The "you may think of it as the object's address in memory" part is
misleading, and should be removed from the docs. While it's true that
you may think of it that way, such thinking just leads you to make
assumptions which are not universally true.
I agree with Marko that this is not a definition. It's a collection of
random statements about ids, their use, and some misleading philosophy.
Even the part about "... compares the identify of two objects" is kind
of funky, since it implies that you do indeed have two objects!
Correct, it should say "two operands" to be pedantic.
But in regular English, it is quite normal to say things like "the two
people actually turned out to be same person" (say, Superman and Clark
Kent) and I see no reason not to allow the similar, technically sloppy
but easily understandable idea that the `is` operator tests whether two
objects are in fact the same object.
This concept isn't really that hard to understand. In the expression `a
is b`, you have an object bound to the name "a", an object bound to the
name "b", hence *two objects*, and you want to know if they are the same
object or not.
A definition would be:
"The identity of an object is an integer which never changes during the
lifetime of the object, and which is guaranteed to be distinct from the
identities of all other objects with overlapping lifetimes. A given
identity may be reused for objects with disjoint lifetimes".
You then go on to say:
"The id() function returns the identity of an object. ..."
This definition is wrong. An object's identity and its ID are not the
same. Objects in Python have an identity the instant they are created,
but they may not have an ID assigned to them until much later, if at all.
Both IronPython and Jython lazily assign IDs on request, not on creation.
Here's an example from Jython:
2
IronPython is similar, although the specific IDs may not be the same.
The problem here is that you are wrongly identifying an object's identity
with its identification number. That doesn't apply to people (if you are
American, your identity is not the same as your Social Security number);
it doesn't apply to rocks, or pencils, or kitchen spoons. Why should it
apply to Python objects? The identity of an object is an inherent,
fundamental aspect of the object's existence, not some label stuck to it.
In Jython, the `is` operator can distinguish objects without assigning
them an ID number:
8
In plain English, "identity" has various definitions, but the one needed
here is this:
The state or quality of being identical, or the same; sameness.
[Webster 1913]
where "identical" is understood in the strong sense of:
being the exact same one; not any other
[Wordnet 2006]
rather than the weak sense of merely looking the same, as in "identical
houses". These are common, ordinary words, and no more need specialist
definitions in the Python documentation than do the other common,
ordinary words used, such as "never", "changes" or "created".
An object's identity is that quality which distinguishes it from every
other object. The nature of that quality is not important. Not
withstanding such interesting but irrelevant philosophical questions as
the Paradox Of My Grandfather's Axe, the intuitive, common, plain-English
meaning of identity is all we need to understand object identity, because
it's the same meaning.
