dascore/utils/plotting.py (1)

68-76: Robust handling of NaN/NaT values in coordinates.

The use of nanmin/nanmax with a fallback to index-based extents correctly handles the edge case of all-NaN/all-NaT coordinate arrays, preventing crashes as intended for issue #534.

Minor consideration: The fallback values (0, len(array)-1) are integers, which may cause minor type inconsistency when the original array is datetime64 or timedelta64. While the downstream conversion functions (_convert_datetimes, _convert_timedeltas) can handle integers, you might consider dtype-aware fallbacks (e.g., np.datetime64(0, 'ns') for datetime arrays) for more explicit type preservation. However, given this is an edge case and the current approach works, this is purely optional.

tests/test_viz/test_waterfall.py (1)

183-192: Test validates the fix for incomplete time coordinates.

The test successfully creates a scenario with aggregated time coordinates (NaN/NaT values) and verifies that waterfall() no longer crashes, addressing issue #534.

Consider optionally enhancing test coverage by:

• Asserting that the plot extents fall back to index-based values (0 to N-1) when coordinates are all NaN
• Verifying that the axis labels or other plot properties are correctly set despite missing coordinate data

However, the current test adequately validates the primary requirement (no crash with incomplete coordinates).

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro