Hi. Is it normal - two absolutely different timestamps for the same notification? Thank you.
Description
Event Timeline
@IKhitron - no it should not. The cross-wiki bundle (when the bundle is not expanded) the timestamp should be the most recent one.
Did you mark as read some of the notifications in the cross-wiki bundle and then the bundle timestamp became different or a new message has come when you were on the page? Anyway, I will be checking couple of scenarios to see where that discrepancy may come from.
I wonder if this might be a bug in the transition flag code, but I'll wait to see if Elena can find reproduction steps for this.
I saw on my local wiki two cross-wiki notifications in bundle. Did not pay attention on timestamps. I clicked on the blue dot on one of them. it disappeared, and this is the result.
Generally, the problem is - the timestamps are not promptly, i.e. automatically updated.
Case 1: bundle timestamp > most recent timestamp of any of the notifications
Case 2: bundle timestamp < most recent timestamp of any of the notifications
The steps to reproduce the issue are
- When user A is logged in, user B sends user A a cross-wiki notice/alert.
- User A without refreshing the page, opens the flyout. Several things happen
- if necessary, the name of foreign wiki gets added to the list of foreign wiki bundle list
- the cross-wiki bundle timestamp gets updated to the timestamp of the most recently received notifications
- when the cross-wiki bundle is expanded, there is a mismatch between cross-wiki bundle timestamp and the most recent timestamp of the bundled notifications.
So, user A sees that the bundle timestamp is earlier than any timestamp of any notifications in the bundle.
@IKhitron I agree that the issue should be closed. There were some backend improvements that made timestamps more responsive to updates.
I could not reproduce the case of bundle timestamp > most recent timestamp of any of the notifications
The case of bundle timestamp < most recent timestamp of any of the notifications is still relatively easy to reproduce but only for some edge cases as described in my Case 2).