Home Original page

Python Portability--Not very portable?

W. eWatson wolftracks at invalid.com
Thu Aug 5 23:28:40 EDT 2010 
On 8/5/2010 7:45 PM, geremy condra wrote:
> On Thu, Aug 5, 2010 at 6:50 PM, W. eWatson<wolftracks at invalid.com>  wrote:
>> In my on-again-off-again experience with Python for 18 months, portability
>> seems an issue.
>>
>> As an example, my inexperienced Python partner 30 miles away has gotten out
>> of step somehow. I think by installing a different version of numpy than I
>> use. I gave him a program we both use months ago, and he had no trouble. (We
>> both use IDLE on 2.5). I made a one character change to it and sent him the
>> new py file. He can't execute it. I doubt he has changed anything in the
>> intervening period.
>
> Portability doesn't mean you can use different versions of your
> dependencies and be A-OK. It should be fairly obvious that if the
> behavior of your dependencies changes, your code needs to change to
> ensure that it demonstrates the same behavior. Portability also
> doesn't mean that any given one-character change is valid, so that may
> be your issue as well.
>
>> A further example. Months ago I decided to see if I could compile a program
>> to avoid such problems as above. I planned to satisfy that need, and see if
>> I could distribute some simple programs to non-Python friends. I pretty well
>> understand the idea,and got it working with a small program. It seemed like
>> a lot of manual labor to do it.
>
> What, why were you compiling a program? And why not just use distutils?
>
> Geremy Condra

I checked the one char change on my system thoroughly. I looked around 
on some forums and NGs 4 months ago, and found no one even had a simple 
"compiled program" available to even demonstrate some simple example.

I would think there are some small time and big time Python players who 
sell executable versions of their programs for profit?

disutils. Sounds familiar. I'm pretty sure I was using Py2Exe, and 
disutils might have been part of it.

So how does one keep a non-Python user in lock step with my setup, so 
these problems don't arise? I don't even want to think about having him 
uninstall and re-install. :-) Although maybe he could do it without 
making matters worse.
More information about the Python-list mailing list