David Faure wrote:
> I think you're assuming that we were #including qtestmouse.h directly,
> but we were not. It turns out the tests were including QTestEvent,
> and that's where the QT_GUI_LIB check is now,

Apparently kdebase does include qtestmouse.h directly. So I reverted the 
change to that CMakeLists.txt.

KDevelop doesn't. I'll let apaku or someone else decide how they want to 
handle that, if they want to make changes. (Before committing a revert 
please test against 4.6; I'd be happy to do that test if requested.) 
Though personally I'm inclined to prefer -DQT_GUI_LIB over including 
quasi-public (non-CamelCase '.h') Qt headers.

> so the change is still SIC.
> Which I can accept since it's needed for other reasons, and doesn't happen
> with qmake, and there is a solution, including <QTestGui>.

...and which is why I said the change is "unfortunate" (it is), not 
"something that needs to be reverted" (only if someone thinks of a way 
to solve the original problem that isn't SIC, and from the comment, this 
seems unlikely).

-- 
Matthew
Please do not quote my e-mail address unobfuscated in message bodies.
-- 
I don't see what C++ has to do with keeping people from shooting 
themselves in the foot.  C++ will happily load the gun, offer you a 
drink to steady your nerves, and help you aim. -- Peter da Silva (from 
cluefire.net)