Spack uses the Clingo ASP solver from the POTASSCO group for dependency resolution. Clingo can provide unsatisfiable cores, which are useful for determining causality of errors, but unsatisfiable cores from Clingo are not guaranteed to be minimal and can be too large to be useful. Spack uses the prioritization guarantees in Clingo to instead deduce “error” facts as valid states, and prioritize against them in the solve. Treating error conditions as highly unoptimal successful solves, Spack uses its generalized expression of conditions in the logic program to create causal chains of facts, which can be traced from error fact back through all intermediate states to the user conditions that triggered it. By triggering full causal explainability only when requested, Spack bypasses the small but unpleasant performance penalties of these rules in the logic program in the default case, allowing both fast dependency resolution in the common case and explainability of errors.