However, I think that the choice of a data structure should be driven more by an overall programming style – if you do full OOP, prefer a class. If you do procedural/functional/scripting style programming, simpler data structures might be better indeed. To choice is up to developer and there is no silver bullet.

Also, using a dictionary instead of a class might lead to anemic domain model.

However, in Python I’d use a dictionary to represent the People object instead of a class. Classes really should only be used for complex state management where encapsulation is absolutely necessary.

I think this needs to be emphasized to Java developers coming to Python.

def new_person(firstname=”", lastname=”"): return {„firstname”:firstname, „lastname”:lastname}

persons = [new_person("Jake", "Blake"), new_person("James", "Smith")]

firstnames = [person["firstname"] for person in persons]

The last data structure of choice should always be classes. If you can get away using dictionaries, list and tuples, then use them first.

For example:
x = (i**2 for i in xrange(0,5)) – it won’ t produce a list(x[3] will produce a TypeError) and store it in memory but you can loop over x and sum it’ s elements and every element will be generated while beeing iterated over:
y = sum(x)
print y
>> 30 #because sum(x) == 0^2 + 1^2 + 2^2 + 3^2 + 4^2

And the more info is here: http://wiki.python.org/moin/Generators

It could be that:
„%s %s” % (self.fname, self.lname)

Or
„{fname} {lname}”.format(fname=self.fname,lname=self.lname)