Posted by Jukka K. Korpela on 06/19/05 15:35

Spartanicus <invalid@invalid.invalid> wrote:

> UA preference part of the accept header should be respected, it's
> there for a reason.

Even if experience shows that's incorrect more often than not?

> For example Opera =<7.2x
> doesn't recognize character references in X(HT)ML mode.

Pardon? I've never heard of such a problem. But I know that Opera has
had problems with _entity_ references in XHTML. Actually even Opera
7.54 seems to fail to get entity references right if I use a DOCTYPE
that refers to my own copy of the XHTML DTD. Sounds like Opera still
handles the entity references in some kludge way.

> Opera's
> accept header says that it *prefers* HTML, so it should be served
> HTML.

At least Opera 7.54 seems to say:

Accept: text/html, application/xml;q=0.9, application/xhtml+xml,
image/png, image/jpeg, image/gif, image/x-xbitmap, */*;q=0.1

It has the _same_ (implied) q value (of 1.0) for text/html and
application/xhtml+xml. The HTTP specification does not define the order
of values in an Accept header as significant, so I can't see what might
constitute an expressed preference here.

> More importantly the renderer most people want to use
> application/xhtml+xml for (Gecko) currently refuses to render
> anything until the document has completely downloaded.

That's certainly an efficiency problem. But Mozilla expresses a
preference of application/xhtml+xml over text/html (using q parameter).
If you intend to serve XHTML as XHTML _at all_, then this would surely
be the place to do so.

--
Yucca, http://www.cs.tut.fi/~jkorpela/
Pages about Web authoring: http://www.cs.tut.fi/~jkorpela/www.html

[Back to original message]