MuellerSeb / nml-tools / 37644682602

87%

main: 87%

Pull #74

github

Preserve explicit value locations for dimension errors

Track the last explicit non-null value span separately from the latest consumed input span. Use that location for dimension positivity errors so trailing null assignments cannot move the diagnostic away from the retained invalid value.

Add regression coverage for negative and zero source values followed by null input, later explicit replacements, and default-only state. Ruff, mypy, and the full pytest suite pass (375 tests).

181 of 186 new or added lines in 3 files covered. (97.31%)

5205 of 5972 relevant lines covered (87.16%)

0.87 hits per line