use pandas.to_timedelta function instead of invoking deprecated ctor arg

[marscher]

Changes

Pandas 2.2 deprecated the unit argument of the Timedeltaindex constructor and suggests to use pandas.totimedelta function instaed. This PR changes this as suggested.

Related Issues

properly fixes #909

Checks

• updated CHANGELOG.md
• updated tests/

[github-actions]

Test Results

2 188 tests  ±0   2 187 ✅ +1   2m 45s ⏱️ -20s
    1 suites ±0       1 💤 ±0 
    1 files   ±0       0 ❌  - 1 

Results for commit 2e6c519. ± Comparison against base commit 82b79b1.

This pull request removes 62 and adds 62 tests. Note that renamed tests count towards both.

weldx.tests.test_time.TestTime ‑ test_add_datetime[TimedeltaIndex-False-False-False]
weldx.tests.test_time.TestTime ‑ test_add_datetime[TimedeltaIndex-False-False-True]
weldx.tests.test_time.TestTime ‑ test_add_datetime[TimedeltaIndex-False-True-False]
weldx.tests.test_time.TestTime ‑ test_add_datetime[TimedeltaIndex-False-True-True]
weldx.tests.test_time.TestTime ‑ test_add_datetime[TimedeltaIndex-True-False-False]
weldx.tests.test_time.TestTime ‑ test_add_datetime[TimedeltaIndex-True-False-True]
weldx.tests.test_time.TestTime ‑ test_add_datetime[TimedeltaIndex-True-True-False]
weldx.tests.test_time.TestTime ‑ test_add_datetime[TimedeltaIndex-True-True-True]
weldx.tests.test_time.TestTime ‑ test_add_timedelta[TimedeltaIndex-h-False-False-False]
weldx.tests.test_time.TestTime ‑ test_add_timedelta[TimedeltaIndex-h-False-False-True]
…

weldx.tests.test_time.TestTime ‑ test_add_datetime[to_timedelta-False-False-False]
weldx.tests.test_time.TestTime ‑ test_add_datetime[to_timedelta-False-False-True]
weldx.tests.test_time.TestTime ‑ test_add_datetime[to_timedelta-False-True-False]
weldx.tests.test_time.TestTime ‑ test_add_datetime[to_timedelta-False-True-True]
weldx.tests.test_time.TestTime ‑ test_add_datetime[to_timedelta-True-False-False]
weldx.tests.test_time.TestTime ‑ test_add_datetime[to_timedelta-True-False-True]
weldx.tests.test_time.TestTime ‑ test_add_datetime[to_timedelta-True-True-False]
weldx.tests.test_time.TestTime ‑ test_add_datetime[to_timedelta-True-True-True]
weldx.tests.test_time.TestTime ‑ test_add_timedelta[to_timedelta-h-False-False-False]
weldx.tests.test_time.TestTime ‑ test_add_timedelta[to_timedelta-h-False-False-True]
…

♻️ This comment has been updated with latest results.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.48%. Comparing base (1750f5d) to head (2e6c519).
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #918      +/-   ##
==========================================
+ Coverage   96.45%   96.48%   +0.03%     
==========================================
  Files          95       95              
  Lines        6287     6293       +6     
==========================================
+ Hits         6064     6072       +8     
+ Misses        223      221       -2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[CagtayFabry]

Thanks for the fixes @marscher !

what is the minimum pandas version required for this?
Does it work with 1.5.x still?

[marscher]

what is the minimum pandas version required for this? Does it work with 1.5.x still?

I've added some backward compatibility code for reading old unit suffixes. Currently it only replaces the (used by weldx schema examples) suffix for seconds "S". I am not aware of any other units used in the past by weldx. To ensure we can still load old files with the new version, we should probably complete this list.

[marscher]

backward compat @CagtayFabry

[marscher]

changelog of pandas:

Units ‘H’, ‘T’, ‘S’, ‘L’, ‘U’ and ‘N’ are deprecated and will be removed in a future version. Please use ‘h’, ‘min’, ‘s’, ‘ms’, ‘us’, and ‘ns’ instead of ‘H’, ‘T’, ‘S’, ‘L’, ‘U’ and ‘N’.

[marscher]

Sorry I misinterpreted your question. I did not check, if it works with old Pandas versions. Is this a requirement or can we just pin a minimum version?

[marscher]

I've tested pandas-1.5 with Python-3.11 locally and all tests pass.

[marscher]

The CI failures are unrelated to my changes. I think this one is ready to go.

[CagtayFabry]

I've tested pandas-1.5 with Python-3.11 locally and all tests pass.

If you tested it with 1.5 I am fine with pinning minimum pandas version >=1.5. 😊
I will update the CI to test for that later as well