It doesn't make sense to expose it for a focused timeline.
Note there's no way for timeline to change focus, at the moment; this
would require special handling here, because one could subscribe to the
back pagination status on a live timeline first, then switch the
timeline to the focus mode. The stream wrapper could then observe other
states that aren't expected by the converter method.
This is actually required for one use case: if the event cache triggers
back-pagination in the background *before* a timeline is opened, it's
possible the timeline observes pagination is running at the beginning,
and doesn't know if more back-pagination is required or not.
This wraps the `Paginator::status` stream by reading the
hit_timeline_start() from the room pagination and adding it to the
`Idle` state.
This whole method is a bit broken. It assumes that, when we did an import from
backup, the backup that they came from is the same as the "current" version.
We *could* add another argument, but to be honest I find the whole method a bit
pointless. It's not particularly tied to the `BackupMachine` abstraction. Let's
instead expose `Store::import_room_keys` and
`ExportedRooomKey::from_backed_up_room_key` publicly, and tell callers to use
that directly.
When we are importing a batch of room keys, use the newly-added
`CryptoStore::save_inbound_group_sessions` method instead of
`CryptoStore::save_changes`.
To do this, we need to pass the backup version into `Store::import_room_keys`
instead of just `from_backup` flag.
When we add a batch of inbound group sessions to the store, if they came from a
backup, we need to record which backup version they came from.
`CryptoStore::save_changes` doesn't give us an easy way to set the backup
version (we could add it to the `Changes` struct, but ugh).
So, we add a new method `save_inbound_group_sessions` which takes the backup
version as a separate param. This works fine, because whenever we import
sessions there are no other changes to store.
The new method isn't used outside of tests yet -- that comes in a follow-up
commit.
This moves code around, and tweaks the behavior:
- `WeakRoom::get()` returns an `Option`, that will be None if the client
is missing or the room is missing.
- `PaginableRoom` for `WeakRoom` will now return a default response if
the room could not be reconstructed from the weak room. It's fine to do
so because we're in a shutdown context.
It's possible the timeline starts with a read marker, in which case the
first item won't be a day divider; in that case, check for the next
item, if present.
test_start_with_read_marker
This patch adds another check to ensure `AsVector` generates
`VectorDiff`s that, once combined, produce an expected `Vector`. It
avoids errors when unit testing `VectorDiff` alone.
