Expand description
Stage 2 of backend benchmarking for the blob kind: precision and per-operation statistics.
The same shape as persistant_kv::stats. No timing (stage 3) and no ranking.
measure runs each leaf’s test runs times, one attempt each, and reports how often it passed,
how often the backend broke the contract and how often it errored, with the first failure’s text.
Instrumented counts, per operation, the calls and the errors by kind. A leaf is exact only
if every run passed; a fraction is never rounded up, and the contract (stage 1) still decides
eligibility.
Structs§
- Instrumented
- A
BlobBackendthat counts what the wrapped backend does, per operation. It changes nothing. - Leaf
Stats - The result of running one leaf’s test
runstimes on a target. - OpCount
- What one operation did while a test ran: calls, successes and errors by kind.
Functions§
- measure
- Run each leaf’s test
runstimes ontarget(one attempt per run) and count.