Add python 3.11 and 3.12 versions to testing matrix #428
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## aggregated_filters_for_trials #428 +/- ##
=================================================================
+ Coverage 96.09% 96.10% +0.01%
=================================================================
Files 11 11
Lines 2200 2209 +9
=================================================================
+ Hits 2114 2123 +9
Misses 86 86 ☔ View full report in Codecov by Sentry. |
|
@mdeceglie - Regarding numpy > 2. We would need pandas 2.2.2 (https://pandas.pydata.org/docs/whatsnew/v2.2.2.html). However, the bug you found in pandas is still persistent in 2.2.2 (pandas-dev/pandas#55794). Not sure, how to proceed... |
…ions-to-testing-matrix
| @@ -531,6 +536,11 @@ def _combine_losses(self, rollup_period='M'): | |||
| The period on which to roll up losses and calculate availability. | |||
| """ | |||
|
|
|||
| # Allow pandas < 2.0 to use 'M' as an alias for MonthEnd | |||
| # https://pandas.pydata.org/docs/whatsnew/v2.2.0.html#deprecate-aliases-m-q-y-etc-in-favour-of-me-qe-ye-etc-for-offsets | |||
| if rollup_period == "ME": | |||
There was a problem hiding this comment.
Let's update the docstrings for _combine_losses and run to explain the situation. Maybe something like "Default value "ME" triggers monthly rollup period. For other aliases, be sure they are compatible with your version of Pandas"
There was a problem hiding this comment.
Good call. I've updated the docstrings.
| @@ -521,4 +519,4 @@ def test_hour_angle_filter(): | |||
|
|
|||
| # Check that the result is the correct boolean Series | |||
| expected_result = pd.Series([False, False, True, True, True], index=index) | |||
| pd.testing.assert_series_equal(result, expected_result) | |||
| pd.testing.assert_series_equal(result, expected_result, check_names=False) | |||
There was a problem hiding this comment.
What caused the need for this change?
There was a problem hiding this comment.
In the newest pvlib version 0.11.1 there was a bug fix for the hour_angle() function (https://pvlib-python.readthedocs.io/en/stable/whatsnew.html).
The series that's returned does have a new argument Name: equation_of_time.
In order to make the test run for older pvlib versions in which the name attribute is not present, I decided to not add the name to the expected result series but rather don't check for series names.
setup.py
Outdated
| "h5py >= 3.7.0", | ||
| "plotly>=4.0.0", | ||
| "xgboost >= 1.3.3", | ||
| "pvlib >= 0.9.0, <0.12.0", |
There was a problem hiding this comment.
requirments_min specifies 0.11. We should change the minimum here too right?
There was a problem hiding this comment.
Absolutely right, I overlooked that. I've bumped it to 0.11 now.
__init__.py