I had this too on Zen (firefox fork). Hard reloading the page with (ctrl+shift+r) fixed it for me.
1 Like
As mentioned below, we have the discourse hosted by them selves. kde.org might uses own instance or they not updated the same time/version.
I saw on meta.dscourse.org:
Labels and descriptions missing from nested object settings - Bug - Discourse Meta
This would explain why remove cached data works. There were definitely some changes.
Does this remove the saved passwords etc?
It doesn’t, from my (uneducated) observations it clears the cache and forces the browser to load the whole page again, it sometimes fixes problems. (Aka. it unloads the page and loads it again fully, normally a bit of data of the website stays which could have problems)
1 Like
I made a ticket by discourse. They might have some insight what happens.
2 Likes
They gave answer. We should test with save mode option when expecting issues with the theme (fore more details see link above):
# http://discourse.example.com/safe-mode
# For us it would be:
https://discussion.fedoraproject.org/safe-mode
Please do not forget to do what @joshisaurio proposed:
Cltr & Shift & r
Should work on all browsers.
3 Likes
Reloading the page does not work.
Unfortunately, like all big tech companies, their first reply to your issue raised indicates they have not read this thread and have not looked into the issue.
At least they overflow the issue and also mentioned the Java script errors I got on the Debian installation.
We know how it is, from helping here on Fedora. We have to dig deeper and give more information that they can go into details. Other ways it gets an Q&A game 
as we play it so often.
Maybe @mattdm knows more what we made from our side on the hosted instance?
One reason more to keep a mirrored test instance alive 
Please everyone, calm down and stop opening external tickets without explicit/expressive data.
This issue is annoying, but it is not super critical. So give our infra team some time to check out what is going on and if/how-far this is related to code/apps/plugins from us or code from upstream Discourse, in order to then forward to the Discourse team the data that is necessary (if anything is necessary at all).
If people open topics here without adding data about logs, context, configuration, etc., but only elaborations of symptoms, we would do the same. They have not received much to work on atm. More precise information is necessary to tackle this effectively and efficiently, and that is available only to our infra team and those with access to our instance(s).
So please give people time to tackle this effectively and efficiently, in an ordered way 
Feel free to use this topic to exchange and verify mitigations for the meantime.
I have seen so far clearing cache helps some, and in my case, I just set my Firefox from “strict” security settings to “standard”, and then back to “strict”, then reload → that works for some time and can be repeated. Maybe others find other ways that help them.
1 Like
I do agree with that. However a big part of us is volunteering here. So when we have a specific time frame to do so, then we would like to get access to the system as expected. Especially when using a external service, we should keep informing the same way as we do, when we have Fedora internal systems, which we regularly get informed what happens on them.
Related: Making sure you're not a bot!
If anyone finds it useful, you can get to your notifications without using the button, by visiting https://discussion.fedoraproject.org/u/insertusernamehere/notifications
1 Like
Just received this answer from @py0xc3 about the ticket I put on: https://forge.fedoraproject.org/infra/tickets/issues/13126:
We don’t host, but we did several customizations that are maintained by us. We cannot say if the issue is caused by something from us or something from upstream Discourse or some interaction between both (there was definitely a software update of Discourse some days ago, as the appearance of the flag tickets has strongly changed). Afaik, the Discourse work on our side was done in the past by @mattdm , though I don’t know if that is still applicable.
I’m not sure if another upstream ticket without data from our side is sufficient to clarify the case, but maybe you can trigger something through the contract that allows the Discourse team to access our instance(s) to verify 
Related:
https://discussion.fedoraproject.org/t/problems-with-discussion-fedoraproject-org-website-in-brave-and-firefox/180746/51
https://meta.discourse.org/t/fedoras-discourse-instance-made-some-troubles-could-the-support-please-give-some-feedback/395546
I don’t think this is super critical, and I asked users in the Discourse topic to focus for now on identifying and testing mitigations for the meantime rather than filing stuff upstream without expressive data.
To be continued.
Hi all, I escalated this to our Discourse enterprise support team. Hopefully I can get some feedback on this soon. Thanks for your patience while we figure this one out! 
4 Likes
Discourse staff here: I did read the whole topic before replying (on meta) as well as tried to reproduce (admittedly only on macOS / latest firefox - which is the only one I have quick access to)
Discourse being highly customizable it’s impossible to know what an issue might be coming from, especially on instances where 40 themes components (!!!) are enabled 
I’ve disabled one theme component (Better 404) that was generating a JS error but since I can’t reproduce it’s hard to do more.
I haven’t seen anyone tried to reproduce the issue in safe mode. Also, providing reproduction steps would be highly helpful 
5 Likes
It appears that safe mode (disabling theme and plugins) makes things work again.
Without safe mode, the “Search”, “Navigation menu” and “Notifications and account” buttons on the top bar don’t do anything. In safe mode, they work as expected.
1 Like
At this moment it works again, in normal mode using the Brave browser.
Also when using Firefox it works. I am not using safe mode.
Oops I spoke too soon. On the overview page of the discussion forum it does not work, only in this thread.
@decathorpe / @jandemus any JS errors in your browser’s console when the issue happens? If you could also share the backtrace that would be very helpful.
1 Like
I posted mine on the second post in this thread, if that helps.
Yes, I saw that one but unfortunately it’s not really helpful 
EDIT: wait, I can repro! 
1 Like