[Tutor] Expressions, literals, operator precedence

[Jim Byrnes]

On 03/23/2014 11:36 PM, Ben Finney wrote:
> Jim Byrnes <jf_byrnes at comcast.net> writes:
>
>> I am reading Practical Programming - An Introduction to Computer
>> Science Using Python 3.  They give this example:
>>
>>>>> abs(-3)
>> 3
>>
>>>>> -3 .__abs__()
>> 3
>
> That's a poor example, in my opinion. It's not good for an introductory
> text to show calling dunder methods like that on an integer literal.
>
> Perhaps you could communicate with the maintainer of that material, to
> point out the problem with their example. Hopefully they will remove the
> example from an introductory text.
>

To be fair to the authors this was in a section taking about how the 
double underscore was special to Python and showing a little about what 
was going on behind the scenes.  They stressed that programmers would 
almost never use them in code.  I plan on sending them a message 
pointing out the results they show are incorrect.

>> Python 3.3.5 (default, Mar 12 2014, 02:09:17)
>> [GCC 4.6.3] on linux
>>
>>>>> abs(-3)
>> 3
>>
>>>>> -3 .__abs__()
>> -3
>
> Yes, I get the same result as you.
>
> The reason is that the expression is being evaluated as::
>
>      -( (3) .__abs__() )
>
> That is:
>
> * Create the integer object 3
>
> * Get the result of that object's ‘__abs__’ method
>
> * Negate (invert the sign) of the value
>
> (If you really care, see the end of this message[0] for a demonstration
> that this is exactly what happens.)
>
>
> Presumably the expression should be different, as shown in your next
> example::
>
>> If I use a variable it works.
>>
>>>>> x = -3
>>>>> x.__abs__()
>> 3
>
> Yes, this is a better way to do it. But why are we calling the ‘__abs__’
> function directly at all? That is, after all, the point of the ‘abs’
> built-in function: to call the correct method on the object::

I am not really using it.  I was just trying to verify the result shown 
in an example.

>      >>> x = -3
>      >>> abs(x)
>      3
>
>> I am curious as to what is happening.  Is the book wrong?  I checked
>> it's errata and nothing is mentioned.
>
> I think that this is both an erratum, and a demonstration that the
> example is a bad idea for the book entirely. Can you contact the
> maintainer of that work to let them know?
>
>

Thanks for the confirmation and the explanation of what was happening, 
much appreciated.

Regards,  Jim