Re: Horizontal rules

--- Tantek Çelik <tantek@cs.stanford.edu> wrote:
> > completely described in CSS3:
> 
> Untrue.  There are aspects in this style sheet which are not valid
> features in
> any CSS Rec or public WD.

This is true. A better phrase would be 'completely described in potentail
CSS3'.
 
> > HR {border: 1px solid;
> > border-color: windowborder /* ? */;
> > display: block;
> > margin: 6px auto}
> >
> > HR[color] {border-color: attr(color)}
> 
> Strictly speaking, the COLOR attribute of HR and the CSS color property
> do not
> accept the same set of values, and the border-color property in
> particular
> does not accept 'attr()' as a value.

See above. However, there is no public statement of validity of values for
border-color for CSS3. Any differences could be resolved (what, harmony
between CSS and HTML? Surely not!) by minor changes to CSS.

> 
> > HR[align=left] {margin-left: 0 !important /* In a UA style sheet?
> (used
> > because the author might set margins, which would override this
> > declaration. I believe a UA !important is essential if there is a real
> > desire to map HTML attributes to CSS, and I suggest that it raise the
> UA
> > declaration to an author one; i.e., this declaration is equivalent to
> > setting margin-left: 0 in an author style sheet, so specificity works
> > properly */}
> >
> > HR[align=right] {margin-right: 0 !important}
> >
> > HR[size] {height: attr(size)-1} /* Sad that even box-sizing doesn't
> work
> > for this (since the default, size=1 has a box height of 2 pixels (top
> and
> > bottom borders of 1pixel) */
> 
> height: attr(size) is incorrect because unitless length values for
> height are
> illegal (typical use of the SIZE attribute is unitless).  

Yes. I realised this almost immediately, and was about the correct myself,
but you beat me to it - add px to the end of that.

> in addition,
> the
> height property in particular does not accept 'attr()' as a value.

Rather strange in view of this in a WD you edited, don't you think?:

 height: attr(size)em;

(in respect of SELECT)
> 
> attr(size)-1 is undefined because there are no binary arithmetic
> operators
> allowed in CSS length values.

Again this is a potential extension.
> 
> > HR[noshade] {background: windowborder}
> >
> > HR[noshade][color] {background: attr(color)}
> 
> Similarly to the HR[color] rule above, the COLOR attribute of HR and the
> CSS
> background-color property do not accept the same set of values, and the
> background-color property in particular does not accept 'attr()' as a
> value.

As with all of these cases there is no public CSS3 definition so you can't
say that. In any case it seems that it is ok when you write it in a formal
draft, but not ok when I make it in a informal suggestion to this list
suggesting how HTML's HR could be completely described (although not you
say satisfactorily, on the basis of obscure float examples, saying that it
is not possible to describe the behaviour using CSS (your stance on this
does not seem wholly in tune with that of fellow CSS WG members (witness
the 'BR is an inline element' shennanigans) on the issue of CSS-HTML
mapping), and implicitly favouring possibily dozens of properties in order
to describe the behaviour of HR and its interaction with align (and this
itself implies that the interaction of HR and IMG is in some way codified,
which isn't the case - HTML is as vague as can be; I have tested
IMG[align]/HR interaction in three different browsers, and I got three
different results; to say that the result for IMG[align] and HR is so set
in stone and so overwhemingly wonderful that it is necessary to create a
whole raft of exceptions to CSS is, in my view, mistaken) using CSS. You
really should make up your mind.


=====
----------------------------------------------------------
From Matthew Brealey (http://members.tripod.co.uk/lawnet (for law)or http://members.tripod.co.uk/lawnet/WEBFRAME.HTM (for CSS))
__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com

Received on Tuesday, 29 February 2000 16:31:26 UTC