In the test masonry-item-placement-001.html:
<grid>
<item style="grid-column:1">6</item>
<item>5</item>
<item style="margin-top:10px">3</item>
<item style="grid-column:span 2">1</item>
<item>2</item>
<item style="grid-column:3/span 2">4</item>
</grid>
Blink, Webkit, and Gecko all perform placement differently, and we'd like to confirm what the correct behavior is.
Blink's placement looks like this:

Webkit's placement result:

Gecko's placement result (which matches the current associated wpt reference file):

It looks like the main difference is that Webkit places items with explicit rows/columns first, and Gecko doesn't prioritize the minimum running position and places the item displaying 2 at the first position after the auto placement cursor.
It's not specified in the "Layout Algorithm" section of the masonry spec if items with a definite row/column should be placed first: https://drafts.csswg.org/css-grid-3/#masonry-layout-algorithm. In grid, explicitly-placed items are placed first: https://drafts.csswg.org/css-grid-2/#auto-placement-algo, but we wouldn't need that for masonry since we don't have the concern of auto-placed items occupying the exact cell that explicitly-placed items should take up.
@tabatkins @fantasai
In the test masonry-item-placement-001.html:
Blink, Webkit, and Gecko all perform placement differently, and we'd like to confirm what the correct behavior is.
Blink's placement looks like this:

Webkit's placement result:

Gecko's placement result (which matches the current associated wpt reference file):

It looks like the main difference is that Webkit places items with explicit rows/columns first, and Gecko doesn't prioritize the minimum running position and places the item displaying
2at the first position after the auto placement cursor.It's not specified in the "Layout Algorithm" section of the masonry spec if items with a definite row/column should be placed first: https://drafts.csswg.org/css-grid-3/#masonry-layout-algorithm. In grid, explicitly-placed items are placed first: https://drafts.csswg.org/css-grid-2/#auto-placement-algo, but we wouldn't need that for masonry since we don't have the concern of auto-placed items occupying the exact cell that explicitly-placed items should take up.
@tabatkins @fantasai