[Numpy-discussion] isfortran compatibility in numpy 1.10.

[Charles R Harris]

Hi All,

The isfortran function calls a.fnc (Fortran-Not-C), which is implemented
as F_CONTIGUOUS && !C_CONTIGUOUS. Before relaxed stride checking
contiguous multidimensional arrays could not be both and continguous 1-D
arrays were always CONTIGUOUS, but this is not longer the case.
Consequently current isfortran breaks backward compatiblity. There are two
suggested solutions

1. Return `a.flags.f_contiguous`. This differs for 1-D arrays, but is
most consistent with the name isfortran.
2. Return `a.flags.f_contiguous and a.ndim > 1`, which would be backward
compatible.

It is also possible to start with 2. but add a FutureWarning and later move
to 1, which it my preferred solution. See gh-6590
<https://github.com/numpy/numpy/issues/6590> for the issue.

Thoughts?

[Travis Oliphant]

As I posted to the github issue, I support #2 as it is the original
meaning. The most common case of isfortran that I recall was to support
transpositions that needed to occur before calling Fortran-compiled linear
algebra routines.

However, with that said, you could also reasonably do #1 and likely have no
real problem --- because transposing a 1-d array doesn't have any effect.

In NumPy 1.0.1, isfortran was intended to be True only for arrays with
a.ndim > 1. Thus, it would have been possible for someone to rely on that
invariant for some other reason.

With relaxed stride checking, this invariant changed because isfortran was
implemented by returning True if the F_Contiguous flag was set but the
C_Contiguous flag was not (this was only ever previously possible for
a.ndim > 1).

If you choose to go with #1, please emphasize in the release notes that
isfortran now does not assume a.ndim > 1 but is simply short-hand for
a.flags.f_contiguous.

-Travis

Post by Charles R Harris
Hi All,
The isfortran function calls a.fnc (Fortran-Not-C), which is implemented
as F_CONTIGUOUS && !C_CONTIGUOUS. Before relaxed stride checking
contiguous multidimensional arrays could not be both and continguous 1-D
arrays were always CONTIGUOUS, but this is not longer the case.
Consequently current isfortran breaks backward compatiblity. There are two
suggested solutions
1. Return `a.flags.f_contiguous`. This differs for 1-D arrays, but is
most consistent with the name isfortran.
2. Return `a.flags.f_contiguous and a.ndim > 1`, which would be
backward compatible.
It is also possible to start with 2. but add a FutureWarning and later
move to 1, which it my preferred solution. See gh-6590
<https://github.com/numpy/numpy/issues/6590> for the issue.
Thoughts?
_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion

[Sturla Molden]

Post by Charles R Harris
1. Return `a.flags.f_contiguous`. This differs for 1-D arrays, but is
most consistent with the name isfortran.

If the idea is to determine if an array can safely be passed to Fortran,
this is the correct one.

Post by Charles R Harris
2. Return `a.flags.f_contiguous and a.ndim > 1`, which would be backward
compatible.

This one is just wrong.

A compromize might be to raise an exception in the case of a.ndim<2.

Sturla

[Sebastian Berg]

I bet it has all been said already, but to note just in case. In numpy itself we use it mostly to determine the memory order of the *output* and not for safty purpose. That is the macro of course and I think yelling people to use flags.fnc in python is better.

- Sebastian

Post by Sturla Molden

Post by Charles R Harris
1. Return `a.flags.f_contiguous`. This differs for 1-D arrays, but is
most consistent with the name isfortran.

If the idea is to determine if an array can safely be passed to Fortran,
this is the correct one.

Post by Charles R Harris
2. Return `a.flags.f_contiguous and a.ndim > 1`, which would be backward
compatible.

This one is just wrong.
A compromize might be to raise an exception in the case of a.ndim<2.
Sturla
_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion

[Charles R Harris]

Post by Sebastian Berg
I bet it has all been said already, but to note just in case. In numpy
itself we use it mostly to determine the memory order of the *output* and
not for safty purpose. That is the macro of course and I think yelling
people to use flags.fnc in python is better.

Probably all the Numpy uses of `PyArray_ISFORTRAN` should be audited. My
guess is that it will be found to be incorrect in some (most?) of the
places.

Chuck