WebKit hacking, pt. II

I had a bit of a break today and happily got working the patch that was frustrating me yesterday. In the end, after ruminating over many test cases and CSS3 drafts, I concluded that content: inherit, none, and normal should all just get rid of the generated element. This is clearly the behaviour in CSS2.1 (whatever Hixie thought in 2001 when he muddied the waters with a false test case), but the question was whether that would be a forward compatible approach for the monstrously complicated things in the CSS3 module. Follow the details on the bug tracker.