Created attachment 191893 [details] Proposed change 1: Added a new function to resolve both opposite positions. Naming could probably be improved (welcome any comment).

Comment on attachment 191893 [details] Proposed change 1: Added a new function to resolve both opposite positions. Naming could probably be improved (welcome any comment). Passing the tests => good to go!

Comment on attachment 191893 [details] Proposed change 1: Added a new function to resolve both opposite positions. Naming could probably be improved (welcome any comment). View in context: https://bugs.webkit.org/attachment.cgi?id=191893&action=review > LayoutTests/fast/css-grid-layout/grid-item-spanning-resolution.html:106 > + <div class="sizedToGridArea autoSecondRowAutoFirstColumn" data-expected-x="0" data-expected-y="90" data-expected-width="160" data-expected-height="210"></div> The data-expected-{x|y} are wrong, they should be data-offset-{x|y}. Also the resolution assumes that we resolve grid-{end|after} like grid-{start|before}, which is wrong as we need to take into account the direction from which it was made. I will see if keeping this test as part of this change still makes sense without further refactorings.

Comment on attachment 191893 [details] Proposed change 1: Added a new function to resolve both opposite positions. Naming could probably be improved (welcome any comment). View in context: https://bugs.webkit.org/attachment.cgi?id=191893&action=review > Source/WebCore/rendering/RenderGrid.cpp:340 > + // No |positions| means that the item's position will need to be resolved by the auto placement algoritm Nit: algoritm -> algorithm > Source/WebCore/rendering/RenderGrid.h:121 > + // This is the function that should be used for any position resolution as we *have* to > + // resolve both directions at the same time (e.g. auto / 1, span 2 / 4, ...). > + PassOwnPtr<GridSpan> resolveGridPositionsFromStyle(const RenderBox*, TrackSizingDirection) const; Nit: The wording of this comment is awkward. Maybe the function name should somehow suggest that this is for resolving both directions at the same time.

Comment on attachment 191893 [details] Proposed change 1: Added a new function to resolve both opposite positions. Naming could probably be improved (welcome any comment). View in context: https://bugs.webkit.org/attachment.cgi?id=191893&action=review >> Source/WebCore/rendering/RenderGrid.h:121 >> + PassOwnPtr<GridSpan> resolveGridPositionsFromStyle(const RenderBox*, TrackSizingDirection) const; > > Nit: The wording of this comment is awkward. Maybe the function name should somehow suggest that this is for resolving both directions at the same time. Good point, I also discovered that some places (grid size estimations e.g.) won't be able to reuse the function. I will just remove the comment for now.

Created attachment 192245 [details] Patch for landing

Comment on attachment 192245 [details] Patch for landing Clearing flags on attachment: 192245 Committed r145240: <http://trac.webkit.org/changeset/145240>

All reviewed patches have been landed. Closing bug.