Re: Yet another error in float

On Fri, 26 Nov 1999 06:09:07 -0800 (PST),
=?iso-8859-1?q?Matthew=20Brealey?= ( wrote:
> Resent-From:
> @ section 
> 9.7:
>  Relationships between 'display', 'position', and
> 'float'

> properties and the box's containing block.
> Otherwise, if 'float' has a value other than 'none',
> 'display' is set to 'block' and the box is floated.
> -----
> Surely not.
> Should be:
> Otherwise, if 'float' has a value other than 'none'
> and display is 'inline', 'display' is set to 'block'
> and the box is floated.

No, it should be as it is now.  It also applies when display is
list-item, compact, table, etc.  Otherwise separate rules would
be needed for those cases.

> Also should say:
> Table elements, except for elements with display:
> table or display: inline-table should not be floated.

Should it?  The rule as originally written, when combined with the
rules in 17.2.1, handes this case already.  Some have argued that
that handling is not ideal, but that's a serious discussion.

> Also (in calculation of widths of floating elements)
> it should say that if width is omitted from floats,
> should state that refuse to float.

That wouldn't be backwards compatible.  It might not have been a bad
idea originally (for non-replaced elements only), but now that existing
specs say such floats should be made as narrow as possible, it should
probably stay that way.

> I say this because my experience is that far too many
> people omit width from float, and as a result it would
> be better for the concrete behaviour of float not
> working to occur than for the vague statement in the
> spec that width: 0 should result, and which UAs, due
> to its extreme vagueness, and generally unsatisfactory
> nature, do not follow.

The width 0 can be turned into the browser-specific min-width.  (Not
that that's always what's done.)


L. David Baron    Sophomore, Harvard (Physics)
Links, SatPix, CSS, etc.     <URL: >
WSP CSS AC                      <URL: >

Received on Friday, 26 November 1999 13:27:00 UTC