[Numpy-discussion] New behavior for correlate w.r.t. swapped input: deprecate current default for 1.4.0 ?

[Charles R Harris]

On Fri, Nov 27, 2009 at 3:04 AM, David Cournapeau <
david at ar.media.kyoto-u.ac.jp> wrote:

> Hi,
>
>    The function correlate can now implement what is generally accepted
> as the definition of correlation, but you need to request this behavior
> ATM (by default, it still does the wrong thing). Is it ok to deprecate
> the default ? I.e. the behavior is exactly the same, but we warn users
> about the change, and in 1.5.0, the default is to use the new behavior.
>
>
What exactly are the differences? IIRC, there is a change for complex
arguements and what looks like a bugfix for real arguments. Is there a way
to warn *only* when the behavior will be different?

Chuck
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/numpy-discussion/attachments/20091127/58e01fb0/attachment.html>