MAINT: remove support for python 2.6 by aabadie · Pull Request #437 · joblib/joblib

aabadie · 2016-12-02T14:34:12Z

Fixes #435

Achievements:

PY26 global variable dropped
Travis CI matrix updated
Most of existing references to python 2.6 have been removed
python 2.6 test data removed
string formatting style updated

2 things are not done though because I need someone else's opinion:

in doc/Memory.rst : the collision warning is also available with python >= 2.7. Should I just change 2.6 in 2.7 or rephrase ?
in joblib/func_inspect.py: the getfullargspec() docstring suggests to rewrite this method using a backport of Python 3. Should this be done in this PR or does it deserve a dedicated PR ?

lesteve · 2016-12-02T15:15:18Z

in doc/Memory.rst : the collision warning is also available with python >= 2.7. Should I just change 2.6 in 2.7 or rephrase ?

If you have the collision warning in Python 2.7 use Python 2.7.

in joblib/func_inspect.py: the getfullargspec() docstring suggests to rewrite this method using a backport of Python 3. Should this be done in this PR or does it deserve a dedicated PR ?

You know my answer to this one, right ;-) ? Dedicated PR please. To be honest the way we do it in scikit-learn is not that great we use a backport of inspect.signature that does not seem to be very actively maintained. One pyston guy opened a reasonably trivial PR and got no answer there. See testing-cabal/funcsigs#30 for more details.

lesteve

The changes seem fine, but were there not more code like your streaming compression stuff that was Python 2.6 specific?

lesteve · 2016-12-02T15:21:51Z

.travis.yml


 env:
  matrix:
-    - PYTHON_VERSION="2.6" NUMPY_VERSION="1.6"


Use Python 2.7 to keep testing for numpy 1.6

well, travis shows that this change needs new test data to be generated...

Yep good point, should be quick with the scripts, also add py27-np16 to the list of conda envs in the main script regenerating the data for all the envs.

Hmmm the tests pass now although you have not regenerated the data. This is a bit worrisome.

lesteve · 2016-12-02T15:22:09Z

benchmarks/bench_pickle.py

    """Nice bench summary function."""
    summary = """Benchmark summary:
    - Global values:
-        . Joblib version: {0}


Kudos for the nice initiative

aabadie · 2016-12-02T16:20:34Z

The changes seem fine, but were there not more code like your streaming compression stuff that was Python 2.6 specific?

Yes . The cases were also used for some file object (BZ2File) with python 2.7.

lesteve · 2016-12-02T16:27:52Z

Yes . The cases were also used for some file object (BZ2File) with python 2.7.

I am a bit disappointed not to remove more code :).

lesteve · 2016-12-05T08:25:42Z

The Travis error needs to be looked at. We get two warnings when we expect one, not sure why.

======================================================================
FAIL: joblib.test.test_numpy_pickle.test_joblib_pickle_across_python_versions
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/travis/miniconda2/envs/testenv/lib/python2.7/site-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
  File "/home/travis/build/joblib/joblib/joblib/test/test_numpy_pickle.py", line 525, in test_joblib_pickle_across_python_versions
    _check_pickle(fname, expected_list)
  File "/home/travis/build/joblib/joblib/joblib/test/test_numpy_pickle.py", line 457, in _check_pickle
    assert len(caught_warnings) == expected_nb_warnings
AssertionError: 
>>  assert len([<warnings.WarningMessage object at 0x2b2522d8cdd0>, <warnings.WarningMessage object at 0x2b2522e47d10>]) == 1

lesteve · 2016-12-05T08:56:44Z

It looks like we catch a DeprecationWarning from numpy ... DeprecationWarning('The compiler package is deprecated and removed in Python 3.x.',). Maybe the simplest thing to do is to filter warnings manually if w.filename does not contain joblib. Maybe there is a better way to do it more cleanly with a warnings filter ...

lesteve · 2016-12-05T09:12:35Z

I pushed a fix, let's see how this goes.

lesteve · 2016-12-05T09:56:18Z

.travis.yml


 env:
  matrix:
-    - PYTHON_VERSION="2.6" NUMPY_VERSION="1.6"


Hmmm the tests pass now although you have not regenerated the data. This is a bit worrisome.

lesteve · 2016-12-05T09:56:41Z

joblib/test/data/generate_data.sh


 # Change the list according to your local conda/virtualenv env.
-CONDA_ENVS="py26-np16 py27-np17 py33-np18 py34-np19 py35-np19"
+CONDA_ENVS="py27-np17 py33-np18 py34-np19 py35-np19"


Add py27-np16 here.

lesteve · 2016-12-05T10:22:23Z

I generated the data for python 2.7 and numpy 1.6. While doing that I realised that there were some .lzma and .xz files for Python 2.6, that does not make any sense right @aabadie ?

lesteve · 2016-12-05T12:30:44Z

OK, merging this one, thanks!

lesteve · 2016-12-06T06:50:33Z

There are still Python 2.6 stuff, @aabadie could you finish up the cleanup?

$ git grep '2\.6'
README.rst:versions are 2.6+ and 3.3+). Numpy (at least version 1.6.1) is an
joblib/func_inspect.py:    once we drop support for Python 2.6. We went for a simpler
joblib/func_inspect.py:    which is not available in Python 2.6.
joblib/numpy_pickle_utils.py:    Required as e.g. ZipExtFile in python 2.6 can return less data than
joblib/pool.py:    # customize the dispatch table without side effects in Python 2.6
joblib/testing.py:    # Python <= 2.6, we still need nose here
setup.py:              'Programming Language :: Python :: 2.6',

aabadie · 2016-12-06T09:06:58Z

indeed @lesteve, I'll issue another PR soon.

lesteve · 2016-12-06T09:24:17Z

Can you add an entry in CHANGES.rst as well, saying that Python 2.6 support has been dropped in version ? (maybe worth going to 0.11 for this exact reason).

GaelVaroquaux · 2016-12-06T09:43:42Z

(maybe worth going to 0.11 for this exact reason).

+1 for jumping the version number. That said, let's release only when we have significant new features: the goal is not to drop 2.6 and not bring anything to people.

lesteve · 2016-12-06T14:04:23Z

+1 for jumping the version number. That said, let's release only when we have significant new features: the goal is not to drop 2.6 and not bring anything to people.

Yep, agreed.

* commit '0.10.2-55-g660fe5d': (52 commits) MAINT simplify way to skip doctests Do not print frames with IPython internals in format_stack BF(TST): compare for version in pickled filename ignoring full path, and looking for _ as well (joblib#445) [MRG] Remove nose dependency completely. (joblib#441) MAINT add Python 3.5 in setup.py Remove some Python 2.6 specific code [MRG] Make test execution on CI possible through py.test command. (joblib#433) Remove more python 2.6 related code (joblib#440) [MRG] Use commit range for diff in flake8_diff.sh (joblib#439) MAINT remove support for python 2.6 (joblib#437) Improve doc about caching methods (joblib#432) Get the tests passing with py.test MAINT update flake8_diff.sh COSMIT move stdlib imports where they belong [MRG] Replace assert_* methods with "assert" keyword statements. (joblib#430) Reraise when the exception is not a TransportableException (joblib#429) Make sure LICENSE.txt is included in the wheel Add LICENSE.txt Update doc/paralle.rst [MRG+1] Remove all nose imports from test scripts. (joblib#422) ...

* releases: (52 commits) MAINT simplify way to skip doctests Do not print frames with IPython internals in format_stack BF(TST): compare for version in pickled filename ignoring full path, and looking for _ as well (joblib#445) [MRG] Remove nose dependency completely. (joblib#441) MAINT add Python 3.5 in setup.py Remove some Python 2.6 specific code [MRG] Make test execution on CI possible through py.test command. (joblib#433) Remove more python 2.6 related code (joblib#440) [MRG] Use commit range for diff in flake8_diff.sh (joblib#439) MAINT remove support for python 2.6 (joblib#437) Improve doc about caching methods (joblib#432) Get the tests passing with py.test MAINT update flake8_diff.sh COSMIT move stdlib imports where they belong [MRG] Replace assert_* methods with "assert" keyword statements. (joblib#430) Reraise when the exception is not a TransportableException (joblib#429) Make sure LICENSE.txt is included in the wheel Add LICENSE.txt Update doc/paralle.rst [MRG+1] Remove all nose imports from test scripts. (joblib#422) ...

MAINT: remove support for python 2.6

e8b87e4

aabadie added the in progress label Dec 2, 2016

lesteve reviewed Dec 2, 2016

View reviewed changes

adressing comments

5ab4a99

Fix tests with numpy 1.6

4be6801

lesteve reviewed Dec 5, 2016

View reviewed changes

Generated data files for python 2.7 and numpy 1.6

99e90d0

lesteve force-pushed the remove_py26 branch from c8f3283 to 99e90d0 Compare December 5, 2016 10:20

lesteve merged commit 953f388 into joblib:master Dec 5, 2016

lesteve removed the in progress label Dec 5, 2016

lesteve mentioned this pull request Dec 5, 2016

[MRG] Make test execution on CI possible through py.test command. #433

Merged

3 tasks

aabadie mentioned this pull request Dec 6, 2016

Remove py26 bis #440

Merged

aabadie mentioned this pull request Dec 7, 2016

Rewrite getfullargspec function in func_inspect module #442

Closed

aabadie deleted the remove_py26 branch February 27, 2018 15:42

Conversation

aabadie commented Dec 2, 2016

Uh oh!

lesteve commented Dec 2, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lesteve left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

aabadie commented Dec 2, 2016

Uh oh!

lesteve commented Dec 2, 2016

Uh oh!

lesteve commented Dec 5, 2016

Uh oh!

lesteve commented Dec 5, 2016

Uh oh!

lesteve commented Dec 5, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lesteve commented Dec 5, 2016

Uh oh!

lesteve commented Dec 5, 2016

Uh oh!

lesteve commented Dec 6, 2016

Uh oh!

aabadie commented Dec 6, 2016

Uh oh!

lesteve commented Dec 6, 2016

Uh oh!

GaelVaroquaux commented Dec 6, 2016 via email

Uh oh!

lesteve commented Dec 6, 2016

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

lesteve commented Dec 2, 2016 •

edited

Loading