Use cases for del

[Ron Adam]

Scott David Daniels wrote:
> Ron Adam wrote:
> 
>> George Sakkis wrote:
>>
>>> I get:
>>>
>>> None: 0.549999952316
>>> String: 0.498000144958
>>> is None: 0.450000047684
>>
>>
>>
>> What do yo get for "name is 'string'" expressions?
> 
> 
>     >>> 'abc' is 'abcd'[:3]
>     False

Well of course it will be false... your testing two different strings! 
And the resulting slice creates a third.

Try:

ABC = 'abc'

value = ABC
if value is ABC:   # Test if it is the same object
    pass

In the previous discussion I was comparing using a string as an 
alternative to using None as a "flag" object.  Not as a value to be 
calculated.

And just to be clear, I'm not disagreeing with you.

Yes, you can have problems with string comparisons if you create a copy 
instead of pointing a name to the object like you did above.  Something 
to be aware of.

To avoid that you either need to define the flag string as a global name 
or use it strictly in the local scope it's defined in.  Python will also 
sometimes reuse strings as an optimization instead of creating a second 
string if they are equal.  Something else to be aware of.

So I'm not suggesting it is a good idea to use strings in place of None. 
   But I still wonder why bool and other object comparisons are slightly 
slower than string comparisons. (?)

Cheers,
Ron


> You need to test for equality (==), not identity (is) when
> equal things may be distinct.  This is true for floats, strings,
> and most things which are not identity-based (None, basic classes).
> This is also true for longs and most ints (an optimization that makes
> "small" ints use a single identity can lead you to a mistaken belief
> that equal integers are always identical.
> 
>     >>> (12345 + 45678) is (12345 + 45678)
>     False
> 
> 'is' tests for identity match.  "a is b" is roughly equivalent to
> "id(a) == id(b)".  In fact an optimization inside string comparisons
> is the C equivalent of "return (id(a) == id(b) of len(a) == len(b)
> and <elements match>)
> 
> --Scott David Daniels
> Scott.Daniels at Acm.Org