Preparing to release bottleneck 1.2.1

[kwgoodman]

I am getting ready to release bottleneck 1.2.1. This release adds support for NumPy's relaxed strides checking and fixes a few bugs. The only thing left to do is testing.

If you guys (@cgohlke windows, @shoyer xarray and pandas, @toobaz debian) have time please test.

Everyone else, please send in test reports. Does bn.test() pass on your computer? Does bn 1.2.1 (the master branch) run with your code that uses bn?

[cgohlke]

All bottleneck tests pass on Windows with numpy+MKL-1.11.3.

[itdaniher]

All tests pass on Linux / AArch64 / Numpy 1.11.0 on Python3.5.2.

[kwgoodman]

Thanks to @cgohlke I added another bug fix #163 to this release.

@itdaniher and @cgohlke could you test again? And @shoyer do you know someone who can test this release against xarray and pandas?

[itdaniher]

169 (edit: all) tests passed on commit f01a64b / Linux / AArch64 / Numpy 1.11 / Python3.5.2.

[cgohlke]

All tests pass on Windows with numpy+mkl-1.11.3. This also fixes pandas-dev/pandas#15453.

[kwgoodman]

The recent bug fix #163 adds the L for long:

In [1]: bn.nanmin(1)
Out[1]: 1L
In [2]: bn.nansum(1)
Out[2]: 1L

Not sure what I think of that. I could make the bugfix windows only.

[cgohlke]

The recent bug fix #163 adds the L for long:

That is only on Python 2.7. It would be nice to return PyInt objects if possible. PyLong objects are needed on Windows and 32-bit Unix if the asum etc values are 64-bit and larger than LONG_MAX.

[kwgoodman]

OK, that's good to know. Well, the bug fix means that we are now at least returning the correct number. That's good enough for me for this release.

[toobaz]

I can confirm a debian package of f01a64b on amd64 debian testing passes all tests.

[kwgoodman]

Thanks, guys, for all the testing.

I am surprised we haven't heard from @shoyer (he requested this release). I'd like to get a developer of pandas and xarray to test the release. Would be nice if one of them would watch this (low traffic) repo.

[shoyer]

Sorry, I haven't had time to get to this. I did open up pandas and xarray issues, hopefully someone will step up.

[kwgoodman]

Yeah, I figured it was just that you were busy. Either that or your notifications were coming at you like a firehose. Or maybe both.

[shoyer]

xarray passes with bottleneck 1.2.1. @fmaussion added a test against bottleneck master to our Travis-CI configuration, so this should be basically trivial to verify in the future!

[kwgoodman]

Bottleneck 1.2.1 is tagged and released. We'll go without pandas testing for this bug fix release.

Note: I only test (appveyor and travis) with py 2.7, 3.5, 3.6 and numpy 1.12.1. So that is what is officially supported. A bug in numpy 1.12.0 will cause a unit test failure in bottleneck.

Thanks to all who tested.

[toobaz]

We'll go without pandas testing for this bug fix release.

I just ran the pandas test suite: all good.

[shoyer]

Thanks Pietro!

…

On Mon, May 15, 2017 at 7:29 PM Pietro Battiston ***@***.***> wrote: We'll go without pandas testing for this bug fix release. I just ran the pandas test suite: all good. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#168 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABKS1ngDE-rt8GjxKqTgNyGrtisXJNcrks5r6N_agaJpZM4NNLDI> .