# This is bad way of definition default argument. Lets see what will happens # if we will use this in real code. def bad_func(element, aggregator=[]): aggregator.append(element) return aggregator # Here's we're invoking our bad_func a few times. br1 = bad_func(1) print "br1:", br1 br2 = bad_func(2) print "br2:", br2 non_default_aggregator = ["a", "b", "c"] br3 = bad_func(4, non_default_aggregator) print "br3:", br3 br4 = bad_func("new element") print "br4:", br4 # As you can see the result behave in different way we've expect, so what is # actually happens? We shall investigate this! print "br1: {0}, br2: {1}, br4: {2}".format(id(br1), id(br2), id(br4)) print "br1 == br2 == br4:", id(br1) == id(br2) == id(br4) print "br3: {0}, nda: {1}".format(id(br3), id(non_default_aggregator)) print "br3 == nda:", id(br3) == id(non_default_aggregator) # So, as we can see three of these outputs referres to the same object. How to # fix this situation and enforce function to behave correctly? print # This is the way it goes. And it's a good practice. def good_func(element, aggregator=None): if aggregator is None: aggregator = [] aggregator.append(element) return aggregator # Here's we're invoking our good_func a few times. br1 = good_func(1) print "br1:", br1 br2 = good_func(2) print "br2:", br2 non_default_aggregator = ["a", "b", "c"] br3 = good_func(4, non_default_aggregator) print "br3:", br3 br4 = good_func("new element") print "br4:", br4 # Nice. Now we have what we wanted. print "br1: {0}, br2: {1}, br4: {2}".format(id(br1), id(br2), id(br4)) print "br1 == br2 == br4:", id(br1) == id(br2) == id(br4) print "br3: {0}, nda: {1}".format(id(br3), id(non_default_aggregator)) print "br3 == nda:", id(br3) == id(non_default_aggregator) # So as conclusion. We see that if object were created during function # definition than default value will keep referring on it on each invokation.