Lots of minor notification issues

Gombessa

Including this one below—how does one have a negative number of notifications? 🤣

Pallas

Gombessa That negative number things happens to me all the time.

navindra

😲

Wow... that's original... dismissing more notifications than you have? I wonder if it's a race condition or something. Or maybe someone posted something, then deleted it and somehow the notification got dismissed twice... weird stuff. I'll have to play around on the test site and see if I can repro this.

navindra

Seems like it's a thing:
https://discuss.flarum.org/d/22789-notification-for-a-reaction-thumbs-up-displayed-as-1/

It's just a UI bug. Everytime a notification is marked as read, the number goes down by one. If you manage to trigger "mark as read" multiple times for the same notification (could happen with slow network maybe), there's a possibility of the number going negative.

HeartKeys

Oh okay, so the explanation sounds a lot less nefarious than the UI deciding to steal notifications for fun!

Gombessa

Makes sense! Probably happens easily enough where they could just "if n<0 then n=0" it.

rogerch

Gombessa Makes sense! Probably happens easily enough where they could just "if n<0 then n=0" it.

The software engineer in me would prefer to figure out the root cause and fix it. "if n < 0 then n = 0" would prevent a negative count but the fact that the accounting is wrong means there could be situations in which you would miss some notifications.

Josephine

Sometimes I get a red bar with oops! Something went wrong. Please reload the page and try again.

Half an hour ago there were two, I was just reading a topic.

navindra

Josephine That coincides roughly with the time that I rebooted the host server.

The server's been pestering me about pending upgrades and I finally decided to just update and reboot it. I kind of like having infinite uptimes on my machines, but these days it's not very advisable to forgo the reboots especially if security patches are involved.

The update itself did restart the docker containers, so that would have been interruption number #1. Then the host reboot would have been interruption number #2. Both ought to have been very quick < 30 seconds, but that may have been enough to interrupt long running API polling and such.