CI / BLD: Various CI Backports

[gfyoung]

• Pin blosc to 1.14.3 (Pin blosc to 1.14.3 for travis-27 build #22429)
• Migrate to CircleCI 2.0 (CI: Migrate to circleci 2.0 #21814)
• Pin openpyxl to 2.5.5 (BLD: Fix openpyxl to 2.5.5 #22601)
• Bump numpy to 1.9.3 (CI: Bump to NumPy compat to 1.9.3 #22499)
• Fix locale handling (CLN/TST: Cleanup tests in test_ujson.py #21739, Fix Python2 test failures in certain locales #22213)
• Upgrade Cython to >=0.28.2 (Accept constant memoryviews in HashTable.lookup #21688)

[pep8speaks]

Hello @gfyoung! Thanks for submitting the PR.

• There are no PEP8 issues in the file pandas/tests/indexes/datetimes/test_misc.py !

• There are no PEP8 issues in the file pandas/tests/io/json/test_compression.py !

• There are no PEP8 issues in the file pandas/tests/io/json/test_pandas.py !

• There are no PEP8 issues in the file pandas/tests/io/parser/test_network.py !

• There are no PEP8 issues in the file pandas/tests/io/test_excel.py !

• There are no PEP8 issues in the file pandas/tests/scalar/timestamp/test_timestamp.py !

• There are no PEP8 issues in the file pandas/tests/series/test_datetime_values.py !

[gfyoung]

@jreback @TomAugspurger : Looks like we need to backport a couple of other things:

• numpy version updates (for the lib.gfortran errors)

• locale fixes!? (for some reason, py35_ascii is picking up way more locales than it does on master, causing all of the failures on that build - there are some slight differences in the implementation of tm.get_locales, so that might be the cause)

[codecov]

Codecov Report

Merging #22637 into 0.23.x will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           0.23.x   #22637      +/-   ##
==========================================
+ Coverage   91.83%   91.84%   +<.01%     
==========================================
  Files         153      153              
  Lines       49591    49588       -3     
==========================================
- Hits        45544    45543       -1     
+ Misses       4047     4045       -2

Flag	Coverage Δ
#multiple	90.24% <100%> (ø)	⬆️
#single	41.79% <100%> (ø)	⬆️

Impacted Files	Coverage Δ
pandas/util/testing.py	84.21% <100%> (+0.15%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 932de54...c780f95. Read the comment docs.

[gfyoung]

@jreback @TomAugspurger :

* https://travis-ci.org/pandas-dev/pandas/jobs/426221423

Build keeps timing out (seems unrelated with PR though).

• https://circleci.com/gh/pandas-dev/pandas/19464
• https://circleci.com/gh/pandas-dev/pandas/19465

Really weird locale failures.

Any thoughts?

[gfyoung]

@jreback @TomAugspurger : Tried doing some more investigation, but still not sure where the locale failures are coming from. I would like to potentially merge everything up to 2b28893 first, which would address the failures I was seeing in #22634.

[TomAugspurger]

What's our plan for release? I don't really have a strong desire to do a 0.23.5, especially if it's a pain to backport things.

…

On Mon, Sep 10, 2018 at 12:39 AM gfyoung ***@***.***> wrote: @jreback <https://github.com/jreback> @TomAugspurger <https://github.com/TomAugspurger> : Tried doing some more investigation, but still not sure where the locale failures are coming from. I would like to potentially merge everything up to 2b28893 <2b28893> first, which would address the failures I was seeing in #22634 <#22634>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#22637 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABQHIrjGnYfVS3yec0lAGTtOc-rFCyo7ks5uZfsXgaJpZM4Wf8ja> .

[TomAugspurger]

#22213 may be the one to fix locales.

[jreback]

could pass on 23.5 and just do 24 maybe by end of sept

[gfyoung]

What's our plan for release? I don't really have a strong desire to do a
0.23.5, especially if it's a pain to backport things.

@TomAugspurger : The "pain" of backporting has been due to CI changes that in hindsight should have been backported earlier than later. I think we should probably backport changes that:

• Modify existing CI configurations / scripts
• Modify existing tests / testing utilities

[gfyoung]

#22213 may be the one to fix locales.

@TomAugspurger : Ah, thanks for pointing that out. Let me see if that solves anything...

[gfyoung]

@jreback @TomAugspurger : Well would you look at that? That seemed to have done the trick! All tests are passing now with Circle 2.0. PTAL.

[TomAugspurger]

LGTM at a glance, thanks. Feel free to merge when you're ready (probably rebase & merge since it's a backport branch).

[gfyoung]

@TomAugspurger : This is against the 0.23.x branch, so I think we're good here.