or to rephrase zzag:

.right() is off by one
.setRight is off by one

so it cancels out.

Personally I find it so difficult to parse that it's easier to just boycott the method.
Everyone who comes across it and is aware of QRect's ridiculousness ends up having to do a load of additional checking in their heads anyway.

Though we don't have a kwin policy on that and we use QRect::right/bottom in other places, so I can't have a formal objection here.

I just approved another patch that used setRight setBottom in the same method, so we may as well be consistent.

In D27050#603781, @davidedmundson wrote:

I just approved another patch that used setRight setBottom in the same method, so we may as well be consistent.

Hmm, then these two https://github.com/KDE/kwin/blob/master/xdgshellclient.cpp#L1653-L1659 also have to be fixed.

I personally would prefer to continue using QRect::move{Right,Bottom}() and QRect::set{Right,Bottom}() as long as borders match, e.g.

rect.moveRight(anotherRect.right());

I just approved another patch that used setRight setBottom in the same method, so we may as well be consistent.

What I meant, was: you can ship this given we also use setRight(other.right()) elsewhere
I just can't press the button now that it's abandoned.

In D27050#603729, @davidedmundson wrote:

Though we don't have a kwin policy on that and we use QRect::right/bottom in other places, so I can't have a formal objection here.

Let's address this in D27060.