Produce evidence that the checker detects a meaningful failure without rejecting legal controls.
Try these before opening the explanations. This is a self-assessment, not a stored score or certification.
The prior stalled offer. Retention is a relation across sampled edges, including the accepting edge. An accepted-beat-only log cannot establish that the offered payload remained stable.
Neither. The harness passed its expectation that the checker reject that mutation for the intended reason. Course/source/render gates need their own recorded evidence, and a bounded teaching model never establishes complete AXI or product compliance.
Ready to continue when you can produce:
A requirement/stimulus/checker/result matrix with nonzero work counts.
Legal controls, isolated counterexamples, clean repaired replay and recorded limitations.
Assertions are well suited to precisely sampled temporal and protocol properties; scoreboards model and correlate expected end-to-end results and ordering. Their scopes can overlap, and both depend on assumptions, observation points and reference-model correctness.
Common trap: Treating either mechanism alone as proof of system correctness.
An applicable requirement lacks adequate reviewed correctness evidence. A mapped test or hit coverage bin may still be insufficient if no checker observes the required result. Repair the evidence plan or record justified scope/risk disposition; a waiver does not verify the requirement.
Common trap: Treating a coverage bin or waiver as a passing correctness result.
The results cover different boundaries, configurations or scenarios. Compare actual traffic, map, reset, clocks, bridge transformations, firmware and reference-model assumptions. Those differences guide experiments; they do not establish the faulty component without discriminating evidence.
Common trap: Closing the bug as 'VIP issue' or assigning integration as the cause without tracing it.