mpeaton / pyopencv

Automatically exported from code.google.com/p/pyopencv
0 stars 0 forks source link

size_t related compilation errors on certain 64bit platforms #3

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago
What steps will reproduce the problem?
Attempt to compile on certain 64-bit platforms will fail because of 
incompatible conversion from 
size_t* to unsigned int*. Tried on my both platform RHEL5 linux x86_64 and Mac 
OS X 10.5.8.

Related files:
pyopencv/pyopencvext/CvQuadEdge2D.pypp.cpp:32: error: invalid conversion from 
'CvSubdiv2DEdge*' to 'unsigned int*'
pyopencv/pyopencvext/MatND.pypp.cpp:56: error: invalid conversion from 
'size_t*' to 'unsigned 
int*'
pyopencv/pyopencvext/ndarray.cpp:117: error: invalid conversion from 
'npy_intp*' to 'const int*'
pyopencv/pyopencvext/ndarray.cpp:118: error: invalid conversion from 
'npy_intp*' to 'const int*'

Original issue reported on code.google.com by zhu....@gmail.com on 9 Feb 2010 at 6:17

GoogleCodeExporter commented 9 years ago
Thanks. That is expected. I would love to have PyOpenCV compilable on 64-bit
platforms, but at the moment that desirable feature that is beyond my reach. I 
will
try to address the issue when I have time and have a 64-bit platform to test 
with. :)

Cheers,
Minh-Tri

Original comment by pmtri80@gmail.com on 10 Feb 2010 at 1:23

GoogleCodeExporter commented 9 years ago
For the compilation part this is the only issue I came across. I altered all of 
these to size_t and it was able to 
compile. Haven't tested the runtime though. Will try it later. Thanks a lot for 
your efforts on bringing a great 
Python interface.

Btw, any idea what the so-called new official python interface is based upon?

Original comment by zhu....@gmail.com on 10 Feb 2010 at 3:41

GoogleCodeExporter commented 9 years ago
Oh really? That was surprising. I thought there were a lot more than just the 
size_t
issue. I wish I had a 64-bit platform to test with. 

As far as I know, the new official python interface is based on ctypes and 
OpenCV's C
interface. James, the author, wrote some scripts to automatically parse the 
function
list and generate corresponding wrapping code. I don't know how far that 
project has
gone though. I haven't checked it out myself.

Cheers,
Minh-Tri

Original comment by pmtri80@gmail.com on 10 Feb 2010 at 9:19

GoogleCodeExporter commented 9 years ago
Hi zhu.146,

I have been trying to make PyOpenCV compatible with 64-bit platforms for a 
while,
although I still have not owned a 64-bit platform. Recently, it has been 
successfully
built on 64-bit OpenSUSE (see issue 11) without making any modification. If you 
have
time, may I ask you to try to build PyOpenCV 2.1.0.wr1.0.2 on your 64-bit 
platform
and see if it works properly? Thanks.

Cheers,
Minh-Tri

Original comment by pmtri80@gmail.com on 4 May 2010 at 5:57

GoogleCodeExporter commented 9 years ago
Good news: The latest version, PyOpenCV 2.1.0.wr.1.1.0, is now compilable on a 
64-bit Ubuntu 10.4 platform and appears to work properly. Feel free to test it 
on your 64-bit platforms.

Cheers,
Minh-Tri

Original comment by pmtri80@gmail.com on 18 Jul 2010 at 10:17