Python Portability

[W. eWatson]

On 8/8/2010 4:08 AM, David Cournapeau wrote:
...
>> python-Numpy-1.2.0. No scipy anything. Well, this is interesting. I just
>> noticed Martin v. Loewis on the Python 2.5 entry. That's you, right?
>
> You are conflating so many issues at the same time, it is very
> difficult to follow what you are doing.
>
> Concerning the numpy error: you installed a version of scipy which
> requires a more recent version of numpy than the one you have. More
> concretely, NumpyTest has disappeared since 1.3.0. Unless you have a
> good reason not to, I strongly suggest to just use the last released
> versions of numpy and scipy (1.4.1 and 0.8.0 respectively).
>
> Note also that Enthought Python Distribution exists to exactly avoid
> those issues - they do the packaging hard word so that you don't have
> to.
>
> But none of this has anything to do with one character change or portability,
>
> cheers,
>
> David
Believe me I had no intent of expanding this thread beyond looking for a 
straight and simple insight to Python distribution (portability, 
whatever) and how to get my partner squared away. The general issue 
seems to drifted off on its own accord. I find generally that once I've 
had to post five times on a topic, and haven't gotten a satisfactory 
answer, I'm not going to get one.

Yes, you are quite correct about numpy, but for reasons which I have no 
plans to explain, I want my partner on the same version of numpy as 
myself. I now have the answer I need, and I do not care one more wit 
about a one character change.

I'm done here.