[SciPy-dev] Is it ok to depend on ctypes for scipy code ?

[David Cournapeau]

Robert Kern wrote:
>
> <shrug> Dynamic linking is inherently platform specific. Everyone does it
> differently. It just happens that Windows' choices gets in the way in this
> particular instance.
Well, details are platform specific of course, but at least solaris and 
linux are similar enough (I think there is no difference in System V and 
BSD anymore on recent unices), and OS X using gcc alleviates the dylib 
specificity. But complaining about it won't make it go away anyway, so 
just forget that I said that.
>
> I think you may be reading my statement out of context. Andrew asked about
> building shared libraries for ctypes' use by abusing distutils to build a fake
> extension module from it.
Mmh, do you mean building a library with  

config.add_extension('foo',                                                        

        sources=[join('src', 'vq.c') ])

is not enought to load it through ctypes on windows ? I certainly do not 
want to add any more burden for scipy distribution, as this is already a 
big problem. I will try something else, but it is a bit problematic for 
scipy.sandbox.svm, because I will have to redo all the wrapping, which I 
didn't plan,

cheers,

David