[css-anchor-position] "determine position fallback styles" should be guarded by "If the current style overflows"

[dshin-moz]

(Editorial, probably)

Strictly reading spec:

Fallbacks are evaluated via interleaving...

These modified styles are applied to the element via interleaving

... Through a full layout pass...

During a full layout pass, once a box has determined its fallback styles [...]

... Then at RO time, we determine position fallback styles (After potentially setting/removing last successful position option)

Then, determine position fallback styles for el [...]

... But then, there's no wording about checking if the current style overflows the cb. Such check is made in the loop...

5. If el rect is not fully contained within cb rect, continue.

... But that's skipped for the last successful position option.

If option is currently abspos’s last successful position option, continue.

So the spec makes it sound like if we have a positioned element that does not overflow without fallback, and has an option that also would not overflow, it would take the fallback option after RO events fire. Similarly, if we the non-fallback option overflows, but there are two fallback options that would not overflow, it sounds like the second option would be taken after RO events fire.

[fantasai]

Yeah, this whole section is confusingly written.

[tabatkins]

I'm a little confused by this - @dshin-moz starts quoting from the second paragraph of the section they link to, but the first paragraph already states exactly the condition they're asking to be added:

When a positioned box (shifted by its default scroll shift) overflows its inset-modified containing block, and has a non-empty position options list, it determines position fallback styles to attempt to find an option that avoids overflow.

Are you perhaps reading that first paragraph as non-normative, @dshin-moz ?

[dshin-moz]

Are you perhaps reading that first paragraph as non-normative, @dshin-moz ?

I read that as the intention/overview of the following definitions, but the following sections (Which provide precise terminologies and definitions) do not seem to include/elaborate on the condition, so... Yes, that first paragraph read to me as non-normative .

... Ultimately, yes, agreed that the pieces are there, and this is a minor point of potential confusion.