[SciPy-dev] Some concerns on Scipy development

[pearu at scipy.org]

Hi,

On Tue, 26 Mar 2002, Travis Oliphant wrote:

> The biggest problems was with linalg, and it was related to the fact that
> Pearu changed f2py for the better, but did so withouth changing the
> interface to linalg (who can blame him we all do this in our "spare"
> time).

I was very much aware that my changes will break any interface that
contains multi-dimensional arrays. Normally one would expect some
transition period when a program introduces an incompatibility
feature. I could not afford that for f2py because of my time limitations
now and in future (recall that f2py is still an one-man project run
without any external support).

The issues that the new f2py solves now were just too important
that I suppressed any hesitations about backward compatibility
and fixed the long lasted confusion with the differences in how data are
stored in multi-dimensional arrays in FORTRAN and Numeric Python.

I cannot take any blame on not to fixing the interface to linalg (that I
actually did in linalg2, what else would you call it?) or any
other interface. It is the job of the interface writers, after all, the
new features was introduced to ease their jobs and to avoid constant
queries about data storage issues from them to me and from the users to
them and me. And in case you have not noticed, then f2py as an LGPL
software comes with "NO WARRANTY IS EXPRESSED OR IMPLIED.  USE AT YOUR OWN
RISK."

Regards,
	Pearu