If I recall correctly, we had a hard time getting numba>0.50.1 and the geospatial stack installed together from conda-forge. But, with issues in our current implementation arising with versions of numba below 0.50.0, we should try to encourage folks to upgrade their numba, otherwise we'll see breakage in local statistics for users with outdated numba.

Codecov Report

Merging #143 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #143   +/-   ##
=======================================
  Coverage   77.61%   77.61%           
=======================================
  Files          25       25           
  Lines        3274     3274           
=======================================
  Hits         2541     2541           
  Misses        733      733           

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update fd4323a...d74fe32. Read the comment docs.

Would <50 break? I think the only issue was that we could not pin random number generation inside numba-ized code so the tests results would vary and thus fail. Results would be correct though. No?

Results would fail to compute. If numba is installed and numba fails to compile the function, then the results are not computed.