- From: Geoff Soutter <geoff@volantis.com>
- Date: Mon, 25 Jul 2005 09:48:19 +0100
- To: "L. David Baron" <dbaron@dbaron.org>
- Cc: www-style@w3.org
On Fri, 2005-07-22 at 11:45 -0700, L. David Baron wrote:
> On Friday 2005-07-22 20:36 +0200, Allan Sandfeld Jensen wrote:
> > Sorry to rant, but I have to ask:
> >
> > Why have rules that a redundant and pointless?
>
> Because behavior should be interoperable.
>
> Experience has shown that unspecified behavior leads to pages depending
> on the market leader's behavior. This requires that other
> implementations reverse engineer the market leader. It's much easier if
> they can just read the spec.
Agreed, although IMHO the existing rules are confusing to the point of
ensuring that implemented behaviour is unlikely to be interoperable.
> > It would make sense if counter-reset and counter-increment properties was
> > additive, but they cascade like all other properties, which means that the
> > only way to do two resets on the element is to declare them both at the same
> > time: {counter-reset: dummy 0 dummy 1}. Being able to do that is just
> > pointless. Having to waste time defining the behaviour of pointless
> > declarations is even more pointless.
>
> Sure, there are alternatives for what the behavior should be. It could
> be that the declaration should be considered an error (and thus be
> ignored).
>
> It really doesn't matter much; it just needs to be specified and
> interoperable.
Agreed.
IMHO, having both counter-reset and counter-increment ignore duplicates
(by whatever mechanism) would be the simplest, clearest and most
consistent solution.
--
Geoff Soutter <geoff@volantis.com>
Received on Monday, 25 July 2005 08:48:27 UTC