The MotionMark tests captured an edge case that causes expensive contentfulness computations.
See https://browserbench.org/MotionMark1.1/developer.html (Focus/Multiply tests)
This edge case is not plausible in real websites, but happens in this benchmark:
- There are lots of elements, performing rendering-intensive rendering
- There is nothing "contentful" in the page, so "first contenful paint" has to be computed multiple times.
Created attachment 421136 [details]
Comment on attachment 421136 [details]
If the test elements contained one glyph each, what would happen?
(In reply to Simon Fraser (smfr) from comment #2)
> Comment on attachment 421136 [details]
> If the test elements contained one glyph each, what would happen?
FCP will be marked the moment one of them become visible, and from that point on the check wouldn't run (as it only measures the *first* contentful paint).
Committed r273220: <https://commits.webkit.org/r273220>
All reviewed patches have been landed. Closing bug and clearing flags on attachment 421136 [details].