It's now just the normal cursor when resizing with Alt + Right click drag.

Is it fine if we keep the files in the current directory and just add the additional directories outputs and control in later patches. The hash values should never coincide with these names. So it might be ugly, but work without problems.
Moving the configuration files in a separate configs folder could then be done in KDE6 since with D17007 some of them would then be invalid anyways.

In D17007#364724, @mart wrote:

could a kconfigupdate script be provided to migrate config?
screwing up multiscreen config between updated is quite of an issue

In D16997#361978, @ngraham wrote:

I'll admit just looking at the new UI, it's not entirely clear what either of the options do. Can you help me understand? Thanks Roman!

I replied to your mail to the plasma-devel thread. We can continue the discussion there. Maybe we find an alternative solution for the radio buttons better suited for what we want to communicate to the user.

In D17012#362329, @ngraham wrote:

My vote is to re-work this patch to do that instead.

Dolphin does this already. Might be an idea to only disable the Delete or Move to trash entry though. Not sure.

In D16988#362323, @rooty wrote:

Clipboard works like this but the coding is beyond my abilities. Krunner does this too except it opens up with the search field in focus. This Kickoff patched version can't do that, and I've yet to figure out how to make it be able to.

I believe in you! :) If you need quick help come join the Plasma IRC channel.

In D16988#362160, @rooty wrote:

(1) Kicker takes the focus off the search field when you go through the results by either mouse or keyboard. This makes it easy to go back to the search field by typing or using the arrow keys (to add more words etc.)

Why does it make it easy?

(2) Krunner becomes unfocused if you use the arrow keys, but not if you hover over the search results with your mouse, and it's still possible in either case to append your search same as in Kickoff.
(3) Kickoff (the old, the master and this patch) seems to leave the search field focused at all times while you go through the results, and it's possible to add more characters to the search like in (1) and (2).
(4) Clipboard behaves pretty much like this patch except pressing the Esc key doesn't make it unfocused again.

I like all the changes, but they should go into separate patches:

From a Ux point of view Ui elements should be disabled instead of hidden if they are not available but the user expects them to be there. The "Rename" entry currently gets disabled. You think it makes more sense to hide it instead?

In D16937#362012, @ngraham wrote:

You know what, I tried this out and I like it. The differing line widths don't actually seem all that weird when actually using it for a a bit. I think this is fine.

Other VDG and/or Plasma folks, what do you guys think?

This diff contains some fix, that need to go into earlier diffs in this patch series.

Checking the edid is necessary, because it might be set asynchronously and before that the hash is based on the output name only.

In D16937#361203, @filipf wrote:

Eike's Simple Menu is a good example of separator use - the separator is subtle and doesn't fully break up the the plasmoid:

On top of that he made the search field unfocused by default, which can really help with transparent themes.

In D16937#360676, @rooty wrote:

FWIW make sure to test your changes with all four possible orientations.

I actually figured out how to do that just before my latest edit, my bad. It seems to work sort of fine with the tabs on the left or the right, except for the lower separator (to the left/right of the tabs on the right or left, respectively) that seems to appear shorter than the length of the separator to the left/right of the tab bar before you start scrolling to the bottom.

...

I don't know if this was on purpose or if it's a bug, but the bottom separator becomes thicker to the left/right of the tab bar before/when you start flicking through the menus in the previous design (no opacity settings / width: root.width) before reaching the end (of a given menu)

...

There is one other special case: a configuration with several identical displays. We maybe want the individual output data globally to be overridden, but keep the output values different in this specific configuration.

• Add back Q_SLOTS specifier to be sure

In D16792#359306, @davidedmundson wrote:

git history tracking

Note, that the configReady function was a public slot in an exported class, so is it safe? I would like to change other stuff in this class definition. I don't know why the class is exported and who consumes it.

In regards to backend I hit a wall in that the variable value if display properties should be saved for all configs or only for this specific one need to be communicated between the KCM and the daemon. The communication mostly uses the kscreen backend though, which means it is tunneled through the Wayland compositor or XCB.

• Use SOCKET_OPTION variable

In D16789#358969, @davidedmundson wrote:

I'm not loving the name, the goal of this patchset is to move all of the wayland logic out of the platform.

initWaylandOutput/initXdgOutput are now handled in setEnabled so can be private to AbstractOutput.
We should aim for the same with outputdevice.

I'd rather this was just "AbstractOutput::init()"

Code otherwise is fine

When user changes properties in KCM, he might either decide to change data for the current configuration only set it as new "default" for this output on every other configuration, which does not overwrite with configuration-specific values. How should this be communicated to the user and how should he switch around for the current config operation?

• Remove unrelated change