Add RTCIceTransport method to remove candidate pairs

[sam-vi]

Fixes #170.

This PR extends the RTCIceTransport interface by adding a new method to allow an application to remove candidate pairs that are no longer needed for the peer connection. The new method removeCandidatePairs immediately removes the supplied candidate pairs, which are then sent in non-cancelable icecandidatepairremove events.

---

Preview | Diff

[dontcallmedom-bot]

This issue was mentioned in WEBRTCWG-2023-09-12 (Page 66)

[aboba]

@pthatcher PTAL

[henbos]

FYI reviewers, the PR is now rebased and much smaller than it looked before

[henbos]

I overall like this, just have a question about the "immediate-ness" of this operation

[henbos]

LGTM

[pthatcher]

Seems fine to me except for the sentence I mentioned.

[jan-ivar]

What's the use case that requires this new method?

[dontcallmedom-bot]

This issue was discussed in WebRTC December 5 2023 meeting – 05 December 2023 (PR #175: Add RTCIceTransport method to remove pairs)

[henbos]

Is this ready to merge at the next meeting?

[jan-ivar]

Overall I don't mind this API, so I'm approving it for the sake of progress.

However, I worry there are some API races here that we still need to sort out. We appear to have two async methods, select and remove basically, which appear as if they can race. We should clarify what's supposed to happen then.

For instance, what happens if someone calls both methods at once with the same argument?

My recommendation of adding an internal slot in a follow-up might help clarify this.

[sam-vi]

Overall I don't mind this API, so I'm approving it for the sake of progress.

However, I worry there are some API races here that we still need to sort out. We appear to have two async methods, select and remove basically, which appear as if they can race. We should clarify what's supposed to happen then.

For instance, what happens if someone calls both methods at once with the same argument?

My recommendation of adding an internal slot in a follow-up might help clarify this.

Thank you!

I think the internal slot and validations should address a lot of the race conditions. I would certainly want to fix any remaining issues once the proposed changes are submitted. I am happy to continue discussing and iterating.

[henbos]

OK let's merge this, @sam-vi vill you care a follow-up issue regarding internal slots and validation?