W3C home > Mailing lists > Public > www-style@w3.org > February 2012

Re: Features and fixes incompatible with backward compatibility

From: L. David Baron <dbaron@dbaron.org>
Date: Sun, 19 Feb 2012 12:25:05 -0800
To: www-style@w3.org
Message-ID: <20120219202505.GA26075@pickering.dbaron.org>
On Sunday 2012-02-19 14:32 -0500, Boris Zbarsky wrote:
> On 2/19/12 2:21 PM, Matthew Wilcox wrote:
> >If anyone has examples of such issues, please could you document them
> >in this thread?
> The fact that "display" and "position" and "float" are separate
> properties (or at the very least that the "absolute" and "fixed"
> values of "position" are not just values of "display", and likewise
> for left and right floating).  Not sure that "everyone" agrees on
> this, though.

I certainly agree.

A few others off the top of my head (though I'll likely think of
more later):

  Form controls should have been display types, even if not defined
  in CSS levels 1 and 2.  Setting a form control to display:inline
  or display:block should have made it not look like a form control
  anymore.  (I think Mac IE 5 did that, but I didn't have a chance
  to change Gecko in time to keep that behavior Web-compatible.)

  CSS should have had 2 different types of blocks, one for a frame
  (a region of formatting), and one for something that just causes
  line breaks.  This would have enabled having an inline formatting
  model that more closely matches design tradition (though still
  avoiding accidental overlap) by having the line spacing between
  lines rather than around them.  It also would have let us solve
  bullet displacement around floats without an indent-edge-reset
  property (as proposed).

  Margin collapsing shouldn't have involved collapsing through

  Mutually exclusive media types (screen, projection, print,
  handheld) shouldn't have existed; media queries are the way to go.

  !important should have used a delimiter other than !.

  The 'clip' property should not have existed.

  'auto' margins should not have existed; we should have had
  block-align instead.

  Link pseudo-classes should have been :link, :unvisited-link, and
  :visited-link.  Or more likely (for privacy) just :link, plus
  properties to change the color and background for visited-ness.

  Perhaps the descendant combinator in selectors shouldn't have been
  easier to type than all the other combinators.


𝄞   L. David Baron                         http://dbaron.org/   𝄂
𝄢   Mozilla                           http://www.mozilla.org/   𝄂
Received on Sunday, 19 February 2012 20:25:28 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:08:12 UTC