Re: Maintained matrix of CSS properties v. browsers/rendering engines ?

On Tuesday 2007-01-30 18:46 +0100, Allan Sandfeld Jensen wrote:
> It has lots of information, but is not entirely neutral: The author has a 
> tendency to simply declare any differences to Firefox a flaw in other engines 
> regardsless of what the standard says, but since CSS 2.1 is clarified pretty 
> much the same way by the working group, he is not entirely wrong ;)

As somebody involved in Gecko/Firefox development and on the CSS
working group, I think you're exaggerating a bit here.  We've had to
make a significant number of changes to match clarifications and
changes in CSS2.1, such as the ones listed below.  I'd note that I
supported many of these changes in the working group because of the
improvements in interoperability.

Some examples:

Rewriting the rules used to position absolutely positioned elements
and the meaning of 'overflow' and 'clip' (part of the CSS2 errata,
changed in Gecko way back around 2000/2001, although I don't
remember if we ever really implemented clip correctly per CSS2).

Rewriting the z-index rules, which required us to rewrite most of
our painting code:

Clarifying that overflow:hidden behaves like 'scroll' and 'auto'
rather than like 'visible' in terms of layout:

Changing the definition of computed value so it no longer includes
anything that requires layout:

Clarifying the way baseline alignment is defined:

There was also counters, where I stopped work on implementation
because I knew the spec was so vague that necessary clarifications
would force significant changes to the implementation.  I finished
that after the group agreed on the clarifications, although I think
before the draft containing them was published:


L. David Baron                                <URL: >
           Technical Lead, Layout & CSS, Mozilla Corporation

Received on Tuesday, 30 January 2007 20:20:42 UTC