• Remove check for currentPort which is no longer needed

Thanks for the hint! I didn't know I could use FormFactor directly to reference the Device's formFactor property. Indeed now it looks much better.

• Use the formFactor attribute of the Device directly

• Use form_factor from device to select the icon for a profile

I wrote in the summary above the contents of the string variables we could work with so that it's clear I couldn't find any other way to extract that information. If you have any other idea (any other variable to use for the match?) I'm happy to test it.

Perfect, thanks, I'll do that

If I move it to the end, then the player device will show a headset correctly anyway since no other condition matches but for the capture device, a microphone icon will be shown instead of the headset.

Hmm, you're right, there's no audio-headset icon in breeze, it's inherited from oxygen. Should we use it while we request an audio-headset icon for breeze? or should I revert that change and use headphones in the meantime?

• We have a headset icon, so use it for headsets

It seems other parts of kdepim already really require KF5 5.28, so there's no point in requiring less here

• Change back REQUIRED_KF5_VERSION to KF5_VERSION

• Set the required KF5 version (5.26.0) with a variable

• Set the ECM dependency with a variable

In D3661#68474, @vkrause wrote:

The change is fine with me (and 5.26 indeed still works), the problem I see is that the version bumps here happen automatically by the KF5 release scripts AFAIK, ie. this will be a very short-lived change.

In D3317#61947, @dfaure wrote:

Did you test this with kdesu? According to the comment, the whole point here is to save a file that will be used by kdesu, i.e. as another user. Since RuntimeLocation is normally per-user, that other user might not have access to the file
(well ok, root would, but I suppose kdesu also allows logging in as another user?)

In D3317#61881, @aacid wrote:

I really know nothing about this, my only involvement with this change was settiing it's umask "properly-ish".

• Moved testfindpreferredimage to an autotests directory

@hpereiradacosta thanks for reviewing this. Can you cherry-pick the commit for Plasma/5.8?

• Increase width by MenuButton_IndicatorWidth+2 instead of adding a new enum entry

• Adjust the distance method to work with QUAD HD resolutions

Converted the test to use QTest

• Converted the test to use QTest

I don't know why, but it seems arc diff doesn't work anymore and I have to use arc diff HEAD~7 to include all 7 commits

Include all 7 commits

All issues done

• Only call PostLastGuardedFrame when OpenGLCompositing is set

Include all 6 commits

• Use only one timer to detect freezes and other optimizations
• Test 3 frames for freezes by default

Btw, I tested the last commit on a virtual machine simulating the freeze with a sleep. I'll check tomorrow what's the minimum value for m_maxFramesTestedForSafety that works to detect the real freeze and update its default value.

• Remove most overhead caused by the freeze-detection thread

How many frames do we need to detect the freeze? I think 30 frames is too much by default. It's adding quite some stress on startup (lots of write to config, lots of additional threads started). Assuming we freeze with tripple buffer when trying to get the next buffer 4 frames should be sufficient to trigger the freeze?

I tested this patch but it didn't work at detecting the freeze. Fortunately, after debugging kwin for a bit I worked on this and made a version that detects the freezes correctly.

Oops, sorry, I wrote a reply yesterday but forgot to submit it :/

• Add QUrl::AssumeLocalFile and assume Qt > 5.4.0

• Move logic to show/hide avatars from components/SessionManagementScreen.qml to Main.qml

Thanks for your comments. I should warn you that this is the second time I touch qml code, so apologies in advance :).