[Python-Dev] A proposal has surfaced on comp.lang.python to redefine "is"

[Andrew Koenig]

It's not my proposal, and I don't know if I'm for it or against it, but I
thought I'd mention it because I think it's interesting.

The observation is that if an object is immutable, there's no legitimate
reason to know whether it's distinct from another object with the same type
and value.  Therefore, the proposal is to change the definition of "is" as
follows:

	1) Unless x and y are both immutable, the meaning of "x is y"
	   remains as it has always been.

	2) If x and y have different types, "x is y" is false (as ever).

	3) If x and y are scalars, the definition of "x is y" is
	   changed to mean "x == y"

	4) Otherwise x and y are immutable collections of the same type,
	   in which case "x is y" is true if and only if every component
	   of x is the corresponding component of y (i.e. yields True when
	   compared using "is").

This change would clear up some seeming anomalies, such as:

	>>> 1000 is 1000
	True
	>>> x = 1000
	>>> y = 1000
	>>> x is y
	False
	>>> x = 42
	>>> y = 42
	>>> x is y
	True

The only disadvantage I can think of is that some comparisons might become
much slower (for example, using "is" to compare two long strings that share
a long common prefix).  However, I have a hard time understanding why anyone
would actually want to use "is" on immutable objects, and if one wants to
force a true object-identity comparison, one can always use id(x)==id(y).

So my question to the more experienced members of this group:  Is this idea
worth considering?  If not, why not?  (I'm partly expecting a response of
"It's not Pythonic," but I don't actually know why not).