Array color input for Canvas2D

[fserb]

Allow RGB and RGBA colors to be specified by a sequence of floats on canvas2D.

Working proposal: https://github.com/fserb/canvas2D/blob/master/spec/array-color-input.md

(cc @whatwg/canvas )

[litherum]

In general, the WebKit team feels this is a good direction.

A few comments:

1. It should be possible to specify a color space along with these color values, like how the CSS color() function does.
2. Presumably, CSS Typed OM will want a color type. We should make sure that we don't have two competing color types. Instead, the same type should work for both.

[fserb]

We had a conversation with the CSS WG and CSS Typed OM is very close to have a Color type. Given that, we are working on modifying this to simply support the Typed OM color type described here: https://drafts.css-houdini.org/css-typed-om-1/#colorvalue-objects

[svgeesus]

the current proposal

[annevk]

#6609 too.

[annevk]

Given

ISSUE 6 TODO add stringifiers

in CSS Typed OM there is no need for this as developers could pass these objects anyway and implementations could optimize the stringification away if they wanted to. The one perhaps annoying edge case would have if someone had overwritten the stringification behavior.

Generally we avoid overloading if (default) stringification ends up doing the same thing. (Although to be fair those code paths might have been less hot than this one.)

[mysteryDate]

ISSUE 6 TODO add stringifiers

You're referring to CSSColorValue objects themselves? From the perspective of Canvas2D style input they definitely don't need to stringify.

[annevk]

I think you might be missing my point? I'm saying that if they will stringify I see less of a need to accept them directly as in most cases you cannot tell the difference. Implementations could optimize those cases where there is no difference if they wanted to.

[domenic]

In particular, if they stringify, then the existing setter in spec and implementations will automatically call that stringifier (like it does for every JS object) so there's no need to write spec text to handle it.

[mysteryDate]

Oh yeah, I definitely missed your point. Thanks for clearing that up.

[annevk]

Closing this as we decided in #6609 that stringification of CSSColorValue objects is sufficient. Folks interested in this issue might want to follow #8708 for new developments in this space.