Csswg-drafts: [css-logical] Should the `inset` shorthand allow quirks in their lengths like the individual properties do?

[emilio]

I'd prefer if not, but the current spec doesn't say. In quirks mode, some (non-zero) lengths don't require units, so for example <div style="position: relative; top: 100"> works.

I'd prefer if we didn't have to do this for inset, given it's a new property, and that the inset- longhands don't allow this looks like.

[emilio]

/cc @Loirooriol

[SelenIT]

I believe the unitless length quirk is applied only to properties explicitly listed in the Quirks Mode Living Standard, isn't it?

[emilio]

Yes, though the definition of the inset property use <'top'>.

[emilio]

Also the Blink implementation (though not shipped) supports this: data:text/html,<div style="position: relative; inset: 100 0 0 0">Foo.

So not sure whether the spec needs fixing and/or whether Blink needs a bug on file.

[Loirooriol]

Ah, I overlooked this. Blink allows the quirk, because the easy way to parse the shorthand was using the longhands.

I don't have a strong opinion, if inset remains a shorthand of only the physical properties it may make some sense to allow the quirk in order to have a consistent syntax. But if it's going to become a shorthand of both logical or physical longhands like fantasai wants, then I wouldn't allow it.

[Loirooriol]

@tabatkins was against spreading the quirk any further (but I was asking about logical longhands).

[emilio]

I agree with tab on that. I'd prefer to keep the quirks where we know it's needed. In this case it's clearly not needed, since the property is new.

[SelenIT]

Isn't the <top> notation in the new properties definition just a shorthand for the auto | <length-percentage>, not the literal mapping to the top property? The notation <‘top’>{1,4} in the inset shorthand definition suggests so. If it is the case and the quirk is applied in the parse time, I suppose that logical properties should be parsed without quirks, and the current Blink behavior should be considered a bug.

[css-meeting-bot]

The CSS Working Group just discussed Should the `inset` shorthand allow quirks in their lengths like the individual properties do?, and agreed to the following:

• RESOLVED: Do not allow quirks in 'inset' shorthand

The full IRC log of that discussion

<dael> Topic: Should the inset shorthand allow quirks in their lengths like the individual properties do?

<dael> github: https://github.com/w3c/csswg-drafts/issues/3525

<dael> fantasai: Top left bottom and right prop in quirks mode allow px without px unit. 100=100px. DO we have same quirk for logical longhands? Related is what about inset shorthand?

<fantasai> (inset shorthand shorthands the physical properties, top/left/bottom/right)

<dael> TabAtkins: I don't see a reason to allow quirks in new properties. Need to not rely on top because that imports quirks

<bradk> No quirks

<dael> fantasai: We can add a note saying this notation doesn't import quirks

<dael> astearns: I'm agreeing shorthand shouldn't have quirks and deal with that in spec definition as well as add test cases

<dael> astearns: Other opinions?

<dael> astearns: Obj to not allowing quirks in 'inset' shorthand?

<dael> RESOLVED: Do not allow quirks in 'inset' shorthand

[emilio]

I think we can close this unless anybody thinks the spec needs more clarification. Feel free to reopen if so. @ericwilligers added wpt tests for this (thanks!).

[ewilligers]

Blink is fixed.

[Loirooriol]

WebKit is also fixed.