[Manny7717]

What

fornav() resolves a fill value from the input data (NaN for floats, -999 for integers, or the caller's fill argument) but then passed hardcoded np.nan as both the input and output fill to the C resampler (_fornav.fornav_wrapper). Two consequences (reported in #689):

• Integer swath data cannot be resampled — NaN cannot be written into an integer output array.
• A user-specified fill (e.g. fill=0) was ignored for float data: uncovered pixels still got NaN.

Change

Pass the resolved fill value as both input and output fill to fornav_wrapper. Default behavior is unchanged (float data with fill=None still gets NaN, matching the documented default); only fill-specified and integer paths change. The dask EWA resampler already forwarded its fill value and is unaffected.

Tests

• New test_fornav_integer_swath_with_fill: int8 swath + fill=0 resamples without error; output contains only data and fill values.
• New test_fornav_fill_value_used_in_output: float32 swath + fill=0 — uncovered pixels are 0, no NaN anywhere.
• Both tests fail on the previous implementation (verified) and pass with the fix.
• pytest test_ewa_fornav.py test_dask_ewa.py → 89 passed, 20 skipped (pre-existing skips). ruff clean.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.90%. Comparing base (82b485c) to head (4a46786).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #741      +/-   ##
==========================================
+ Coverage   93.87%   93.90%   +0.03%     
==========================================
  Files          89       89              
  Lines       13888    13924      +36     
==========================================
+ Hits        13037    13075      +38     
+ Misses        851      849       -2     

Flag	Coverage Δ
unittests	93.90% <100.00%> (+0.03%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[djhoese]

Thanks for fixing this. The fix looks good. Could you please combine the two new tests outside of the unittest TestCase class and use pytest's parametrize? This will also require changing the assert method calls to normal assert X conditions.

[djhoese]

Thanks for this. I'm sorry it took so long to merge such an obvious change. Other EWA resampling PRs that were merged made some conflicts. Claude also pointed out to me that the integer fill value was out of the range for the integer dtype (int8, default was -999.0). I've tacked on a fix for that to this PR. I rebased your branch and force pushed it. This should be good to merge once CI finishes. Thanks again.