To avoid an O(n^2) situation with every relations container trying to process
every redaction that may occur in a room, this switches to an event-level
notification, so that the specific relations container who cares can listen to
just the events it wants to know about.
This special cases pending relation events to go directly to the timeline sets
and ignores the `pendingEventOrdering` option. This feels a bit strange in the
code, so we should revisit this choice when we stabilized relation support.
use sendRoomKeyRequest with a new resend flag, instead of cancelRoomKeyRequest,
when requesting keys, so that we make sure that we send a new request if there
is no previous request
fixes https://github.com/vector-im/riot-web/issues/6838
There is a legacy top-level 'membership' field on events. We were
interning the normal one in 'content', but not this one, so the
legacy field was still keeping a copy of the string 'join' / 'leave'
etc in memory for every member event.
This make the interning code know about this field.
My test account had 37MB (shallow) of Date objects knocking around
in memory. This gets rid of them. I can't see any appreciable
difference in the time taken to switch rooms (where now we recreate
a bunch of Dates that previously would have been cached).
We were relying on being able to override toString in DecryptionError, which
(a) doesn't work thanks to https://github.com/babel/babel/issues/3083, and (b)
was a bit naughty anyway. Instead, just add a detailedString property and use
that.
... instead of having it call event.setClearData.
The main advantage of this is that it fixes a race condition, wherein apps
could see `event.isDecrypting()` to be true, but in fact the event had been
decrypted (and there was no `Event.decrypted` event on its way).
We're also fixing another race, wherein if the first attempt to decrypt failed,
a call to `attemptDecryption` would race against the first call and a second
attempt to decrypt would never happen.
This also gives a cleaner interface to MatrixEvent, at the expense of making
the `megolm` unit test a bit more hoop-jumpy.
These terms were somewhat confusing (and, in the case of megolm, misleading),
so replace them with explicit senderCurve25519Key and claimedEd25519Key fields.
Remove some we don't care about. Set some other ones we do care
about but don't currently adhere to to warn. Set the max warnings
threshold to the current number of warnings, so we don't introduce
more of them. Fix a bunch of legit lint errors and add exceptions
to various places in the test code that does funny things with
'this'.
