@kfunk
Hmm, it seems kdevelop tries to popup an error or something for lldb tests. Maybe due to unsupported lldb version?

Get rid of "property depends on non-NOTIFYable property" warning, as well as unneeded code.

Address comments. Seems I can update the diff directly.

The patch implements what I mentioned.

For reason I don't understand, onPositionChanged doesn't fire for me even with hoverEnable set to true. So the last point won't work. But otherwise it works well.

Sorry I should have been more specific.

what about set moved variable according to global mousre position instead? I know this is possible with QCursor, but not sure if it's possible at qml side, though. I haven't looked at the code yet, but if there's c++ component, we can add a qobject derived class calling QCursor::pos(), and expose it to qml.

Yeah I still don't understand the overall design well. Here's my understanding, please correct me if I'm wrong

CUDA is compatible with both C and C++. But clang only accepts c++11 etc for -std. For example, with c11, clang says

error: invalid argument '-std=c11' not allowed with 'CUDA'

• Make the cuda config dialog less confusing
• Add comments explaining how cuda auto detection works

I mistakenly set wrong include paths for Clang 3.9 and 4.0. That's why I was getting so many errors. After fix that up, everything works as expected even with -nostdinc flags.

Sorry I misread the last comment. No it's not affected by nostdinc or nostdinc++, as in current code those are not added to the command line.

Yeah there're nocudainc and nocudalib. But without cuda includes, some cuda structures like __device__ and __global__ will be undefined.

A bit more investigation shows that both clang and nvcc implicitly add many headers when compiling the file. And our parser doesn't do that, thus the error.

Hmm, I can compile fine with nvcc from CUDA 8.0, despite a few warnings about uninitialized variables.

Rebase on master

By the way, is it ok to land to 5.1 branch? Or maybe master is better?

I name it after "wild pointer" because both of them are dangling around :P

Simplify a bit

It's now ready for review.

Added and fixed remaining test cases, and rebased to latest HEAD.

Rebased on 5.1

@kfunk yes I think so, it's just bug fixes, and for a better first time impression people get about the LLDB plugin with the 5.1 release. I'll rebase.

Use a different way to implement this.
The first version is flawed in that it'll overwrite user preference,
and it doesn't handle missing GDB plugin.