Re: "non-zero top border" in 8.3.1

On Sun, 16 Sep 2007, L. David Baron wrote:
> > 
> > I want to hear from the browsers what they want to do. We'll talk 
> > about it in F2F, if you are not coming (or are you?) I hope you can 
> > call in for this one.
> So I think there are two pretty good arguments in favor of making the 
> change Alex proposes:
>  * it seems to match implementations better (although I think this could
>    use a little more verification)
>  * it gets along better with incremental rendering -- a float at the
>    beginning of an empty block is treated the same as the same float
>    would be if the block had content after the float.  This seems pretty
>    important to me.  And, in general, the position of an empty block is
>    the same (top) position it would have if it had content, which seems
>    to be pretty useful in implementation.
> So, after the discussion we had at the meeting, I'm leaning in favor of 
> the proposed change.

My reasons for being strongly against this change are not technical. I 
think it is a bad idea to be changing this section again. To summarise, my 
objections are:

 * That we don't know what this change will imply. I think it would be 
   better to wait for a wider and more detailed study of the side effects 
   of this change before we go in and start poking at what was previously 
   assumed to be very stable text.

 * That the previous text hasn't been given adequate time for implementors 
   to attempt to implement it.

 * That repeatedly changing such complicated parts of the spec will 
   continue to make the CSS working group look incompetent.

 * That this change would further lower the respect that implementors have 
   in the CSS working group, leading to the standards getting ignored.

 * That if we make this change, it could be interpreted by the community 
   as being an admission that fundamental changes are again fair game, 
   which could lead to rather unwelcome pressure being put on the group.

 * That it is FAR too late (by years) to be making changes to previously 
   agreed text.

BTW, if this section does change, someone should take the responsibility 
for updating all the margin collapsing tests:

(I certainly won't have time to do it; I'd estimate it would take a margin 
collapsing expert somewhere in the region of two weeks of solid work to 
actually go through each test and verify that it is still correct.)

Ian Hickson               U+1047E                )\._.,--....,'``.    fL       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Tuesday, 18 September 2007 18:57:24 UTC