2.44 KB, text/html
According to the CSS working group (per the August 2000 F2F ) nested floats do not affect the formatting context of the ancestors of the float, and do not get the outer float to shrink wrap about them. We do this correctly when the outer float also has text in it (), but when a float is directly contained within another float, we mess up and shrink wrap the inner float. See: http://www.hixie.ch/tests/adhoc/css/box/float/001.html -- References --  It's a little brief, but: http://lists.w3.org/Archives/Member/w3c-css-wg/2000JulSep/0119.html  http://www.hixie.ch/tests/adhoc/css/box/float/002.html  http://www.hixie.ch/tests/adhoc/css/box/float/003.html
Nesting floats is not very common. Suggesting delaying until post-mozilla0.9.
This is actually a decision I'd like to revisit in the working group - I'd like to see the outer floats shrink-wrap.
Accepting, and, due to low-impact, moving to future. Also, adding [FLOAT] to summary...
Reconfirmed using FizzillaCFM/2002070913. The testcase doesn't appear as described. Setting All/All.
CSS2.1 isn't really any more clear on this issue than previous CSS versions.... What's the current state of the spec on this?
I think the status is dbaron and I mildly disagree. I want to say that floats are not affected by floats within them (hence my filing this bug); dbaron wants to say that a root formatting context grows to fit all floats, including floats (IIRC). The WG should discuss this in the coming months.
OK. Over to layout:floats; hixie, feel free to take this pending WG discussion....
WG discussion has changed this bug slightly...
You're right about what we should do, but the spec doesn't require it. 10.3.5 is intentionally vague.
Yes, technically "shrink-to-fit" is not defined, but I'm _reasonably_ sure that there's no way what we do can count as "shrink-to-fit"...
Also visible when there's a right float in an abs-pos block.
Created attachment 189864 [details] floated and absolutely positiond divs containing floats My testcase shows that the bug is also present when a right-floated block contains another right-floated block. So any block that is floated or absolutey positioned and contains a right-floated block will be effected by this problem. BTW, since fixing this bug would fix the second line of the Acid2 test, shouldn't this be given more attention once Gecko 1.8 is branched? (This is my first contribution to Bugzilla, so let me know if I did something weid.)
According to dbaron this is fixed on his REFLOW branch.
Fixed on trunk by reflow branch landing.