[Tutor] When are strings interned?

[Lie Ryan]

Angus Rodgers wrote:
> Hello, world!
> 
> This is my first post to the Tutor list (although I've already
> posted to comp.lang.python a couple of times).
> 
> I'm currently reading Chapter 4 of Wesley Chun's book, "Core 
> Python Programming" (2nd ed.).
> 
> I find this, in Python 2.5.4, on my Win98SE system (using IDLE):
> 
>>>> n = "colourless"
>>>> o = "colourless"
>>>> n == o
> True
>>>> n is o
> True
>>>> p = "green ideas"
>>>> q = "green ideas"
>>>> p == q
> True
>>>> p is q
> False
> 
> Why the difference?

In this particular case, the reason is because "colourless" can be used
as an identifier. Now, before I continue, I need to warn, never to rely
on any sort of interning behavior; as they are strictly implementation
detail, implementation specific, and is NOT the behavior of Python (but
only the behavior of CPython).

Identifier are strings that can be used for names (variables). In python
most name lookups are actually a dict lookup (except for locals).
Strings that can be used as identifier are interned to speed up these
lookups. Identifiers cannot contain spaces, that's why green ideas are
not interned.

And... have I told you not to rely on this behavior? NEVER rely on this
implementation details. Once you get bitten by it, you'll regret it.