[Edu-sig] generic objects (including lambdas) as dict keys?: not usually a problem

[kirby urner]

I just discovered for myself what many must already know: lambda
expressions may serve as dict keys (I knew values).  That means they may
partake of the mutability of some functions, in having mutable default
parameter values.


In [122]: d = {lambda x=[10]: print(x.append(11), x) : 3}

In [123]: list(d.keys())[0]()

None [10, 11]


In [124]: list(d.keys())[0]()

None [10, 11, 11]


In [125]: list(d.keys())[0]()

None [10, 11, 11, 11]


In [126]: list(d.keys())[0]()

None [10, 11, 11, 11, 11]

That's not going to corrupt the dict though.  In general, callable objects
may serve as keys, with lambda expressions a special case. So what if the
object points to mutable content.


In [127]: class Callable:
     ...: def __init__(self):
     ...:     self.the_list = []
     ...: def __call__(self, *args):
     ...:     self.the_list.extend(args)


In [128]: obj = Callable()

In [129]: the_dict = {obj : 3}

In [130]: list(the_dict.keys())[0]("troll farm")

In [131]: obj.__dict__

Out[131]: {'the_list': ['troll farm']}

On the other hand, if you define __eq__ vis-a-vis something mutable, then
the dict-maker will have 2nd thoughts about the viability of your
enterprise.

So nothing that earth-shaking.  It's the Python we all know.

Just thought I'd say hi.  More soon.

Kirby Urner
Portland, OR
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/edu-sig/attachments/20180320/0e4823f0/attachment.html>