Equal sets with unequal print and str() representations

[Ganesh Gopalakrishnan]

Thanks to all who replied - also to Ben. I had foolishly assumed that 
the same set exhibits the same rep on at least one platform. Like any 
bug, the falsity of my assumption took months to expose - till then, 
things had worked fine. Needless to say I'm new to Python.  (The double 
printing is because I tend to work within an Emacs inferior shell.)

Cheers,

Ganesh

On 10/16/11 8:23 PM, Ben Finney wrote:
> Ganesh Gopalakrishnan<ganesh at cs.utah.edu>  writes:
>
>> This probably is known, but a potential pitfall (was, for me) nevertheless.
>> I suspect it is due to hash collisions between 's3' and 's13' in this
>> case?
> What is the actual problem? What behaviour is occurring that doesn't
> match your expectation?
>
>>>>> S1==S2
>> S1==S2
>> True
>>>>> str(S1)
>> str(S1)
>> "{'s8', 's13', 's2', 's0', 's7', 's6', 's4', 's3', 's14'}"
>>>>> str(S2)
>> str(S2)
>> "{'s8', 's3', 's2', 's0', 's7', 's6', 's4', 's13', 's14'}"
>>>>> str(S1) == str(S2)
>> False
> Right, that's all correct (though I don't know why some of your
> expressions are being shown twice). A deliberate property of a set is
> that its items are unordered.
>      <URL:http://docs.python.org/library/stdtypes.html#set>
>
> Since sets are unordered, the string representation may show the items
> in an arbitrary and unpredictable sequence. Don't write any code that
> depends on a predictable sequence of retrieval from an unordered
> collection.
>
> So what did you expect instead, and what supports that expectation?
>