Chris Colbert wrote:
>
> Quick note: By linker bug I really meant a bug in your make system,
> not in
> the linker itself :-) But I guess that's how you interpreted it.
>
> Dag Sverre
>
> _______________________________________________
>
>
>
> Here is a pure cython example that exhibits the same behavior. The
> cimports are simply .pxd files of header prototypes.
>
>
> #---------- purecython.pyx -----------------#
> cimport c_cxcore
> cimport c_highgui
> cimport c_cv
> cimport c_constants
>
> cdef tester():
> cdef c_cxcore.IplImage* img1
> cdef c_cxcore.IplImage* img2
> img1 = c_highgui.cvLoadImage("2.bmp", 1)
> img2 = c_cxcore.cvCreateImage(c_cxcore.cvSize(640, 480), 8, 3)
> c_cv.cvResize(img1, img2, c_constants.CV_INTER_AREA)
> c_highgui.cvNamedWindow("test", 1)
> c_highgui.cvShowImage("test", img2)
> c_highgui.cvWaitKey(0)
>
> def test():
> tester()
Well, you never free the memory for img1 and img2, but that shouldn't
matter.
Anyway, just to be clear, you are saying that this crashes, but the
exact same C code, that is
IplImage* img1;
IplImage* img2;
img1 = cvLoadImage("2.bmp", 1);
... and so on
does not crash?
(If the pure C code ever crashes you must bring this to the OpenCV
forums instead.)
In that case, do manual inspection of the Cython-generated C source --
is there any difference at all (up to variable names etc.) between
Cython-generated and manual C code?
--
Dag Sverre
_______________________________________________
Cython-dev mailing list
[email protected]
http://codespeak.net/mailman/listinfo/cython-dev
