Adapting code to multiple platforms

[Jeffrey Barish]

I have a small program that I would like to run on multiple platforms
(at least linux and windows).  My program calls helper programs that
are different depending on the platform.  I think I figured out a way
to structure my program, but I'm wondering whether my solution is good
Python programming practice.

Most of my program lives in a class.  My plan is to have a superclass
that performs the generic functions and subclasses to define methods
specific to each platform.  I envision something like this:

class super:
    '''All the generic stuff goes here'''

class linux_subclass(super):
    def func(self):
        '''linux-specific function defined here'''

class windows_subclass(super):
    def func(self):
        '''windows-specific function defined here'''

And then in main I have:

inst = eval('%s_subclass' % sys.platform)(args)

to create an instance of the appropriate subclass.

I realize that I will have to name the subclasses appropriately
depending on what values get assigned to sys.platform. (On my platform,
sys.platform is 'linux2', so I would actually need class
linux2_subclass(super).  I don't know what value gets assigned on a
windows platform.)

The other way I thought of -- surrounding all the platform-specific code
with if statements to test sys.platform -- seems clearly worse.  I'm
just getting up to speed on Python and OOP, so I'm wondering whether I
have missed something obvious.  I'm hoping that the strongest rebuke
would be that I found something obvious.
-- 
Jeffrey Barish