Re: Scope of :hover

Michael wrote on Tuesday, June 3, 2003 at 8:14:24 PM:

> Please no, that will make parsing selectors very confusing, as
> :hover by itself has an implicit universal selector according to the
> grammar, making it identical to *:hover.

If it's a horrible burden to implement *:hover and :hover differently,
okay. It makes the behavior less useful, but that isn't required for
the core idea. A possible variant is letting :hover and *:hover apply
only to certain elements similar to Win IE (so they will be identical
selectors), but still letting h1:hover apply to h1 elements. If that's
still considered "very confusing" I don't know what to do. Is the rest
of CSS comparatively easy to implement?

As a CSS author, I think it's confusing that *:hover and a:link:hover
could theoretically be identical in a CSS2 UA. I wish they had
originally said :hover and *:hover apply to everything, but I realize
changing it to that today causes a hell of a problem for documents
authored specifically to Win IE (so about 90% of the Web). It's
probably more workable to only have :hover to apply to certain
elements, but that doesn't preclude allowing constructs like h1:hover
for those that want them. Those that don't simply won't write h1:hover
since it would be useless to them.

Does that sound workable?

-- 
John Lewis

Received on Tuesday, 3 June 2003 21:39:37 UTC