vmpstr

> At this point it's a race between the decode and the task that was queued while updating the image data. If the decode finishes first, both promises will be...

The non-normative note on https://html.spec.whatwg.org/multipage/embedded-content.html#dom-img-decode seems to imply that updating the image data also happens in a microtask, but the spec doesn't agree. I agree that we need to fix...

To rephrase (please correct me if I misinterpreted what you mean), ```js img.src = "foo.png"; img.decode(); // do something here that causes // 1. the browser image cache to evict...

Ah, yeah I'm not really sure what happens in that situation if you just append the img to the DOM without decode(). Presumably if the new data is loaded, then...

> > It also seems to mean that doing the second img.src = "foo.png" doesn't conceptually change anything. > > Why not? Until that happens, the image is pointing to...

> I mostly want the spec to make sense and browsers to actually be following it. Right now we have neither... Yes, I agree. We should fix the spec to...

We ran into a similar issue in a context of #9890 where we'd like to have a property that limits how far up the tree `view-transition-name` is exposed. The initial...

I think a timeout idea would work to solve this as well. I agree with you that there may be better architectural solutions (async searching) but I don't intend this...

In that case, this might be OK to close since the behavioral differences from the unique set in the implementation would not be observable. But yeah, cleaning up the spec...

Follow up question here is on the sizing of the root capture: 1. Do we size ::vt to the padding box of the scope? If so, is the root capture...