Release 0.15.0 #752

orbeckst · 2016-03-01T23:57:39Z

Discussions for release 0.15.0

~~Note that with semantic versioning no new features can be in this release. If we want new features, we will have to make it 0.15.0.~~ -- Contact Analysis happened, so this is now a full release.

(edited by Richard)

kain88-de · 2016-03-02T08:48:36Z

How should the branches then be handled? Due to me merging the new contact analysis develop is now targeting 0.15.0. We can cherry-pick from develop to master and publish from master directly once we solved the important bugs for 0.14.0. For the 0.15.0 release @richardjgowers would then have to merge develop into master instead of just rebasing (this might lead to some commits appearing twice in the history).

Sorry for creating the mess by merging the contact analysis branch so eagerly

richardjgowers · 2016-03-02T09:46:44Z

I don't see the point in a 14.1 it's just going to cause headaches. We'll
call the next one 15 (contact and deprecations) and topology becomes 16

On Wed, 2 Mar 2016 08:48 kain88-de, [email protected] wrote:

How should the branches then be handled? Due to me merging the new contact
analysis develop is now targeting 0.15.0. We can cherry-pick from develop
to master and publish from master directly once we solved the important
bugs for 0.14.0. For the 0.15.0 release @richardjgowers
https://github.com/richardjgowers would then have to merge develop into
master instead of just rebasing (this might lead to some commits appearing
twice in the history).

Sorry for creating the mess by merging the contact analysis branch so
eagerly

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

orbeckst · 2016-03-02T16:06:39Z

On 2 Mar, 2016, at 02:46, Richard Gowers wrote:

I don't see the point in a 14.1 it's just going to cause headaches. We'll
call the next one 15 (contact and deprecations) and topology becomes 16

Yes, that works, too.

Oliver Beckstein * [email protected]
skype: orbeckst * [email protected]

kain88-de · 2016-03-18T10:23:04Z

@richardjgowers should we set the release target to a more realistic end of april instead of 8 days ago?

orbeckst · 2016-03-21T05:43:06Z

I set it to April 1 but @richardjgowers should change it... I just didn't want to see the red "milestone delayed" messages anymore, especially while we're in the GSoC applicant vetting period.

I also edited the description, it's going to be more than just bug fixes.

Its most important function is to getting us closer to the semi-mythical day when #363 will finally be merged... no pressure @richardjgowers and @dotsdl ... ;-)

kain88-de · 2016-03-21T15:41:51Z

I was thinking how to motivate @richardjgowers and @dotsdl. What was left to do to merge the branch?

jbarnoud · 2016-03-21T15:45:21Z

@kain88-de tests among other things. Last time I had a look, a bunch of tests where failing in the 363 branch.

A procedure to pinpoint regressions would be useful too as develop evolved quite a bit since issue-363 started.

richardjgowers · 2016-03-21T15:48:56Z

Haha I'm finishing up my thesis, so not unmotivated but maybe a little preoccupied!

The hold up to this release is we wanted deprecations of things that will change once 363 gets merged, for example SegmentGroup.positions doesn't give you per-atom behaviour. So anything that isn't an AtomGroup doing AtomGroup things needs a warning.

richardjgowers · 2016-03-21T15:52:09Z

And then to get the PR done, yeah lots of tests. The old AtomGroup tests are a little haphazard and rely on _atoms among things, so they need rewriting.

dotsdl · 2016-03-21T16:21:10Z

Hey all, just like @richardjgowers, I'm not unmotivated but preoccupied. Been on a tear to release datreant.core, and hopefully doing that by the end of today. I'll be returning to and finishing #698 by the end of the week so we can move this release along.

orbeckst · 2016-03-23T21:03:25Z

FYI: There are a bunch of methods in the code that are scheduled for removal in 0.15.0, for instance AtomGrouyp.selectAtoms().

kain88-de · 2016-04-18T07:39:36Z

OK I think we are almost done with this release. There is very little left to do. The release won't contain much and we can aim for a very big release 0.16.0 at the end of the summer.

TODO

fix warnings introduced in All deprecations on AtomGroup and friends getters and setters. #698, check fix pos warnings #828, Removed some use of deprecated methods #825
fix up new new contacts class see Contact improvements #792

kain88-de · 2016-05-03T20:35:01Z

@MDAnalysis/coredevs what is still needed to make this happen before the end of the month?

mnmelo · 2016-05-04T14:27:02Z

I'm still busy with #799, which turns out to be a deep rabbit hole for several reasons. I'm fine with leaving that for 16.0, but am currently devoting more time to it so I can get rid of it asap.

richardjgowers · 2016-05-04T19:50:17Z

Ok so 0.15 release this weekend!

On Wed, 4 May 2016 15:27 mnmelo, [email protected] wrote:

I'm still busy with #799
#799, which turns out to
be a deep rabbit hole for several reasons. I'm fine with leaving that for
16.0, but am currently devoting more time to it so I can get rid of it asap.

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#752 (comment)

orbeckst · 2016-05-04T20:55:33Z

Did we remove deprecated names such as selectAtoms() #752 (comment) ?

kain88-de · 2016-05-05T08:19:30Z

@mnmelo We can just add a warning about the missing cython package for development installations in the next release. Then you have some time for a proper fix.

@orbeckst Nope not yet.

kain88-de · 2016-05-05T08:21:42Z

@orbeckst These deprecations have been moved by @dotsdl to 0.16.0 .
As far as I can see there are no more deprecations scheduled for this release.

kain88-de · 2016-05-06T17:22:06Z

@richardjgowers can we also make a minor release for GridDataFormats? I changed some stuff there to and I think it be nicer to have a non-dev version on conda.

orbeckst · 2016-05-06T17:23:59Z

On 6 May, 2016, at 10:22, kain88-de [email protected] wrote:

@richardjgowers can we also make a minor release for GridDataFormats? I changed some stuff there to and I think it be nicer to have a non-dev version on condo.

@kain88-de – you can make a GridDataFormat release yourself, you should have all the authority that you need (if not, tell me what’s missing). Just update the MDA setup.py accordingly.

Oliver Beckstein * [email protected]
skype: orbeckst * [email protected]

kain88-de · 2016-05-06T18:41:48Z

I'm not aware I have rights to push to pypi too. I can make everything on github and conda that's about it.

orbeckst · 2016-05-06T19:16:05Z

I will add you. What’s your username on PyPi?

On 6 May, 2016, at 11:41, kain88-de [email protected] wrote:

I'm not aware I have rights to push to pypi too. I can make everything on github and conda that's about it.

Oliver Beckstein * [email protected]
skype: orbeckst * [email protected]

kain88-de · 2016-05-06T20:12:48Z

@orbeckst thanks that's a nice offer. But I have enough responsibilities as it is and don't want to have rights for PyPi right now.

orbeckst · 2016-05-06T20:32:11Z

On 6 May, 2016, at 13:12, kain88-de [email protected] wrote:

@orbeckst thanks that's a nice offer. But I have enough responsibilities as it is and don’t want to have rights for PyPi right now.

Then you have to tell me when to do a new release – apparently, in the past I was the one who did it because there’s no other maintainer on PyPi.

(I am more than happy to also hand this to @richardjgowers if he wants to, of course ;-))

Anyway, open an issue for a new GridDataFormats release and we can quickly push it out.

Oliver Beckstein * [email protected]
skype: orbeckst * [email protected]

richardjgowers · 2016-05-09T10:51:41Z

So we're getting some (intermittent) failures on the Mac OS 2.7 minimal job. It looks like it's failing to create new file handles, which last time was caused by Universes not closing properly (and so each test creating a new handle, eventually hitting the upper limit on these). Looking into this.

kain88-de · 2016-05-10T20:13:10Z

The builds pass again. What did you change?

richardjgowers · 2016-05-10T20:15:19Z

Literally nothing :S

jbarnoud · 2016-05-10T20:17:52Z

On 10/05/16 22:15, Richard Gowers wrote:

Literally nothing :S

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#752 (comment)

I restarted a build on travis earlier today. The failing badge was
bothering me. Now I realize I made it more difficult to investigate, sorry.

richardjgowers · 2016-05-11T15:29:30Z

So on the full Linux build, we got this:

======================================================================

ERROR: test_density_from_Universe (MDAnalysisTests.analysis.test_density.Test_density_from_Universe)

----------------------------------------------------------------------

Traceback (most recent call last):

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/numpy/testing/decorators.py", line 147, in skipper_func

    return f(*args, **kwargs)

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/MDAnalysisTests/analysis/test_density.py", line 102, in setUp

    self.universe = mda.Universe(self.topology, self.trajectory)

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/MDAnalysis/core/AtomGroup.py", line 4353, in __init__

    self.load_new(coordinatefile, **kwargs)

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/MDAnalysis/core/AtomGroup.py", line 4889, in load_new

    self.trajectory = reader(filename, **kwargs)    # unified trajectory API

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/MDAnalysis/coordinates/XDR.py", line 80, in __init__

    self._load_offsets()

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/MDAnalysis/coordinates/XDR.py", line 114, in _load_offsets

    data = read_numpy_offsets(fname)

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/MDAnalysis/coordinates/XDR.py", line 61, in read_numpy_offsets

    return {k: v for k, v in six.iteritems(np.load(filename))}

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/numpy/lib/npyio.py", line 399, in load

    pickle_kwargs=pickle_kwargs)

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/numpy/lib/npyio.py", line 162, in __init__

    _zip = zipfile_factory(fid)

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/numpy/lib/npyio.py", line 92, in zipfile_factory

    return zipfile.ZipFile(*args, **kwargs)

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/zipfile.py", line 770, in __init__

    self._RealGetContents()

  File "/home/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/zipfile.py", line 811, in _RealGetContents

    raise BadZipfile, "File is not a zip file"

BadZipfile: File is not a zip file

I'm going to assume it's just a travis hiccup and redo the build

kain88-de · 2016-05-11T15:36:09Z

What the heck? I ran the full testsuite in the morning on my laptop everything passed.

kain88-de · 2016-05-12T07:47:31Z

@richardjgowers where do we stand on build? Just assume that travis has some random hickups lately? Or do we have a general problem with file handles that only pop up randomly?

richardjgowers · 2016-05-12T07:51:10Z

Yeah all seems good now

On Thu, 12 May 2016 08:47 kain88-de, [email protected] wrote:

@richardjgowers https://github.com/richardjgowers where do we stand on
build? Just assume that travis has some random hickups lately?

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#752 (comment)

kain88-de · 2016-05-12T08:01:49Z

So a release this weekend?

richardjgowers · 2016-05-12T09:00:25Z

Yeah I wanted to do it last weekend but we had failures everywhere. Is the
conda build process documented enough for me to do it? Or did you want to
do it?

On Thu, 12 May 2016 at 09:01 kain88-de [email protected] wrote:

So a release this weekend?

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#752 (comment)

kain88-de · 2016-05-12T09:24:52Z

@richardjgowers I can do it after you made the release. I will also add documentation for it on the wiki soon, it is a little bit more involved since you need to build it on a special virtual machine.

richardjgowers · 2016-05-12T12:08:45Z

Ok we got the weird failure again here. There's lots of errors, but it looks like it's because we can't open files, many because we've got too many open simultaneously? (IOError: [Errno 24] Too many open files: '/Users/travis/build/MDAnalysis/mdanalysis/miniconda/envs/pyenv/lib/python2.7/site-packages/MDAnalysisTests/data/adk_dims.dcd')

We had this before which was caused by Universes not deleting properly, causing handles to remain open, but the extra checks in #323 should be reporting if this is happening?

The only other change I can think of is the tempdir, but this was just copy pasting the code into our repo right?

richardjgowers · 2016-05-16T08:55:34Z

Ok, master branch updated, we're live on Pypi, docs uploaded to github and pypi docs. @kain88-de sorry about the delay

kain88-de · 2016-05-16T09:29:04Z

@richardjgowers thanks for publishing. Don't worry about the delay it's still weekend in Germany thanks to national holidays

kain88-de · 2016-05-16T11:17:44Z

https://anaconda.org/MDAnalysis/mdanalysis/files

conda release is online

dotsdl · 2016-05-16T18:33:00Z

Sweet, thanks @richardjgowers and @kain88-de for doing the heavy lifting!

orbeckst added the release label Mar 1, 2016

orbeckst added this to the 0.14.1 - Bugfixes milestone Mar 1, 2016

richardjgowers changed the title ~~Release 0.14.1~~ Release 0.15.0 Mar 2, 2016

richardjgowers self-assigned this May 16, 2016

kain88-de closed this as completed May 16, 2016

Release 0.15.0 #752

Release 0.15.0 #752

Comments

orbeckst commented Mar 1, 2016

kain88-de commented Mar 2, 2016

richardjgowers commented Mar 2, 2016

orbeckst commented Mar 2, 2016

kain88-de commented Mar 18, 2016

orbeckst commented Mar 21, 2016

kain88-de commented Mar 21, 2016

jbarnoud commented Mar 21, 2016

richardjgowers commented Mar 21, 2016

richardjgowers commented Mar 21, 2016

dotsdl commented Mar 21, 2016

orbeckst commented Mar 23, 2016

kain88-de commented Apr 18, 2016 • edited Loading

TODO

kain88-de commented May 3, 2016

mnmelo commented May 4, 2016

richardjgowers commented May 4, 2016

orbeckst commented May 4, 2016

kain88-de commented May 5, 2016

kain88-de commented May 5, 2016 • edited Loading

kain88-de commented May 6, 2016

orbeckst commented May 6, 2016

kain88-de commented May 6, 2016

orbeckst commented May 6, 2016

kain88-de commented May 6, 2016

orbeckst commented May 6, 2016

richardjgowers commented May 9, 2016

kain88-de commented May 10, 2016

richardjgowers commented May 10, 2016

jbarnoud commented May 10, 2016

richardjgowers commented May 11, 2016

kain88-de commented May 11, 2016

kain88-de commented May 12, 2016 • edited Loading

richardjgowers commented May 12, 2016

kain88-de commented May 12, 2016

richardjgowers commented May 12, 2016

kain88-de commented May 12, 2016

richardjgowers commented May 12, 2016

richardjgowers commented May 16, 2016

kain88-de commented May 16, 2016

kain88-de commented May 16, 2016

dotsdl commented May 16, 2016

kain88-de commented Apr 18, 2016 •

edited

Loading

kain88-de commented May 5, 2016 •

edited

Loading

kain88-de commented May 12, 2016 •

edited

Loading