There's no need for this property stuff

sounds very sensible +1

Services Done (in a branch)

kwin
ksmserver
kcminit
plasmashell
krunner

From the viewpoint of baloo_file, the process just hangs, which is bad.

I found out why this was added. 7e4b9688d52f55828ba4a5b82e9d8fe5b1a5ff94

We need to find something more reliable than noDecoration, that includes so many other things.

More generally, should we be porting towards Kirigami.Units everywhere?

I thought the bug was that we weren't checking the RC from KService::serviceByStorageId before using it

I don't get why it would overlap if it's not visible since we're in a RowLayout here?

This patch fixes the issue, but I do still see the sidebar

FWIW, Plasma 5.17 uses PC2, so I have no real objections to this patch in the context of this patch. I'm only questioning if that frameworks change is safe, given if it is causing changes elsewhere

It still sounds like you're saying "we've just made a breaking change in frameworks "

Please ensure that you've run the relevant plasma-framework/tests/components in qmlscene

I'm going to push soon if I don't hear soon. It solves a real world problem and it's had one approval.

In more long term we need to fix ConfigModule do you think is better to do a setNeedsSave(false) after loading or emiting signal in all case ?

Given we have that .endsWith("Wayland") check, there isn't any rush to remove it

About KCModuleQML didn't investigate a lot but KCModule (parent class) emit a changed false on showEvent (the way the handle to set need save to false.

Take this code:

The SDDM KCM is broken right now because the kde_settings.conf file is never created with the right permissions. This means that the KCM can't read the settings file.

This patch however fixes that for me

I still don't fully understand the bug and the fix

Cool, I'm happy as long as there's a known reason.

Can we add something to x11_client_test ?

What I don't understand from this patch is that plasmathemeexplorer which has the same anchors but uses a gridview doesn't have this problem
Nor plasma NM's Add Connection Dialog

Urgh. I suspect this was a hidden bug from when runnermanager never emitted teardown.

Do you have commit access?

What the original code was trying to do was distribute the whitespace among cells, right now if you resize to be 5.8 columns wide we want to spread that out, not have a gap on the white.

The component changes seem unrelated

Seems sensible +1