On Fri, Sep 19, 2014 at 8:59 PM, Steven D'Aprano <steve+comp.lang.pyt...@pearwood.info> wrote: > Here's a proof of concept of what would be allowed: > > import random > class MyDict: > def __init__(self, items): > self._items = list(dict(items).items()) > self._flags = [False, False, False] > def keys(self): > k = [item[0] for item in self._items] > self._check(0) > return k > def values(self): > k = [item[1] for item in self._items] > self._check(1) > return k > def items(self): > k = self._items[:] > self._check(2) > return k > def _check(self, i): > self._flags[i] = True > if self._flags == [True, True, True]: > random.shuffle(self._items) > self._flags = [False, False, False]
Also, this can't possibly offer the same guarantee. Watch: d = MyDict(some_lot_of_items) d.values(); d.items() # mutate the dict in whatever way you like pairs = zip(d.keys(), d.values()) This might well create mismatched pairs, because after generating the keys() return value, the list gets shuffled, prior to generating values() in the same expression. This would not be allowed. ChrisA -- https://mail.python.org/mailman/listinfo/python-list
