RE: [CSS2.1] Flow control interop

On Saturday, November 15, 2014 12:42 PM, Greg Whitworth wrote:
> > IE is actually correct since the margin of the cleared element, which
> > is the issue, should be maintained. There is nothing in the spec that
> > says the margin of a box should be thrown out in this case. What
> > FF/Blink/Webkit are doing is incorrectly throwing out the margin-top
> > and this prevents the entire properties use. This is a really big
> > issue. As an example if there is text, as is the common case of floats
> > and clear, we see that all browsers are consistent
> > http://jsfiddle.net/d83gt4d8/7/. This to me seems like FF/Blink/Webkit
are
> buggy when there is no text or previous content before the float.
> 
> Agreed, they are inconsistent and that is why I am asking them to align on
> something, preferably us since we do have the consistency and spec
> compliance (thanks for pointing out the errata). But the important
question
> is, is Gecko/Blink/Webkit willing to align with IE on this?
> 

I have continued to think on this and there are additional issues as well.
The FF/Blink/Webkit implementations actually will prevent floats and margins
from working correctly in resize situations where content reflows and the
margins are disappearing. This can cause spacing issues inconsistent with
what the developer would expect. Additionally there are issues with negative
margins that are not expected as well and need to be taken into account
here.

Examples of issues on top of the initial issue and why we should use the
spec definition: 
http://jsfiddle.net/d83gt4d8/9/
http://jsfiddle.net/d83gt4d8/10/
http://jsfiddle.net/d83gt4d8/11/
http://jsfiddle.net/d83gt4d8/14/
http://jsfiddle.net/d83gt4d8/15/

As you can see there are many inconsistencies with negative margins. We need
to get the entire picture together before we make a decision on which
solution is best. 

FF/Blink/Webkit all seem to have an issue with negative margin on a float
that incorrectly places the float in the wrong place. 

FF/Blink/Webkit all seem to have inconsistencies on resize as the text flows
too where elements jump position and margins collapse and not collapse, or
clearance is applied incorrectly.

--
Thanks,
Arron Eicholz

Received on Monday, 17 November 2014 14:22:51 UTC