If you set the viewport size to 680 x 500 you can see the following:
For the first one the LCP size is based on the whole scroll height of the text element rather than just the visible amount.
For the second it's only the visible amount.
For the third it's the visible amount that would be visible on-screen (i.e., it's cut off by being out of the viewport and somewhere in between the first and second size).
I have trouble understanding the algorithm in these cases but wanted to check if the first and third scenarios are correct? i.e. should we consider the whole element size in this (including the non-visible scrollable area) or it should it only be the visible size?
If the former will try to document this better at https://web.dev/articles/lcp, if the latter this looks like a Chrome bug.
Firefox incidentally has the same size for all three tests so either way it seems we have an interop bug that could be better specified as to what is correct.
Look at these tests:
overflow: autooveflow: hiddenoverflow: autoand overflowing offscreenIf you set the viewport size to
680x500you can see the following:I have trouble understanding the algorithm in these cases but wanted to check if the first and third scenarios are correct? i.e. should we consider the whole element size in this (including the non-visible scrollable area) or it should it only be the visible size?
If the former will try to document this better at https://web.dev/articles/lcp, if the latter this looks like a Chrome bug.
Firefox incidentally has the same size for all three tests so either way it seems we have an interop bug that could be better specified as to what is correct.