[SciPy-Dev] Ticket #995, All scipy.constants.constants functions "choke" on non-array sequence input

[Skipper Seabold]

On Thu, Jun 17, 2010 at 4:10 PM, Vincent Davis <vincent at vincentdavis.net> wrote:
> On Thu, Jun 17, 2010 at 1:58 PM, Skipper Seabold <jsseabold at gmail.com> wrote:
>> On Thu, Jun 17, 2010 at 3:52 PM, Vincent Davis <vincent at vincentdavis.net> wrote:
>>>
>>>
>>> Sent from my iPhone
>>>
>>> On Jun 17, 2010, at 1:24 PM, Skipper Seabold <jsseabold at gmail.com>
>>> wrote:
>>>
>>>> On Thu, Jun 17, 2010 at 3:23 PM, Vincent Davis <vincent at vincentdavis.net
>>>> > wrote:
>>>>> On Thu, Jun 17, 2010 at 1:18 PM, Vincent Davis <vincent at vincentdavis.net
>>>>> > wrote:
>>>>>> The problem is that for example C2F is
>>>>>>
>>>>>> def C2F(C):
>>>>>>     return 1.8 * C + 32
>>>>>> So array_like (list, tuple...) does not work.
>>>>>>
>>>>>> What is the preferred solution?
>>>>>>
>>>>>> 1, Change the docs to say "scalar or array" and return a warning
>>>>>> if C
>>>>>> is not a scalar or array
>>>>>> 2, Ad the ability to handle all array_like.
>>>>>>
>>>>> def C2F(C):
>>>>>     return np.add(np.multilpy(1.8, C), 32)
>>>>>
>>>>> I think that accepts array-like and scaler and returns the same
>>>>
>>>> Or just C = np.asarray(C)
>>>>
>>>> This is quite often what is done.
>>> But if I start with a list I expect a list back.
>>>
>>
>> I don't know of any functions that preserve list inputs.
>>
>> In [3]: np.multiply([1,2,3,4,5],[2,3,4,5,6])
>> Out[3]: array([ 2,  6, 12, 20, 30])
>
> Your right, it preserves a scaler but not a list.
> Ok then would you prefer it accepts array_likem or scaler and returns
> a array or scaler?
>

You can just say accepts array_like and use asanyarray (point taken) I think.

Skipper