range Failure?

[Bengt Richter]

On Sun, 10 Mar 2002 19:34:36 -0500, Peter Hansen <peter at engcorp.com> wrote:

>Bengt Richter wrote:
>> 
>> On Sun, 10 Mar 2002 11:07:18 -0500, "Colin J. Williams" <cjw at sympatico.ca> wrote:
>> [...]
>> >It might say:
>> >
>> >     range([start= 0,] stop[, step= 1])
>> 
>> Still, the two optional []'s are not valid in all combinations,
>> (which is what you fix with your improved text ;-)
>> so maybe it should say something like:
>> 
>>      range([start= 0,] stop)          # where step defaults to 1, or
>>      range(start, stop [, step= 1])
>
>Why not be very explicit, rather than fancy and confusing?  
>The above is an improvement over the original, but perhaps 
>this is better?
>
>  range(stop)                   # start is 0, step is 1
>  range(start, stop [, step=1])
>
Yes, that is better.

>Answering my own question: the first of the three is pretty bad.
>The second one suffers from implying that Python can have optional
>positional arguments which come before non-optional ones.  The 
>third suffers from implying (perhaps?) that Python can have 
>overloaded methods.
>
>Since range() doesn't take keyword arguments, even my second
>one above could be considered a little bit confusing, if you 
>think that implies you can say range(1, 6, step=2) which you
>cannot.
>
I guess it's a matter of defining the grammar of usage documentation.

Regards,
Bengt Richter