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

Re: Updated XHTML 2.0 draft published

From: Toby A Inkster <tobyink@goddamn.co.uk>
Date: Wed, 7 May 2003 21:56:36 +0100
To: Masayasu Ishikawa <mimasa@w3.org>
Cc: www-html@w3.org, www-html-editor@w3.org
Message-ID: <20030507205636.GB25340@ophelia.goddamn.co.uk>
My comments:


Someone (sorry, can't remember who) recently suggested that obeying semantics
should be made a part of conformance for documents. This of course would 
require manual checking. So instead, I recommend a two-tiered 
conformance test. A document should be said to be "level 1 valid XHTML2" 
if it validates against any of the DTD, Scheme or Relax NG definitions 
and "level 2 valid XHTML2" if it also meets the semantic requirements.

Level 1 validation could be machine checked, whereas level 2 validation 
would have to be manually checked.

As well as ruling out, say, using <h> elements for making text bold, 
level 2 conformance should also rule out using <div class="heading"> for 

5.5 - LinkTypes

Yes to redirect. It's a hack of sorts, but better than what we have now.
Also add "disclaimer" (we have "copyright", so why not?). "P3Pv1" should 
be deprecated in favour of the more general "privacy" which may point to 
a P3P v1 file, or alternatively some other machine- or human-readable 
privacy policy.

There should be an "original" link type to link to the original version 
of a file, from a mirror.

There should be "earliest", "earlier", "later" and "latest" link types. 
For example, the XHTML 2 specification could link (rel earlier) to XHTML 
1.1 and (rel earliest) to XHTML 1.0 (or HTML 1.0?).

There is a "start" link type, but no "end".

Add an "accessibility" link to let people know about a page's 
accessibility features (such as a list of accesskeys).

Some links have written by them "Some user agents also support the 
synonym foobar". Who cares? We're defining a new standard here.

5.5 - MediaDesc

Currently defined as "A comma-separated list of media descriptors as 
described by [CSS2]". Why limit ourselves to CSS2? CSS3 is just around 
the corner. But why limit ourselves to CSS at all?

Better definition would be "A comma-separated list of media descriptors 
as described by the standard for the preferred styling language of the 

6.1 - Title

Example has a wierd escaping problem.


What do we do if more than one edit is made to and element?

<p edit="changed" datetime="2003-01-13T13:15:30Z"
   edit="deleted" datetime="2003-01-17T17:15:30Z"> .. </p>

6.5, 6.6

<p src="foo" cite="bar" href="baz" type="image/png"> .. </p>

Which resource should be of type image/png. Foo? Bar? Baz? All of them?


The troubled style attribute. If we're going to keep it, can't we at 
least make it a little more useful. For example, say I want to use an 
inline style to set the background color of each paragraph to yellow:

	<p> .. </p>
	<p> .. </p>
	<p> .. </p>

It would be nice to do it just once using:

<section style="
p {
	background-color: #FFC;
        <p> .. </p>
        <p> .. </p>
        <p> .. </p>

rather than:

        <p style="background-color: #FFC;"> .. </p>
	<p style="background-color: #FFC;"> .. </p>
	<p style="background-color: #FFC;"> .. </p>

Also, must this data be of type "text/css"? Or are we going to give the 
type attribute yet another usage?!

7.0 - security tag



"Every XHTML document must have a title element in the head section." 
Are any of the following to be considered valid?

<title />
<title> </title>


Deprecate h1-h6? No. Discard them? Yes. If it is really neccessary to 
add some kind of backwards compatibility measure, include a "level" 


Paragraphs can now contain lists. Yay! It always irked me having to end 
paragraphs with things like "such as:"


Add a "sectiontype" attribute to describe the purpose of the section. It 
should have a fixed set of possible values (like rel). I'd suggest at 

	* Glossary
	* Introduction
	* Conclusion
	* Appendix
	* Contents
	* Index
	* Note
	* Bibliography
	* Example
	* Chapter
	* Subchapter

9.3, 9.6, 9.9, 9.14

Too much emphasis on a narrow field of endevour. Either remove all 
(except 9.3) of these elements, or add other elements corresponding to 
parts of text commonly use in the arts, humanities and other areas of 
science. If the latter, I'd suggest at least:

	<person /> 

	<publication /> for books, journals and even musical or 
	cinematic works.

For example:

<p><publication xml:lang="de" cite="#marx">Das Kapital</publication> was
written by <person id="marx">Karl Marx</person></p>


The <dfn/> element is awkward. Get rid of it.

8, 9

We are really crying out for a better system of quoting and referencing.


Attack and destroy!


I'm not entirely sure why this exists. It's more or less just a 
two-column table semantically.


What is the semantic difference between these:

<link rel="next" href="foobar"/>
<link rev="prev" href="foobar"/>

I propose that rev be removed.


For all the examples, why do you propose that only search engines would 
be interested in them?


The profile attribute is stupid. Use rel="scheme.X", as per the 
RFC defining embedding Dublin Core into HTML.


Standby element. Good idea. Should possibly specify (or at least 
recommend) that no external objects should be embedded in the standby
element. For example:

<object data="mymovie.gif">
		<object data="mymovie.avi" />

Display an AVI while we're waiting for the GIF to load?


Do we even need an <object/> anymore? We can embed using the src 
attribute on any element.

Toby A Inkster BSc (Hons) ARCS | mailto:tobyink@goddamn.co.uk | pgp:0x6A2A7D39
    aim:inka80 | icq:6622880 | yahoo:tobyink | jabber:tobyink@a-message.de
            http://www.goddamn.co.uk/tobyink/ | "You've got spam!"

Received on Wednesday, 7 May 2003 16:56:56 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:06:03 UTC