities
  defined in this recommendation or in the XML recommendation) for which
  the user agent has processed no declaration ..., the entity reference
  should be processed as the characters ... that make up the entity
  reference."

This implies IMO that it is generally assumed that conforming user
agents recognize the entities defined in the XHTML DTDs. Some
clarification would be nice, however.

And <http://www.w3.org/TR/xhtml1/#h-A2> states:

  "The XHTML entity sets are the same as for HTML 4 ..."

Opera 7.x _is_ broken anyway - it discards "&auml;", but recognizes
"&amp;auml;" as the "ä" character.

> Character entities are about solving an input problem and an encoding
> problem. XML solves the encoding problem by requiring all conformant
> XML processors to grok UTF-8 and UTF-16.

According to
<http://lists.w3.org/Archives/Public/www-html/2002Aug/0136.html>, it is
even considered to remove character entity references from XHTML 2.0. I
like that idea.

-- 
<http://schneegans.de/>

Received on Sunday, 11 May 2003 05:21:48 UTC