So one thing I need to investigate further is that it feels to me that intersection observer running before the animation frame callbacks seems wrong, I think it should run substantially after them, with a bunch of other stuff in between, since rAF is primarily a writing callback (so should be early, before the natural style/layout), whereas IntersectionObserver is primarily a reading callback (so should be late, after the natural style/layout). But I suspect that some of my concern there may be that I think where HTML describes the running of the animation frame callbacks being wrong. I think this requires a good bit of digging into (to see what current specs say, and how interoperable existing implementations are); it's related to our #72.

I have concerns about the rootMargin being a string based API. This should be using CSS Typed OM if not now, at least some point if the future. The spec should at a minimum have a note about a typed OM version of the margin.

Gentle ping to remind you that we're looking for issues to be filed by the end of next week. Thanks.

Discussed on today's call. Sounds like the delivering timing needs a look.

/cc @szager-chromium

Discussed on today's telecon.

I'd like to leave this open to monitor the progress on w3c/IntersectionObserver#263 and whatwg/html#3246 etc.

Per today's F2F, now for me to discuss open issues with @szager-chromium

Pinged @szager-chromium; unclear where things are now. Will circle back.

I want to hopefully close the loop on the comment by @dbaron:

So one thing I need to investigate further is that it feels to me that intersection observer running before the animation frame callbacks seems wrong, I think it should run substantially after them, with a bunch of other stuff in between, since rAF is primarily a writing callback (so should be early, before the natural style/layout), whereas IntersectionObserver is primarily a reading callback (so should be late, after the natural style/layout).

You are quite right that IntersectionObserver should run after rAF callbacks; I just created a pull request for that change. The original language there was a mistake.

This sidesteps the issue of when exactly style/layout/paint gets updated, but that issue is better addressed in the separate spec issues you called out.

Yep, that PR improves things, thanks.

... but I think, as I mentioned in https://github.com/w3c/IntersectionObserver/issues/263#issuecomment-448418727, that the steps should be still later. I think this requires writing a bunch of tests to check what happens today, though...

@szager @dbaron what is the state of this now?

We separated out the more general bits into #489, but we're perhaps still leaving this open to wait on w3c/IntersectionObserver#263.

@plinss and I are looking at this in a breakout in our virtual "Cork" face-to-face.

So we were leaving this open for w3c/IntersectionObserver#263, which is in turn waiting on whatwg/html#3246, which is in turn waiting on w3c/csswg-drafts#5115. While it would be nice to fix that sequence of things, we probably don't need to leave this review open for it.

These are also all referenced from #489.

OK, we discussed in mini-plenary and we're happy closing this issue at this point. Thanks for requesting TAG review... a long time ago. Sorry this took so long.

Thank you @dbaron and @plinss.