strides wrong for dimension = 1 objects with relaxed strides

[juliantaylor]

if compiled with NPY_RELAXED_STRIDES_CHECKING=1

>>> x =np.array ((1,))
>>> print x.strides
(9223372036854775807,)

it goes into thise code in numpy/core/src/multiarray/ctors.c:3739

#if NPY_RELAXED_STRIDES_CHECKING
            else {
                not_cf_contig = 0; 
            }
            if (dims[i] == 1) { 
                /* For testing purpose only */
                strides[i] = NPY_MAX_INTP;
            }
#endif /* NPY_RELAXED_STRIDES_CHECKING */

[njsmith]

What's the bug? Is it doing this when NPY_RELAXED_STRIDES_CHECKING is unset
or something?

http://docs.scipy.org/doc/numpy-dev/release.html#npy-relaxed-strides-checking

On Mon, Dec 2, 2013 at 11:44 AM, Julian Taylor notifications@github.comwrote:

x =np.array ((1,))
print x.strides
(9223372036854775807,)

it goes into thise code in numpy/core/src/multiarray/ctors.c:3739

#if NPY_RELAXED_STRIDES_CHECKING
else {
not_cf_contig = 0;
}
if (dims[i] == 1) {
/* For testing purpose only /
strides[i] = NPY_MAX_INTP;
}
#endif / NPY_RELAXED_STRIDES_CHECKING */

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4091
.

Nathaniel J. Smith
Postdoctoral researcher - Informatics - University of Edinburgh
http://vorpus.org

[juliantaylor]

it was reported on the mailinglist
http://mail.scipy.org/pipermail/numpy-discussion/2013-December/068392.html

without relaxed stride it returns (8,)
I realize now this was probably intentional, the code documentation could be improved.

[seberg]

Yeah, I did that intentionally to make stride bugs that could happen when RELAXED_STRIDES is used more obvious. Of course it should not stay for a release version...