[css-typed-om-1] Does computedStyleMap() always return a new object?

[bzbarsky]

https://drafts.css-houdini.org/css-typed-om-1/#computed-stylepropertymapreadonly-objects

If so, it should be annotated [NewObject].

That said, why would it? There's no reason for it too, afaict, just like attributeStyleMap and styleMap don't.

[darrnshn]

Blink currently returns a new object each time. I'm not sure what the reason is, my guess is maybe it's to match getComputedStyle()?

[bzbarsky]

Right, but for getComputedStyle() it's sort of a function of window and element, so returning new thing each time sort of makes sense because caching it would be a bit weird and complicated. In this case we just have an element.

[tabatkins]

Yeah, we should be able to return the same object each time, I think. It's explicitly live, so the get() calls will always reach down to the real element anyway. I guess that means it should use a [SameObject] annotation?

[bzbarsky]

That would work.

[bzbarsky]

(Same object every time means it needs to get cached on the element somewhere, so there's that tradeoff.)

[tabatkins]

Ah, I don't actually use [SameObject] - that's for attributes. I just need to specify in the method that it caches on first access.