| Age | Commit message (Collapse) | Author |
|---|
|
The Spec now allows it, even if with a disclaimer that
it's subject for further removal.
|
|
Hopefully manages cases like #347 in a better way.
|
|
|
|
That's a matter of another upcoming commit.
|
|
|
|
|
|
|
|
Update the cache for sure if unread counters were changed
|
|
|
|
|
|
|
|
In Quaternion, I was able to reproduce a process where
`d->notificationCount` and `d->highlightCount` were changed in
`Room::updateData()` meanwhile `roomChanges` indicated `NoChange`
thus the cache became permanently inconsistent with running state.
Without deep understanding of `roomChanges` logic, I've applied
this straightforward change which made this phenomenon
unreproducible.
|
|
Closes #344.
|
|
Closes #326.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This changes from internal linkage for every translation unit including
e2ee.h to external linkage across translation units but without the need
to create a .cpp file just to define and init the constants.
|
|
|
|
|
|
...because /W4 is explicitly added instead.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
[skip ci]
|
|
They've been deprecated for almost a year by now.
|
|
|
|
The fallback to libQMatrixClient still remains.
|
|
|
|
[skip ci]
|
|
|
|
|
|
|
|
std::string is still a thing, after all.
|
|
|
|
|
|
[skip ci]
|
|
|
|
|
|
140 impose coding standard
|
|
|
|
That never really worked.
|
