* Resume playing audio at the appropriate time
* Re-emit call events (they were the exception before - all other events
were re-emitted through the MatrixClient)
* Fix an audio/video typo
This function is *always* called when a MatrixClient is created, e.g. in
an appservice. If the environment does not support WebRTC, this is not
necessarily an error; it is expected in many situations.
Fix a small typo
Signed-off-by: Dexter Chua <dec41@srcf.net>
After a fair emount of investigation, it looks like there's not
really a particularly elegant way to have an environment where
setTimeout has a return type that makes sensible assertions in both
node and the browser. Ideally we want it to be something that asserts
that we don't try to use it as a NodeJS.Timeout (ie. call the methods)
because that will break in a browser. In practice, this involves making
wrappers or redefining the timeout functions or some similar kind of
hackery, the evils of which probably offset having perfect typing.
https://matrix.to/#/!bEWtlqtDwCLFIAKAcv:matrix.org/$vpFWl7p1_8A858RAccO0gud3sNVWIDNxNELRjdqaZQ4?via=matrix.org&via=mozilla.org&via=vector.modular.im
for the discussion
Don't send events all the way via the mock HTTP backend: we're not
trying to test that here. This meant we weren't actually getting
into the right state because the request to send the invite never
actually returned. Now this works, we need to clear the invite timer
otherwise jest has a timer hanging around at the end of the test
(plus we should be doing it anyway).
Send party_id on events and check the party_id of incoming events matches
Includes a basic test to assert that it actually does: we should
build out a decent test suite for calls as there's a lot of edge-case
functionality that can break and slip through the cracks (eg. glare).
This is a start.
Fixes https://github.com/matrix-org/matrix-js-sdk/issues/1511
Start the migration to v1 VoIP by supporting m.call.reject, which
we'll send if the caller says they're v1. Our version stays as v0
for now, until we speak the rest of v1.
Honour the default reaosn in a hangup being user_hangup.
* Set 'connecting' state before sending answer, otherwise it can
race with ICE connecting
* Ignore completed ice connection state: connected is what we care about
* Null-check remotestream when stopping media
* Comments
Use the new local-age field for deciding whether a call is still
valid or not. Also add a grace period so we don't ring half a second
before the call becomes invalid.
