STR: 1. Load attached testcase, with grid enabled (layout.css.grid.enabled) EXPECTED RESULTS: In the testcase's lower section, the "b" cell should be slightly skinnier than in the upper section. (Its 75% width should make it occupy only 75% of its 50px-wide grid area.) ACTUAL RESULTS: In the testcase's second line, the "b" cell is *wider* than in the upper section. (It looks like it's taking 75% of the grid's full width) Chrome 48 (w/ Experimental Web Platform Features enabled) gives EXPECTED RESULTS. Nightly 44.0a1 (2015-10-21) gives ACTUAL RESULTS. (Mats, feel free to dupe as-needed if this is already tracked somewhere.)

I think this is just because we haven't landed that code yet? (bug 1163435) I was waiting for a CSSWG decision on the matter before landing it but that never materialized. I think the consensus now (at Mozilla) is that we should ship what we believe is the right thing to do, which is to resolve it against the size in the same axis, so I think we can land that patch now because that what it does IIRC. (the patch is probably bit-rotten by now though) (There's also a separate bug for abs.pos., bug 1163565.)

Ah, thanks.