Re: [ctypes-users] My ctypes interface fails on OSX 10.8 (Mountain Lion) but works on Red Hat Linux 6

[Diez R. <de...@we...>]

> 
> I've been working on a ctypes interface to udunits2
> (http://www.unidata.ucar.edu/software/udunits/udunits-2/udunits2.html) and
> have hit a strange issue when testing on OSX 10.8 (Mountain Lion).
> 
> I have built a udunits2.so (on Linux) and a udunits2.dylib (on OSx) which I
> can interface to with the attached c code (ud.c) from both the command line
> and via ctypes as per "example1.py". This works great across both Linux and
> OSX.
> 
> However, the point of me using ctypes is to avoid the need for writing the
> wrapping C code, so attempting to replicate "ud.c" with python ctypes code in
> "example2.py" works great on Linux, but results in a udunits failure on OSX.
> 
> I'm not sure what I'm doing wrong, the fact that this code works beautifully
> from C means that I am suspicious about my ctypes code rather than the udunits
> library itself is at fault, but I can't understand why my ctypes code would
> work on one OS and not on the other.
> 
> Should you need it to understand my C code the udunits2 C docs can be found
> here: http://www.unidata.ucar.edu/software/udunits/udunits-2/udunits2lib.html
> 
> I'm using python 2.7.3, udunits 2.1.24 and the builtin ctypes python module on
> both Red Hat 6 and OSX 10.8.
> 
> Does anyone have any idea what I'm doing wrong?


I don't see any "real" ctypes usage here. All I see is loading a lib, and
then using it as if all arguments were integers. Which looks suspicious to
me, and might cause issues because of e.g. Word-size differences under
various OSes. 

So you should try & proper declare function parameters and result  types,
and then try again.

Diez