User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:2.0b8) Gecko/20100101 Firefox/4.0b8 Build Identifier: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:2.0b8) Gecko/20100101 Firefox/4.0b8 Refer to the third example on http://oli.jp/bugs/mozilla/flexbox-overflow.html (“Outline on box element children is padded by the same width”) for more info. Reproducible: Always Actual Results: When adding a 1px outline to a box element’s child, the outline is separated from the child by the same amount, showing 1px of the background. The separation matches the width of the outline Expected Results: The outline should be drawn immediately outside the element’s area, as it is in Firefox 3.6.10
Oli, if this behavior changed since 3.6, would you be willing to find the regression range? roc, it looks like we're including the block's outline somewhere in the box code when doing sizing.... Are we maybe looking at overflow areas somewhere?
Boris, I’m not actually sure how to find a regression range (or ahem what it is), and after filing 6 flexbox bugs today I’d like to pass if possible ;) sorry for the non-minimal test case, I got lazy after the first couple and combined em.
Oli, a regression range is basically "when the problem appeared". Usually found by downloading nightly builds so as pin it down to a day. The regression range here is in December 2009: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=3ff17b03644e&tochange=44c392db6672 dbaron, you have some overflow changes in there.... This changeset looks suspicious: http://hg.mozilla.org/mozilla-central/rev/038564bc6eb7
David, see end of comment 4.
Just FYI this is regarding OLD Flexbox (display: box;), not NEW Flexbox (display: flex;), and the chances of it being fixed are probably not great. For more background: https://bugzilla.mozilla.org/show_bug.cgi?id=625694