[Numpy-discussion] Re: Vote: complex64 vs complex128

[Perry Greenfield]

On Apr 4, 2006, at 4:40 PM, Travis Oliphant wrote:
>
> Totally agree.   I don't see the argument that Complex64 is a 
> "precision" description.  To a new user it could go either way 
> depending on their previous experience.  I think most new users won't 
> even use the bit width names but will instead use 'complex' and be 
> done with it...
>
>> I think it's time that we start taking backwards compatibility with 
>> previous
>> releases of numpy seriously and not break numpy code without clear, 
>> significant
>> gains in usability.
>>
> +1
>
The issue that just won't go away. We did it the current way for 
numarray initially and were persuaded to switch to be compatible with 
Numeric.

I agree that it isn't obvious what the number means for complex. That 
ambiguity will always be there. Unless we did a real user test to find 
out, we wouldn't know for sure what future users would most likely 
expect.

But in the end, pick one and let's not change it again (or even talk 
about changing it). It doesn't matter that much to me which it is.

Perry