Border fix

[akuchibotla]

Addressing issue #1669 and looking for feedback

In particular, I made sure there were no NaNs in the swirl mapping as well as no numbers less than 0. The arithmetic seemed to work incorrectly when radius was 0 (DivideByZero) but I feel as though my patch for that is still a little hacky. Is there a Numpy number abstraction that works as a "very small number"?

[stefanv]

How about np.isclose?

[stefanv]

np.finfo(float).eps

[stefanv]

Let's also add a test to ensure that this remains fixed in the future.

[akuchibotla]

@stefanv I'm confused about your comment to use np.isclose rather than the equality check I'm using. This error seems to occur only when the radius is set to exactly 0 so I'm not sure why checking if the radius is close to 0 would be helpful here.

[stefanv]

Ah, I thought some other small numbers would also trigger the error. Have
you checked even for numbers as small as 1e-15? If this are fine then the
exactly zero check is fine.
On Feb 19, 2016 10:54 PM, "Anand Kuchibotla" notifications@github.com
wrote:

@stefanv https://github.com/stefanv I'm confused about your comment to
use np.isclose rather than the equality check I'm using. This error seems
to occur only when the radius is set to exactly 0 so I'm not sure why
checking if the radius is close to 0 would be helpful here.

—
Reply to this email directly or view it on GitHub
#1952 (comment)
.

[emmanuelle]

@akuchibotla can you also add a test to check that the issue of #1669 is solved? Thanks!

[akuchibotla]

Not 100% sure if this is the kind of test you're looking for

[soupault]

np.allclose?

[soupault]

@akuchibotla @emmanuelle What do you think about this? :

1. Use checkerboard as in the original issue;
2. Swirl the image;
3. Calcualte histogram / sum of pixels over ~2-pix wide frame;
4. Compare with reference value.