kicad-developers team mailing list archive

[Dick Hollenbeck <dick@xxxxxxxxxxx>]

On 01/12/2013 11:15 AM, Wayne Stambaugh wrote:
> I've been attempting to resolve my Python scripting build issues on 
> MinGW32 on Windows and I have a few questions for the folks who wrote 
> the cmake build code.  Why are we not using cmake's FindPythonInterp() 
> instead of specifying the python executable on the path?  Specifying the 
> Python interpreter on the command line at build time seems unnecessary 
> to me.  CMake's FindPythonInterp supports defining a a specific version 
> of Python by defining Python_ADDITIONAL_VERSIONS.  I'm guessing we would 
> want to do that because python2 is currently defined as the default 
> Python interpreter so I'm also assuming that the scripting code 
> generated does not work with Python3.
>
> There is also a link bug on MinGW because rt is defined as one of the 
> link libraries in SWIG_LINK_LIBRARIES.  Actually, someone attempted to 
> fix it by adding and conditional statement to only add rt on Linux but 
> they forgot to take it out of the original list.


No, the sequence was the other way around.  The rt was added by me recently, and I
probably broke the windows DLL doing it, while fixing the linux DEBUG DLL.

So just removing that is evidently not going to be adequate.  Please test the Linux DEBUG
DSO build with scripting to verify it still works.

Linux + Debug + DSO

is the problem case.

The missing symbol without librt was the profiling helper, clock_gettime() and it was
getting pulled in because GetRunningMicroSecs() is evidently still in a Debug build, which
in and of itself is fine.
That should be allowed.

Thanks for fixing the Windows build.

Dick