feat(ui): Add m.room.tombstone to the room list required_state

[Hywan]

This patch adds the m.room.tombstone state event to the list of events in required_state used by the RoomListService. The goal is to offer the possibility for the consumers to know whether a room has been tombstoned or not.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.87%. Comparing base (71abbeb) to head (aad7af3).
Report is 14 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4211      +/-   ##
==========================================
- Coverage   84.89%   84.87%   -0.02%     
==========================================
  Files         272      272              
  Lines       29142    29142              
==========================================
- Hits        24739    24734       -5     
- Misses       4403     4408       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[kevinaboos]

Echoing my comments from the Matrix Rust SDK room here, as requested:

I tested this PR out in Robrix, but unfortunately was not able to get any tombstoned info at all. Here's what I'm doing, just in case I missed a key step.

When I receive a diff with a new room from the room list service, i then attempt to call several different tombstone-related functions (which I think are all doing mostly the same thing under the hood):

• room.tombstone() always returns None.
• room.get_state_event_static::<RoomTombstoneEventContext>().await also always returns None.
• room.get_state_events(StateEventType::RoomTombstone).await() always returns an empty vec.

i know for a fact that I do have tombstoned rooms on the account that I'm testing with, so I'm expecting them to have tombstoned room state. Here's a side-by-side example showing how both rooms exist as separate rooms in the same session: the new version of Element Web/Desktop on the left, and the old tombstoned one on the right. If I understand tombstoning correctly, I'd expect the old version of the room (on the right) to have a tombstone state event.

[poljar]

What's the state of this PR? Do we need this or do we expect users to enable tombstone events themselves?

When I receive a diff with a new room from the room list service, i then attempt to call several different tombstone-related functions

Are you sure that the server sent the event to you? As far as I can see, we don't treat those events any differently from other important room state events, if we receive it we put it in the cache.

[kevinaboos]

What's the state of this PR? Do we need this or do we expect users to enable tombstone events themselves?

Hi @poljar, thanks for the response. I would argue that tombstone state should be a part of the sync service's required state, since without it, the client may accidentally display multiple old tombstoned rooms as separate unrelated rooms without being able to properly determine that they're replacements of one another.

When I receive a diff with a new room from the room list service, i then attempt to call several different tombstone-related functions

Are you sure that the server sent the event to you? As far as I can see, we don't treat those events any differently from other important room state events, if we receive it we put it in the cache.

I'm not 100% sure, no. I think that's part of the problem -- even with this change, I don't get any tombstone events from any rooms.

How might I accurately determine whether I've received the tombstone events from the server? (besides using the existing SDK APIs for this, like https://matrix-org.github.io/matrix-rust-sdk/matrix_sdk/struct.BaseRoom.html#method.tombstone)

[poljar]

How might I accurately determine whether I've received the tombstone events from the server? (besides using the existing SDK APIs for this, like https://matrix-org.github.io/matrix-rust-sdk/matrix_sdk/struct.BaseRoom.html#method.tombstone)

I like to use mitmproxy for this then you can observe the full /sync response.

[Hywan]

@erikjohnston Do you know if sliding sync on synapse filters out the m.room.tombstone events by any luck, or mess with them somehow?

[Hywan]

I wonder if this can be solved by element-hq/synapse#17540.