[css-values-4] Consider changing the specified behavior of <ratio>: 0 / 0

[emilio]

https://drafts.csswg.org/css-values-4/#ratios says:

The computed value of a is the pair of numbers provided, unless both numbers are zero, in which case the computed value is the pair (1, 0) (same as 1 / 0).

Is there a strong reason this is a computed value change rather than just a "behaves as"?

There isn't really a problem with this, but it feels odd to make it serialize to something else for no good reason, and it seems weird to serialize to 0 / 0 in specified values and 1 / 0 in computed values.

[Loirooriol]

I guess this can affect interpolations from/to 0 / 0depending on #4953

[tabatkins]

I don't think that's important, since 0 / 0 is an error case anyway. ^_^

But yeah, I think we were just imprecise in wording there. Should be fine to just make it behave as 1/0, but continue to compute to 0/0, if that's easier/better in the implementation.

[cbiesinger]

So, in the wpt/css/css-sizing/aspect-ratio/parsing/contain-intrinsic-size-valid.html test, it expects that the serialized value of 0/0 is 1/0. The spec only says the computed value should be 1/0. Is the test correct?

[tabatkins]

It depends on which value is being serialized. The test is incorrect, since it's serializing the specified value; that should still be 0 / 0.

[tabatkins]

I've fixed the test, btw.

[cbiesinger]

I thought I fixed the test :p

[tabatkins]

lol, i think i didn't merge the PR for mine