[Tutor] TypeError: dict objects are unhashable

[A.T.Hofkamp]

Lie Ryan wrote:
> Emile van Sebille wrote:
>> Mark Tolonen wrote:
>>>>>> [dict(n) for n in set(tuple(n.items()) for n in l1)]
>>
>> Anyone know if the ordered items for two different dicts where dicta 
>> == dictb is guaranteed the same?  I know the ordering is unspecified 
>> but you can depend on the sequence of keys matching data and I think 
>> it'll also match items.
>>
>> but for all DA's and DB's below, is that true?
>>
> 
> Considering how dictionary is implemented, I don't think so. The 
> dictionary is internally implemented as a hash table, and two dictionary 
> with identical items may have different sized hash table, the algorithm 
> guarantees efficient usage of memory (it resizes when it's too small or 
> too big). On a resize, the hashes are recalculated and the ordering 
> rescrambled.

And a short demo to confirm:

Python 2.3.4 (#1, Jul 25 2008, 14:24:21)
[GCC 3.4.6 20060404 (Red Hat 3.4.6-10)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
 >>> x = range(10)
 >>> y = x[:]
 >>> y.reverse()
 >>> d1 = dict([(i*1000, i) for i in x])
 >>> d2 = dict([(i*1000, i) for i in y])
 >>> d1.items()
[(0, 0), (4000, 4), (1000, 1), (8000, 8), (6000, 6), (9000, 9), (2000, 2), 
(5000, 5), (7000, 7), (3000, 3)]
 >>> d2.items()
[(8000, 8), (4000, 4), (2000, 2), (0, 0), (9000, 9), (6000, 6), (5000, 5), 
(3000, 3), (1000, 1), (7000, 7)]


Sincerely,
Albert