[motion-1] Fix offset-path parsing test

[danielsakhapov]

Test List

https://wpt.fyi/results/css/motion/parsing/offset-path-parsing-valid.html

Rationale

Update tests to be up-to-date with the spec
web-platform-tests/wpt#40120

[danielsakhapov]

@BorisChiou @nmoucht

[BorisChiou]

Thanks. The parsing test and computed value test are out-of-date of course. I'm also trying to add more tests for shapes/coord-box in parsing-valid and computed tests.

[nt1m]

The margin-box one looks OK. I'm not sure about the omitting the position part. @emilio @Loirooriol, should we always serialize the position here?

[BorisChiou]

About the serialization, the current spec requires us to serialize <position> always, https://drafts.csswg.org/css-shapes-1/#basic-shape-serialization. However, there is a spec issue talking about this: w3c/csswg-drafts#8695.

[Loirooriol]

https://drafts.csswg.org/css-shapes-1/#funcdef-basic-shape-ellipse

The position argument defines the center of the ellipse. This defaults to center if omitted.

https://drafts.csswg.org/css-backgrounds-3/#valdef-background-position-center

center: Computes to 50%

https://drafts.csswg.org/cssom/#serializing-css-values

If component values can be omitted or replaced with a shorter representation without changing the meaning of the value, omit/replace them.

So it seems to me that ellipse(50% 60%) should NOT serialize as ellipse(50% 60% at 50% 50%).
In fact, ellipse(50% 60% at 50% 50%) should serialize as ellipse(50% 60%).

Same for the circle(100px) above.

[emilio]

Agreed with @Loirooriol, that test fix is wrong.

[emilio]

(in principle, at least).

The spec says:

omitting components when possible without changing the meaning

Which seems true in this case. But then in the non-normative example it says:

Omitting components means that some default values do not show up in the serialization. But since always uses the 2- or 4-value form, a default is not omitted.

I don't quite understand why we shouldn't omit it, but I'm really jet-lagged so maybe I'm not parsing that right.

[BorisChiou]

BTW, the spec of offset-path says:

If circle() or ellipse() is used, and an explicit center position is not given, they default to using the offset starting position, rather than their standard default.

So now I feel weird if we serialize it as 50% 50% if the user doesn't give the center position. We do not use 50% 50% as default value for offset-path. Serializing it as 50% 50% may mislead people.

[BorisChiou]

cc @tabatkins

[BorisChiou]

I filed a spec for this: w3c/fxtf-drafts#514.

[foolip]

Does the current spec text require some specific behavior, or is it just unclear about the behavior? I'm wondering if we should wait until w3c/fxtf-drafts#514 is revolved before making any test changes, or if it can be done in two stages.

[BorisChiou]

Maybe just accept both cases for now. e.g.

// https://github.com/w3c/csswg-drafts/issues/8695
test_valid_value("offset-path", "circle()", ["circle()", "circle(at 50% 50%)"]);

Or something like this.

[BorisChiou]

OK, the comment in w3c/csswg-drafts#8695 (comment) makes sense to me. We should omit at <position> if it was not given. And preserve the at <position> even if it specifies the default value, for the same reason.

[danielsakhapov]

@BorisChiou @nt1m updated the test, please, have a look

[BorisChiou]

Mind updating this as well?

replace
test_valid_value("offset-path", "circle(100px)", "circle(100px at 50% 50%)");
with
test_valid_value("offset-path", "circle(100px)");

For this proposal: "We definitely need to omit the at <position> when it was omitted by the author, for offset-path to work correctly", from w3c/csswg-drafts#8695 (comment)

[danielsakhapov]

done

[nt1m]

The test looks good to me now.