Immutability of Floats, Ints and Strings in Python

[Ned Batchelder]

On Friday, November 25, 2016 at 6:34:00 AM UTC-5, Ned Batchelder wrote:
> On Friday, November 25, 2016 at 6:24:47 AM UTC-5, Nikunj wrote:
> > Hi All,
> > 
> > Out of curiosity, I wanted to understand the reason behind having different memory location for two identical floats . This is unlike ints or strings. Tried googling but couldn't find anything concrete. Any links or references would be appreciated!
> > 
> > Example:
> > 
> > For FLOATS:
> > ==========
> > 
> > >>> l = 1.3
> > >>> id(l)
> > 140421602788216
> > 
> > >>> k = 1.3
> > >>> id(k)
> > 140421602788240
> > 
> > >>> k == l
> > True
> > 
> > >>> k is l
> > False
> > 
> > For INTS and STRINGS:
> > =================
> > >>> i = 2
> > >>> o = 2
> > >>> id(i), id(o)
> > (140421602779712, 140421602779712)
> > 
> > >>> i is o
> > True
> > 
> > >>> a1 = 'hi'
> > >>> a2 = 'hi'
> > >>> a1 is a2
> > True
> 
> It's more complicated than that:
> 
>     (using Python 2.7.10)
>     >>> i = 2002
>     >>> o = 2002
>     >>> i is o
>     False
>     >>> a = 'what!?'
>     >>> b = 'what!?'
>     >>> a is b
>     False
> 
> A Python implementation can choose when to reuse immutable objects and
> when not to.  Reusing a value has a cost, because the values have to
> be kept, and then found again. So the cost is only paid when there's
> a reasonable chance that the values will actually be needed again.
> And that cost has to be weighed against the opposite cost of simply
> making a new object instead.

(Sorry, forgot an important point) The other reason to reuse objects is
so that equality testing can be shortened with an identity check. Because
strings are (sometimes) reused, CPython can test if two strings are equal
by first checking if they are precisely the same string object, and only
if they are not, by examining individual characters.

String comparison is an important operation that needs to be fast,
especially when you consider the use of strings as keys in dicts. So the
cost of reusing strings is easily worth it because of the time it saves
when doing string comparisons.

> 
> To answer your direct question: floats are so numerous and varied, it's
> reasonable to guess that you won't often need the same one.  And the cost
> to make a new one is so low, that it's simpler just to not reuse any of
> them.

And: floats are rarely checked for equality, and very very very rarely used
as dict keys, so there's no gain by short-circuiting the equality check.

--Ned.