-
-
Notifications
You must be signed in to change notification settings - Fork 2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Stuck notification in thread #23921
Comments
I reproduced this issue in the Delight room. The room shield appears as red. The notification counter appeared as grey. Selecting the thread did not reset the counter. When talking into the room, it seems that the notification disappeared. |
One cause of this is when the edit event is the latest thing in the unread thread, I wonder if the load-thread-from-server API doesn't include the edit event and thus it not being in the timeline results in the RR being set on the wrong event (the original event ID rather than the ID of the edit event) I have show hidden events enabled. Given how frequent edits are, this will be happening whenever the latest event in a thread is an edit. And entirely destroys productivity by wedging an unread dot on the space, the room and the thread in the room list. |
The client seemingly has no way to know the real latest event id in the thread as the server doesn't give it a total order list of all the events, only the 1st-order related events (thread responses), edits and other 2nd-order relations are bundled inside the 1st-order relations' unsigned fields. This issue impacts all users & devices who were not open/online at the time a thread response gets edited, all the while until another response is added to the thread. Based on this and how common edits are, I'd argue most users will see this from every now and then, so this sits just below |
This The Currently trying to find an appropriate fix for this issue |
@gsouquet does that apply for threads entirely loaded from the server? I doubt the full m.replace event will be received in those cases |
* Remove extensible events v1 field population on legacy events ([\matrix-org#3040](matrix-org#3040)). * Improve hasUserReadEvent and getUserReadUpTo realibility with threads ([\matrix-org#3031](matrix-org#3031)). Fixes element-hq/element-web#24164. * Remove video track when muting video ([\matrix-org#3028](matrix-org#3028)). Fixes element-hq/element-call#209. * Make poll start event type available (PSG-962) ([\matrix-org#3034](matrix-org#3034)). * Add alt event type matching in Relations model ([\matrix-org#3018](matrix-org#3018)). * Remove usage of v1 Identity Server API ([\matrix-org#3003](matrix-org#3003)). * Add `device_id` to `/account/whoami` types ([\matrix-org#3005](matrix-org#3005)). * Implement MSC3912: Relation-based redactions ([\matrix-org#2954](matrix-org#2954)). * Introduce a mechanism for using the rust-crypto-sdk ([\matrix-org#2969](matrix-org#2969)). * Support MSC3391: Account data deletion ([\matrix-org#2967](matrix-org#2967)). * Fix threaded cache receipt when event holds multiple receipts ([\matrix-org#3026](matrix-org#3026)). * Fix false key requests after verifying new device ([\matrix-org#3029](matrix-org#3029)). Fixes element-hq/element-web#24167 and element-hq/element-web#23333. * Avoid triggering decryption errors when decrypting redacted events ([\matrix-org#3004](matrix-org#3004)). Fixes element-hq/element-web#24084. * bugfix: upload OTKs in sliding sync mode ([\matrix-org#3008](matrix-org#3008)). * Apply edits discovered from sync after thread is initialised ([\matrix-org#3002](matrix-org#3002)). Fixes element-hq/element-web#23921. * Sliding sync: Fix issue where no unsubs are sent when switching rooms ([\matrix-org#2991](matrix-org#2991)). * Threads are missing from the timeline ([\matrix-org#2996](matrix-org#2996)). Fixes element-hq/element-web#24036. * Close all streams when a call ends ([\matrix-org#2992](matrix-org#2992)). Fixes element-hq/element-call#742. * Resume to-device message queue after resumed sync ([\matrix-org#2920](matrix-org#2920)). Fixes matrix-org/element-web-rageshakes#17170. * Fix browser entrypoint ([\matrix-org#3051](matrix-org#3051)). Fixes matrix-org#3013. * Fix failure to start in firefox private browser ([\matrix-org#3058](matrix-org#3058)). Fixes element-hq/element-web#24216. * Correctly handle limited sync responses by resetting the thread timeline ([\matrix-org#3056](matrix-org#3056)). Fixes element-hq/element-web#23952.
Steps to reproduce
Unclear, but this is what happened:
Outcome
What did you expect?
The notification count should go down from 2 to 0
What happened instead?
The notification count went down from 2 to 1, and the thread remained marked as unread in the thread list. Restarting the client was necessary to clear the notification.
Operating system
NixOS unstable
Browser information
Firefox 107.0.1
URL for webapp
develop.element.io
Application version
Element version: c473cb6-react-9914b0bafd23-js-ccab6985ad55 Olm version: 3.2.12
Homeserver
Synapse 1.72.0
Will you send logs?
No
The text was updated successfully, but these errors were encountered: