According to the documentation, random.shuffle() should accept a sequence. But it also accepts dicts, in which case it does nothing, expensively:

py> x = dict.fromkeys(range(10**6))
py> random.shuffle(x)
py> str(x)[:55] + "...}"
'{0: None, 1: None, 2: None, 3: None, 4: None, 5: None, ...}'


I'm not sure if it is better to explicitly test for and reject dicts and mappings, or just document that sorting dicts will apparently succeed while doing nothing.

Do we want to support sorting OrderedDicts? That at least makes sense, since they have an order, but again shuffle() seemingly works but actually does nothing.

I lean towards explicitly testing for dicts and raising.

It is also pointless to shuffle a list object repeating the same item, but that is no reason to add a special case. Is there a real problem with allowing dictionaries and OrderedDict? It seems to behave sensibly if you give each item a unique value:

>>> D = {i: i * 100 for i in range(5)}
>>> D
{0: 0, 1: 100, 2: 200, 3: 300, 4: 400}
>>> shuffle(D)
>>> D
{0: 300, 1: 100, 2: 200, 3: 0, 4: 400}

Generally speaking, trying to shuffle a dict D already blows up, unless D's keys are the integers range(len(D)).  In that case, D is indistinguishable from a list in the sense that both map range(N) to values via __getitem__.  `shuffle()` does no type checks whatsoever now, and will shuffle any object that maps range(N) to values.

Fine by me the way it is.

I concur with Tim.  These are just duck-typing facts of life.