bug: thread root aggregated latest event missing sender field #2167

New issue

Open

opened 2026-08-22 14:43:47 +00:00 by ajbura · 4 comments

ajbura commented

2026-08-22 14:43:47 +00:00

What is the bug?

m.thread relations added by server to the thread root unsigned content is missing required sender property.

Steps to reproduce

send an event.
send a thread reply
refresh the app
observe the root event unsigned content

Server name and version

26.7.3

Additional context

No response

### What is the bug? `m.thread` relations added by server to the thread root unsigned content is missing required `sender` property. ### Steps to reproduce 1. send an event. 2. send a thread reply 3. refresh the app 4. observe the root event unsigned content ### Server name and version 26.7.3 ### Additional context _No response_

c11y-thread-root-event-missing-latest-event-sender.json

3.2 KiB

ajbura added the

Bug

Status

Needs Investigation

labels

2026-08-22 14:43:47 +00:00

nex added

and removed

labels

2026-08-22 16:07:27 +00:00

ajbura commented

2026-08-22 17:21:27 +00:00

Author

it looks like the content of event is assigned to the latest_event instead of assigning the event itself.

link to problem area:

relations.latest_event = Raw::from_json(event.content().to_owned());

it looks like the `content` of event is assigned to the `latest_event` instead of assigning the `event` itself. link to problem area: https://forgejo.ellis.link/continuwuation/continuwuity/src/commit/9491be928d6ded74e8a2750dc52b7f299b9fd446/src/service/rooms/threads/mod.rs#L93

Henry-Hiles added the

Difficulty

Easy

label

2026-08-22 17:26:58 +00:00

Jade commented

2026-08-22 17:38:38 +00:00

Owner

Ah, this looks like a pain/problem because those functions are pre-continuwuity code and write the thread data directly to the database representation of the PDU. Might have to be fixed as a part of an (already needed) larger refactor, for historic messages at least

nex commented

2026-08-22 18:11:50 +00:00

Owner

I'm already working on refactoring how we handle unsigned data (and also backfilled events, but that's a separate thing), ftr. Would that perhaps be something I need to add to my list?