W3C home > Mailing lists > Public > www-style@w3.org > February 2004

Re: [CSS21] response to issue 54

From: Jukka K. Korpela <jkorpela@cs.tut.fi>
Date: Mon, 9 Feb 2004 21:17:52 +0200 (EET)
To: www-style@w3.org
Message-ID: <Pine.GSO.4.58.0402092042070.3547@korppi.cs.tut.fi>

On Mon, 9 Feb 2004, Bert Bos wrote:

> CSS WG response:
>
>     We don't consider TYPE on lists to be presentational the same way
>     as COLOR or FONT are.

Sorry to interfere, but I really have great difficulties in seeing what
this tries to say. Clause 6.4.4 is very obscure. There was an informal
comment about one attribute, TYPE, being actually presentational in some
elements but not in others. This is hardly debatable under any reasonable
interpretation of the word "presentational". I'm not sure whether the
comment was meant to suggest a change or not.

But before the response, I thought I understood what 6.4.4 means. It gives
a long list of attributes and says that any attribute that is not in the
list should be considered presentational. So apparently this does not
exclude the possibility that some of the attributes in the list might be
interpreted as presentational, in some situations at least.

But after the response I don't know what to think about it. Does the
response mean that TYPE is presentational in the sense meant in 6.4.4
but the WG thinks something else about it in some other sense? And why
would such other thoughts affect the intepretation of 6.4.4, or the
specification as a whole?

There's probably some complicated reason why 6.4.4 does not simply list
the attributes that _are_ presentational, specifying them by element if
necessary (since HTML attributes may have quite different meanings
depending on element).

Maybe the reason is that the wording "The UA may choose to honor
presentational attributes in the source document. If so, these attributes
are translated to the corresponding CSS rules - -"
is intentionally obscure too. Does it really forbid UAs from honoring
attributes that are _not_ declared as presentational in 6.4.4?
It doesn't actually say so. And what about attributes that _cannot_ be
translated into corresponding CSS rules, such as, incidentally, the
TYPE attribute in OL? Should we draw the conclusion that such attributes
must not be honored?

Besides, what's the point of specifying something in terms of
"corresponding CSS rules", when no specified correspondence is given
anywhere? Of course we can draw _some_ conclusions, but what is really,
say, the CSS equivalent of <font size="1"> or <dl compact>?

I hadn't noticed that 6.4.4 also says: "For XHTML and other languages
written in XML, no attribute should be considered presentational."
So if I use HTML, the <font color="..."> is presentational, but if I use
XHTML, it is not? And if I use XML, then nothing I say in markup can ever
be presentational?

Please find a new word for what you mean by "presentational", since it
apparently deviates quite a lot of what "presentational" means in common
language.

I think 6.4.4 as a whole could be replaced by a few simple sentences, such
as:

The rendering of documents may be affected by markup, too, even by
elements and attributes that are only meant to suggest presentational
features. The processing of such markup by browsers supporting CSS
should be described in the documentation of browser style sheets,
to the extent that the presentational features are describable in CSS.

For HTML documents, including XHTML documents, browsers' default style
sheets should contain rules that correspond to specified or commonly
accepted meanings of elements and attributes that are essentially
presentational, such as
b { font-weight: bold; }
font[face] { font-family: attr(face); }

***

It would be useful to list down the exact rules, really. The so-called
default style sheet in appendix D is a hopeless mixture of attempt at
empirical survey, wishful thinking, and weird ideas like
abbr, acronym   { font-variant: small-caps; letter-spacing: 0.1em }
and it addresses issues far beyond the simple question "what do
presentational HTML features mean in CSS terms"?

-- 
Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/
Received on Monday, 9 February 2004 14:18:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:26 GMT