• Add a manual test

Yeah I can't upload screenshots etc. through Arc so I always need to upload those afterwards.

• Fix qmldir file
• Also register meta type of MetricPrefix so it can actually be used
• Add a manual test to test how Formatter works from QML

• More minor corrections

• Correct comment

Sorry, I was waiting for it but apparently completely missed the original patch. Everything looks fine to me.

• Don't use multiple commands to achieve what one can also do

In D27505#614240, @davidedmundson wrote:

Would be good to dump this test qml somewhere in the repo. A folder full of tests which just have one component in a bunch of configurations.
I've been pushing for that in p-f, and it catches so many things.

It most likely does, since Icon also has no implicit width/height it will default to 0 and crash when there's no special layout properties specified.

Now pushed with correct author information. :) Thanks!

Nice catch.

@heikobecker This patch has been open for a long time... Do you have commit access or do you need someone to commit it for you?

@heikobecker This patch has been open for a long time... Do you have commit access or do you need someone to commit it for you?

@jakobkummerow Ugh, do you have an email address and full name for me? I thought they were in the commit but apparently not (or arc ate them). Then I can fix the authoring information of that commit.

Both use QtQuick Controls 1 ScrollView, which works quite a bit different if I recall correctly. So they are not really comparable.

That's even better, thanks. Do you have commit access?

I agree with what @cfeck says. I can also already think of at least one case where even a 256KiB buffer would be too small, which is a dual-cpu Epyc system, where you can potentially have 128 cores / 256 threads.

In T10470#219979, @ndavis wrote:

I believe @ahiemstra mentioned something about adding a way to put some kinds of actions in the widget header. It might be worth considering when that would be more desirable than putting all buttons at the bottom.

So maybe this isn't actually a destructive action at all. After all, an app with unsaved changes or active processed will block the logout/shutdown absent user interaction.