Updated condition to skip for pytables build issue on numpy 1.15 #22098 #22522

NikhilKumarM · 2018-08-27T14:05:32Z

closes TST: remove skip for pytables build issue on numpy 1.15 #22098
passes git diff upstream/master -u -- "*.py" | flake8 --diff

TomAugspurger

Could you fetch and merge master? That should fix the circleCI failure.

TomAugspurger · 2018-08-27T14:23:45Z

pandas/compat/numpy/__init__.py

@@ -74,5 +75,6 @@ def np_array_datetime64_compat(arr, *args, **kwargs):
           '_np_version_under1p12',
           '_np_version_under1p13',
           '_np_version_under1p14',
-           '_np_version_under1p15'
+           '_np_version_under1p15',
+           '_np_version_equal1p15'


nint: add a trailing comma

Is the comma required after '_np_version_equal1p15'?

No, but it makes the next diff nicer.

TomAugspurger · 2018-08-27T14:23:47Z

pandas/tests/io/test_pytables.py

@@ -2192,7 +2192,7 @@ def test_unimplemented_dtypes_table_columns(self):
            pytest.raises(TypeError, store.append, 'df_unimplemented', df)

    @pytest.mark.skipif(
-        not _np_version_under1p15,
+        _np_version_equal1p15,
        reason=("pytables conda build package needs build "


You could update this message to link to the upstream numpy issue.

…anch_02

NikhilKumarM · 2018-08-27T16:05:55Z

I have fetched and merged master. But still shows a build error. Anything wrong with my code.

TomAugspurger · 2018-08-27T16:10:47Z

Looks like numpy 1.15 was installed.

Anaconda doesn't seem to have 1.15.1 built yet https://repo.continuum.io/pkgs/main/linux-64/. Will probably be up this week.

jeffyancey · 2018-08-27T18:45:16Z

pandas/compat/numpy/__init__.py

@@ -15,6 +15,7 @@
 _np_version_under1p13 = _nlv < LooseVersion('1.13')
 _np_version_under1p14 = _nlv < LooseVersion('1.14')
 _np_version_under1p15 = _nlv < LooseVersion('1.15')
+_np_version_equal1p15 = _nlv == LooseVersion('1.15')


Is the pattern to change this to

_np_version_under1p16 = _nlv < LooseVersion('1.16') _np_version_equal1p16 = _nlv == LooseVersion('1.16')

when numpy 1.16 comes out? Or is this a one off to meet a current need?

TomAugspurger · 2018-08-27T18:52:46Z

This is for 1.15.0 exactly. It'd probably be best to define a skipif for NumPy 1.15.0 in https://github.com/pandas-dev/pandas/blob/master/pandas/util/_test_decorators.py and just use that in the test.

…

On Mon, Aug 27, 2018 at 1:47 PM Jeff ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In pandas/compat/numpy/__init__.py <#22522 (comment)>: > @@ -15,6 +15,7 @@ _np_version_under1p13 = _nlv < LooseVersion('1.13') _np_version_under1p14 = _nlv < LooseVersion('1.14') _np_version_under1p15 = _nlv < LooseVersion('1.15') +_np_version_equal1p15 = _nlv == LooseVersion('1.15') Is the pattern to change this to _np_version_under1p16 = _nlv < LooseVersion('1.16') _np_version_equal1p16 = _nlv == LooseVersion('1.16') when numpy 1.16 comes out? Or is this a one off to meet a current need? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22522 (review)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABQHIu-1HRG3_ocxKBbOoJciz0eyZQAmks5uVD7DgaJpZM4WN2cT> .

jreback · 2018-08-28T12:04:04Z

pandas/tests/io/test_pytables.py

@@ -2192,9 +2192,9 @@ def test_unimplemented_dtypes_table_columns(self):
            pytest.raises(TypeError, store.append, 'df_unimplemented', df)

    @pytest.mark.skipif(
-        not _np_version_under1p15,


don't define anything new here, juse use LooseVersion explicity to check for 1.15.0 as this is the only version affected.

I have changed it. Could you please have a look at it @jreback. let me know if changes required.

…anch_02

jreback · 2018-08-29T12:24:56Z

pandas/tests/io/test_pytables.py

-        not _np_version_under1p15,
-        reason=("pytables conda build package needs build "
-                "with numpy 1.15: gh-22098"))
+        LooseVersion(np.__version__) == LooseVersion('1.15'),


I think you need 1.15.0 here

I think we can use StrictVersion here?

You are right. I see LooseVersion('1.15') and LooseVersion('1.15.0') are not equal. But, StrictVersion('1.15') and Strict Version('1.15.0') are equal. We can use StrictVersion, in that we can use '1.15' or '1.15.0'. If we are using LooseVersion, then it should be '1.15.0'.

is it safe to use StrictVersion then?

I have made changes with LooseVersion('1.15.0'). I think this will suffice. Please let me know if we have to use StrictVersion instead. @jreback @TomAugspurger

…anch_02

jreback · 2018-08-31T10:09:42Z

thanks!

…as-dev#22098 (pandas-dev#22522)

Updated condition to skip for pytables build issue on numpy 1.15 #22098

966741d

TomAugspurger reviewed Aug 27, 2018

View reviewed changes

NikhilKumarM added 2 commits August 27, 2018 10:34

Merge branch 'master' of https://github.com/pandas-dev/pandas into br…

583211f

…anch_02

Updated the message for skipping and added trailing comma

cc2ea5b

jeffyancey reviewed Aug 27, 2018

View reviewed changes

jreback requested changes Aug 28, 2018

View reviewed changes

jreback added the Testing pandas testing functions or related to the test suite label Aug 28, 2018

NikhilKumarM added 2 commits August 28, 2018 08:38

Merge branch 'master' of https://github.com/pandas-dev/pandas into br…

90690bf

…anch_02

Explicitly checking numpy version 1.15 for pytables test

d6cba4c

jreback requested changes Aug 29, 2018

View reviewed changes

jreback added this to the 0.24.0 milestone Aug 29, 2018

NikhilKumarM added 2 commits August 29, 2018 10:00

Merge branch 'master' of https://github.com/pandas-dev/pandas into br…

b252c1e

…anch_02

Changed LooseVersion(1.15) to LooseVersion(1.15.0)

3337b74

jreback merged commit 285eb1e into pandas-dev:master Aug 31, 2018

Sup3rGeo pushed a commit to Sup3rGeo/pandas that referenced this pull request Oct 1, 2018

Updated condition to skip for pytables build issue on numpy 1.15 pand…

e22de52

…as-dev#22098 (pandas-dev#22522)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Updated condition to skip for pytables build issue on numpy 1.15 #22098 #22522

Updated condition to skip for pytables build issue on numpy 1.15 #22098 #22522

NikhilKumarM commented Aug 27, 2018

TomAugspurger left a comment

TomAugspurger Aug 27, 2018

NikhilKumarM Aug 27, 2018

TomAugspurger Aug 27, 2018

TomAugspurger Aug 27, 2018

NikhilKumarM Aug 27, 2018

NikhilKumarM commented Aug 27, 2018

TomAugspurger commented Aug 27, 2018

jeffyancey Aug 27, 2018

TomAugspurger commented Aug 27, 2018 via email

jreback Aug 28, 2018

NikhilKumarM Aug 28, 2018

jreback Aug 29, 2018

jreback Aug 29, 2018

NikhilKumarM Aug 29, 2018 •

edited

Loading

NikhilKumarM Aug 29, 2018 •

edited

Loading

NikhilKumarM Aug 30, 2018 •

edited

Loading

jreback commented Aug 31, 2018

Updated condition to skip for pytables build issue on numpy 1.15 #22098 #22522

Updated condition to skip for pytables build issue on numpy 1.15 #22098 #22522

Conversation

NikhilKumarM commented Aug 27, 2018

TomAugspurger left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

NikhilKumarM commented Aug 27, 2018

TomAugspurger commented Aug 27, 2018

Choose a reason for hiding this comment

TomAugspurger commented Aug 27, 2018 via email

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

NikhilKumarM Aug 29, 2018 • edited Loading

Choose a reason for hiding this comment

NikhilKumarM Aug 29, 2018 • edited Loading

Choose a reason for hiding this comment

NikhilKumarM Aug 30, 2018 • edited Loading

Choose a reason for hiding this comment

jreback commented Aug 31, 2018

NikhilKumarM Aug 29, 2018 •

edited

Loading

NikhilKumarM Aug 29, 2018 •

edited

Loading

NikhilKumarM Aug 30, 2018 •

edited

Loading