[css-backgrounds] Change canonical serialization for box-shadow

[csnardi]

The spec declares the syntax for a shadow in box-shadow to be:

<shadow> = inset? && <length>{2,4} && <color>?

However, Firefox, WebKit, and Chrome all serialize this as <color>? && <length>{2,4} && inset?. This matches the serialization for drop-shadow() (see https://drafts.fxtf.org/filter-effects/#funcdef-filter-drop-shadow) and text-shadow (see https://drafts.csswg.org/css-text-decor-3/#text-shadow-property), which both are declared as <color>? && <length>{2,3}.

For consistency, <shadow> should be defined as:

<shadow> = <color>? && <length>{2,4} && inset?

in order for serialization to match 3 out of the 4 major browsers and text-shadow/drop-shadow().

[fantasai]

Pushed out tentative edits, waiting for WG approval.

[css-meeting-bot]

The Working Group just discussed [css-backgrounds] Change canonical serialization for box-shadow, and agreed to the following resolutions:

• RESOLVED: switch the order to match current implementations

The full IRC log of that discussion

<dael> Topic: [css-backgrounds] Change canonical serialization for box-shadow
<dael> github: https://github.com//issues/2305
<dael> TabAtkins: Going by standard rule that default serialization should match grammar that implies color after link but impl oes color before links. We should change grammer to match impl.
<dael> Rossen_: Looks like edits were pushed out. So only thing we need to do is get consensus.
<dael> Rossen_: Any other opinions as to why we shouldn't do this?
<dbaron> This is part of why I am not a fan of that general rule being imposed without doing compatibility testing first... although I think we may have rolled back the rule a bit?
<bradk> +1
<AmeliaBR> +1 to matching reality & also other properties
<dael> Rossen_: Objections to switch the order to match current impl?
<dael> RESOLVED: switch the order to match current implementations

[csnardi]

@fantasai I created two PRs for tests that should verify this change; they should be linked to this issue.

[TalbotG]

@csnardi If you need more tests, you know that I have 15 tests targeting box-shadow in

https://www.gtalbot.org/BrowserBugsSection/CSS3Backgrounds/

https://www.gtalbot.org/BrowserBugsSection/CSS3Backgrounds/box-shadow-021.xht
...
https://www.gtalbot.org/BrowserBugsSection/CSS3Backgrounds/box-shadow-036.xht

[csnardi]

@TalbotG Thanks, I didn't know that! I'll take a look at them a little bit later and add in any parsing cases I missed to the first pull request linked above.