DEP: Remove datetime64 deprecation warning when constructing with timezone by rmhowe425 · Pull Request #24193 · numpy/numpy

rmhowe425 · 2023-07-15T23:30:08Z

This commit changes the deprecation warning for np.datetime64 to a UserWarning. Updates to the doc string for np.datetime64 are also provided to explicitly state that timezones for datetime64 objects are not available.

Closes #23904

seberg · 2023-07-17T14:41:54Z

Thanks, it's too bad that we didn't have one, but could you add a test to cover the path when a datetime.datetime object includes timezone information? (C-code coverage says that this path is never hit.)

seberg · 2023-07-17T14:47:04Z

    or datetime format.
+
+    When parsing a string to create a datetime object, if the string contains
+    a timezone, the timezone will be dropped.


What's the exact logic? A userwarning is given (I think it would be good to mention), we do have a Warnings heading in principle, too.

Can we explain (and check) what we actually do? And do we actually want to keep it for all of them?

Timezones differences without the Z are simply applied

A Z (and I guess also trailing) is simply ignored, the naive timezone is assumed to be UTC?

Is that correct?

charris · 2023-07-22T20:42:36Z

Could use a release note.

rmhowe425 · 2023-07-30T00:20:36Z

@seberg Looks like I have 4 tests failing because of: ERROR: Problem encountered: No BLAS library detected!.

Is this issue stemming from something that I have done, or the build environment?

mattip · 2023-07-30T04:13:34Z

Probably the CI changes. You may need to rebase off main and force push or merge main into your branch.

rebase

rmhowe425 · 2023-07-30T13:37:34Z

Thanks @mattip! That fixed my issue!

mattip · 2023-07-30T15:12:45Z

+        with pytest.warns(UserWarning, match=msg):
+            t0 = np.datetime64('2023-06-09T12:18:40Z', 'ns')
+
+        t0 = np.datetime64('2023-06-09T12:18:40', 'ns')


The tests no longer belong in this file, which is intended to test deprecations. Are the use cases covered in other tests?

Ah good point! Moving the tests to test_datetime.py.

From what I can tell, these tests are not covered in other places.

…ecation messages.

seberg

I still think it isn't really clear what is going on, I don't really recall discussing it but I guess we did and the docs are the only thing that needs updating.

I think numpy datetimes are naive but we are parsing them as if they were UTC?

I just will note that upgrading the warning might actually make it a lot more noisy in practice (because not everyone sees DeprecationWarnings).

seberg · 2023-07-31T08:27:15Z

+
+    Datetime64 object should be considered to be UTC and therefore have an 
+    offset of +0000.


Suggested change

Datetime64 object should be considered to be UTC and therefore have an

offset of +0000.

For the purpose of parsing time offsets the timezone naive `datetime64`

behaves as if converting to UTC.

Maybe? I don't think the old sentence is correct, the warning is there for a reason after all.

Probably this could be part of a Notes section?

seberg · 2023-07-31T08:29:17Z

-                return -1;
-            }
+            if (PyErr_WarnEx(PyExc_UserWarning,
+                "no explicit representation of timezones available for np.datetime64",


This seems not very clear to me? Maybe also mention that parsing to naive datetime but handling timezone as if we were UTC?

(If that is correct)

seberg · 2023-07-31T08:30:19Z

+            "no explicit representation of timezones available for np.datetime64",
+            1) < 0) {
+                return -1;
+            }


NIT: Please undo this indent. You can over-indent the above.

charris · 2023-09-02T19:04:37Z

close/reopen

charris · 2023-09-05T17:38:43Z

Be nice to finish this up. The test failure looks irrelevant and should go away if this is rebased.

rmhowe425 · 2023-11-26T18:46:10Z

Working on fixing this now!

…ecation messages.

rmhowe425 · 2023-11-26T19:55:26Z

@melissawm I rebased and all of my tests are passing.

The number of lines contributed for my PR does not appear to be correct.....I'm hoping that this won't interfere with reviewing my PR

melissawm · 2023-11-27T17:30:52Z

It looks like something went wrong on your rebase, as you now have a large number of unrelated commits listed here. Do you think you can take a look to see what happened?

charris · 2024-02-05T03:17:53Z

This needs a final review.

mattip

LGTM.

mattip · 2024-02-05T13:20:40Z

I don't really recall discussing it but ...

@seberg should we re-open this for discussion?

rossbar

FWIW the changes here agree with the discussion in #23904 and the triage meeting, so +1 from me as well!

Co-authored-by: Ross Barnowski <rossbar@berkeley.edu>

mattip · 2024-02-07T18:30:04Z

Thanks @rmhowe425

rmhowe425 added 3 commits July 15, 2023 19:06

Changing DeprecationWarning to UserWarning and updating unit tests.

3063f50

Fixing formatting errors based on lint checks.

a4b483a

Updating docstring

0346208

github-actions Bot added the 01 - Enhancement label Jul 15, 2023

Fixing formatting error.

a816177

rmhowe425 changed the title ~~ENH: datetime64: Remove deprecation warning when constructing with timezone~~ ENH: Remove datetime64 deprecation warning when constructing with timezone Jul 17, 2023

seberg reviewed Jul 17, 2023

View reviewed changes

charris changed the title ~~ENH: Remove datetime64 deprecation warning when constructing with timezone~~ DEP: Remove datetime64 deprecation warning when constructing with timezone Jul 22, 2023

charris added the 09 - Backport-Candidate PRs tagged should be backported label Jul 22, 2023

charris added this to the 1.25.2 release milestone Jul 25, 2023

rmhowe425 added 2 commits July 29, 2023 17:12

Updating documentation and unit tests.

e3c8e25

Adding a release note.

1d3c807

rmhowe425 requested a review from seberg July 30, 2023 00:19

charris modified the milestones: 1.25.2 release, 1.26.0 release Jul 30, 2023

Merge pull request #1 from rmhowe425/main

0bf6703

rebase

mattip reviewed Jul 30, 2023

View reviewed changes

Moving unit tests to datetime.py since tests no longer check for depr…

02235b3

…ecation messages.

rmhowe425 requested a review from mattip July 30, 2023 18:18

seberg reviewed Jul 31, 2023

View reviewed changes

charris closed this Sep 2, 2023

charris reopened this Sep 2, 2023

charris removed this from the 1.26.0 release milestone Sep 16, 2023

rmhowe425 added 8 commits November 26, 2023 13:59

Changing DeprecationWarning to UserWarning and updating unit tests.

e416202

Fixing formatting errors based on lint checks.

0d54b2e

Updating docstring

545c2e3

Fixing formatting error.

dddc7e9

Updating documentation and unit tests.

489ec7b

Adding a release note.

a412394

Moving unit tests to datetime.py since tests no longer check for depr…

e883bc0

…ecation messages.

Rebase

327a152

rmhowe425 requested a review from seberg November 26, 2023 19:21

charris modified the milestones: 1.26.3 release, 1.26.4 release Dec 25, 2023

Merge branch 'numpy:main' into dev/datetime64-depr_warning

7c937ad

charris removed the 09 - Backport-Candidate PRs tagged should be backported label Feb 5, 2024

charris removed this from the 1.26.4 release milestone Feb 5, 2024

charris added this to the 2.0.0 release milestone Feb 5, 2024

mattip approved these changes Feb 5, 2024

View reviewed changes

rossbar approved these changes Feb 5, 2024

View reviewed changes

Comment thread numpy/_core/_add_newdocs_scalars.py Outdated

rmhowe425 and others added 2 commits February 7, 2024 07:57

Update numpy/_core/_add_newdocs_scalars.py

9d177d5

Co-authored-by: Ross Barnowski <rossbar@berkeley.edu>

Merge branch 'numpy:main' into dev/datetime64-depr_warning

4addb09

mattip merged commit a315987 into numpy:main Feb 7, 2024

rmhowe425 deleted the dev/datetime64-depr_warning branch June 25, 2024 20:41

pont-us mentioned this pull request Nov 19, 2024

Avoid numpy "parsing timezone aware datetimes is deprecated" warning in tile server xcube-dev/xcube#807

Closed

RonaldAJ mentioned this pull request May 26, 2025

ENH: datetime64: Remove deprecation warning when constructing with timezone #23904

Closed

This comment was marked as spam.

Sign in to view


		Datetime64 object should be considered to be UTC and therefore have an
		offset of +0000.

Uh oh!

Uh oh!

Conversation

rmhowe425 commented Jul 15, 2023

Uh oh!

seberg commented Jul 17, 2023

Uh oh!

seberg Jul 17, 2023

Choose a reason for hiding this comment

Uh oh!

charris commented Jul 22, 2023

Uh oh!

rmhowe425 commented Jul 30, 2023

Uh oh!

mattip commented Jul 30, 2023

Uh oh!

rmhowe425 commented Jul 30, 2023

Uh oh!

mattip Jul 30, 2023

Choose a reason for hiding this comment

Uh oh!

rmhowe425 Jul 30, 2023

Choose a reason for hiding this comment

Uh oh!

seberg left a comment

Choose a reason for hiding this comment

Uh oh!

seberg Jul 31, 2023

Choose a reason for hiding this comment

Uh oh!

seberg Jul 31, 2023

Choose a reason for hiding this comment

Uh oh!

seberg Jul 31, 2023

Choose a reason for hiding this comment

Uh oh!

charris commented Sep 2, 2023

Uh oh!

charris commented Sep 5, 2023

Uh oh!

rmhowe425 commented Nov 26, 2023

Uh oh!

rmhowe425 commented Nov 26, 2023

Uh oh!

melissawm commented Nov 27, 2023

Uh oh!

charris commented Feb 5, 2024

Uh oh!

mattip left a comment

Choose a reason for hiding this comment

Uh oh!

mattip commented Feb 5, 2024

Uh oh!

rossbar left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

mattip commented Feb 7, 2024

Uh oh!

This comment was marked as spam.

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants