nix-result-checks tests both halves of a Nix project
— build-time (does the derivation actually produce what it should) and eval-time (is this pure Nix logic correct) —
under one report format and one CLI (nrc).
Build checks are derivations that always succeed,
capturing pass/fail as data (stdout/stderr/exitCode outputs) instead of propagating it,
so a failing check doesn’t stop the rest of the suite from building.
Eval checks are plain data, evaluated in parallel, with no derivation involved at all.
Both kinds reduce to the same entry shape, so nothing downstream has to special-case which kind of check it’s looking at.