W3C home > Mailing lists > Public > www-html@w3.org > June 2003

The "ContentType" data type (was: Shorten <object> in XHTML 2.0?)

From: Christoph Schneegans <Christoph@Schneegans.de>
Date: Mon, 30 Jun 2003 13:27:43 GMT
To: <www-html@w3.org>
Message-ID: <bdpkte.3qk.1@mail.christoph.schneegans.de>

Christoph Päper wrote:

> In XHTML2 that should IMO be something like
>   <object data="test"
>    type="image/png, image/jpg;q=0.9, image/gif;q=0.8">
>    Test
>   </object>
> combined with server-side content negotiation.

I haven't seen any comments on
yet, although the topic seems to be quite interesting:

· What's the advantage over "traditional" server-driven content
negotiation? Nested objects like

  <object data="foo.png">
    <object data="foo.jpg">
      <object data="foo.gif">

are probably useful for servers that don't support content negotiation.
But if they do, why not write

  <object data="foo">

instead of

  <object data="foo" type="image/png, image/jpg;q=0.9, image/gif;q=0.8">

and let the server do all the work? Imagine you want to add another
representation, e.g. "foo.svg"! You would have to change every occurence
into something like

  <object data="foo" type="image/svg+xml, image/png;0.9, image/jpg;q=0.8, image/gif;q=0.7">

in all of your documents.

· Should "q" factors in a user agent's default "Accept" header taken
into consideration? For example, they could be multiplied with the "q"
factors in the "type" attribute. This is what Apache already does, cf.
section "Apache Negotiation Algorithm" in

· The draft suggests that user agents "first look for a file logo.png,
and then for logo.jpg ... when accessing files in a local filestore". Do
you want a user agent to look for "index.html" etc. when it encounters a
URI reference that ends with "/"?

Received on Monday, 30 June 2003 09:30:34 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:06:04 UTC