[Numpy-discussion] promises, promises

Alan G Isaac aisaac at american.edu
Sun May 11 13:16:44 EDT 2008


>> I suspect part of the problem is that "backward 
>> compatability" is being interpreted in terms of discoverable 
>> behavior, rather than in terms of documented behavior. 

On Sun, 11 May 2008, Gael Varoquaux wrote:
> Not at all. It is to be interpreted in terms of "I have 
> a few dozens kilolines of code left by a student who work 
> with numpy 1.0, if I upgrade numpy, will they still 
> work?". 

That seems like the same thing to me.  What I mean is that 
part of the problem is that the handling of scalar indexing 
is being treated as part of the API contract simply because 
it was discoverable behavior.

To be specific: I do not recall any place in the NumPy Book 
where this behavior is promised.

When behavior is not promised, one feels fewer regrets in 
changing it.

Moving forward, I hope we will find a way to be more careful 
about making a minimum of promises.

Cheers,
Alan






More information about the NumPy-Discussion mailing list