REL: 2.3.1

[effigies]

Built a changelog, and think it's about time for a release. Any objections? Predominantly bug fixes, so I'm calling it 2.3.1 unless Python 3.7 support or nib-diff justifies a new minor version.

Open issues/PRs that could go in quickly

• RF+ENH: nib-diff - allow to specify absolute and/or relative maximal diff to tolerate #661 [PR] (Submitter: @yarikoptic; Reviewer: @chrispycheng)
• FIX: Interpret empty string in SliceThickness DICOM field as missing (default to 1) #652 [PR] (Submitter: @mattcieslak)
• ENH: Add SerializableImage class with to/from_bytes methods #644 [PR] (Submitter: @effigies; Reviewer: @matthew-brett)
• BUG?: MGHHeader.set_zooms() does not persist on I/O roundtrip or affect affine #619 [BUG] (Submitter: @larsoner)

Pre-release checklist

• Review the open list of nibabel issues. Check whether there are outstanding issues that can be closed, and whether there are any issues that should delay the release. Label them!
• Review and update the release notes. Review and update the Changelog file.
• Look at doc/source/index.rst and add any authors not yet acknowledged.
• Use the opportunity to update the .mailmap file if there are any duplicate authors listed from git shortlog -nse.
• Check the copyright year in doc/source/conf.py
• Refresh the README.rst text from the LONG_DESCRIPTION in info.py by running make refresh-readme.
• Check the dependencies listed in nibabel/info.py (e.g. NUMPY_MIN_VERSION) and in doc/source/installation.rst and in requirements.txt and .travis.yml. They should at least match. Do they still hold? Make sure nibabel on travis is testing the minimum dependencies specifically.
• Make sure all tests pass (from the nibabel root directory): nosetests --with-doctest nibabel
• Edit nibabel/info.py to set _version_extra to ''; commit
• Build bots
  • nibabel-py2.7-osx-10.10
  • nibabel-bdist32-27
  • nibabel-bdist32-34
  • nibabel-bdist32-35
  • nibabel-bdist64-27 [buildslave offline]

Adapted from http://nipy.org/nibabel/devel/make_release.html#release-checklist

[yarikoptic]

ok, now that no hypothesis etc, I will give another round of builds!

[yarikoptic]

oh... so the release would be coming from rel/2.3.1 which is diverged from master:

$> git diff origin/master...origin/rel/2.3.1 --stat
 .mailmap             |  1 +
 Changelog            | 34 +++++++++++++++++++++++++++++++++-
 dev-requirements.txt |  2 +-
 nibabel/info.py      |  8 ++++----

will you @effigies remerge master into release branch or the plan is different and you would like to keep some stuff in master away from the release? (although it seems just diff related:

$> git diff origin/rel/2.3.1...origin/master --stat
 .travis.yml                         |   4 +-
 appveyor.yml                        |   2 +-
 dev-requirements.txt                |   1 -
 nibabel/cmdline/diff.py             | 240 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++------------
 nibabel/cmdline/tests/test_utils.py |  50 ++++++++++++--
 nibabel/tests/test_diff.py          | 102 +++++++++++++++--------------
 nibabel/tests/test_scripts.py       |   8 ++-

meanwhile I will locally merge master into rel/2.3.1 for testing

[effigies]

Merged master in. Will update the changelog when I get another minute.

[effigies]

@yarikoptic Nearly ready. Are your tests all green? (There was a very minor pull this morning, but if you want to do another round, fine with me.)

[pauldmccarthy]

Hey @effigies, when are you intending to release? I might have just found an issue in the mess that is keep_file_open+indexed_gzip.drop_handles. Am aiming to resolve it and get a PR up within the next hour or so...

[effigies]

Plan was today, but if there's a bug, I can wait. Can you start a PR with a test that fails, even before fixing it? That will make reviewing simpler.

[pauldmccarthy]

Sure, I'll do so ASAP. I think the existing ArrayProxy are invalid, so I'll fix them as a first step...

[pauldmccarthy]

Started a PR at #679 ...

[effigies]

@yarikoptic Final tests okay? (I can't remember what you called them...)