This extra debug logs may help isolate the cause of
https://github.com/vector-im/element-web/issues/16194.
These changes also fix a related (but most likely different) failure mode: if a
failure occurred in the `encryptionPreparation` async task, we would skip trying
to prepare in all future attempts for that room. This change ensures prep
failures are logged and we resume prep attempts on the next call from the
application.
We may not have managed to set up the backup yet when we get keys. So we should
unconditionally queue up the keys for backup, so that when the backup is set up,
they will be sent instead of dropped.
The timeouts on the two phases of olm session creation are 2 and 10
seconds respectively, so sessions will fail if servers take more
than 10s to respond. Now that we have two phases, we can afford to
wait longer on the second one because the user's isn't waiting for
it to finish before the message will send, so increase it to 30s
so servers have more of a chance to respond.
I think this was attempting to remove the events from _pendingEvents
but a) it wasn't and b) it probably shouldn't be because the retry
itself will re-add them when the decryption attempt starts and remove
them if it succeeds.
Also fix what was presumably a c+p fail.
don't need to do the shorter timeout when doing preparation to encrypt, and
skip the second phase if the first phase already took longer than a normal
otk claim
- allow applications to pre-send decryption keys before the message is sent
- establish new olm sessions with a shorter timeout first, and then re-try in
the background with a longer timeout without blocking message sending
Currently we split the device messages up to limit the number per call,
but that can end up splitting messages to a given users device over
separate API calls. This is fine, but means that the server can't e.g.
bundle them into a single EDU for remote users or sanity check that the
client is sending to the right set of devices (i.e. its device list
cache isn't wrong).
The bulk of this is just export/import changes, though there's a couple pieces to highlight:
* We no longer use default exports. This is because it's discouraged by the JS community, though not in any official capacity.
* We now use `polyfillSuper` for some prototype inheritance because the tests, and sometimes webpack, break on "cannot call EncryptionAlgorithm without 'new'". It's very much a workaround, and definitely not needed when we use real classes.
There is some import shuffling to help keep the imports clean - this was done by my IDE.
