Expand description
Comparison of a finished live re-execution with its historical case (spec 026, 3.7 and 3.8).
This is offline evaluation over documents the host supplies: the
historical bundle (resolved through the host’s synchronous Resolver
at a caller-supplied time), the live run’s rustev.live-reexecution/1
document, its new run record and its plan. Nothing here dispatches a
backend, reads a clock or reaches the runtime: the live run already
happened, through the runtime’s separately named entry point.
The historical case is first verified by spec 004 reproduction. Then an
explicitly named proposal payload or outcome category, chosen by a
versioned task adapter, is compared. Full judgment bytes are not the
agreement metric, because decision, plan, snapshot and time may
legitimately differ. Changed output is live-diverged, never a failed
historical reproduction; unchanged output is agreement on the named
fields only. An expired, erased or otherwise unavailable historical item
makes the comparison unavailable with that reason; nothing is restored
from a digest, and no source document is modified.
Structs§
- Live
Case - One finished live re-execution and the historical case it links to.
Constants§
- EXCLUSIONS
- What every live comparison leaves out.
- LIMITATIONS
- What every live comparison states it does not establish.
Functions§
- compare_
live - Compare a finished live re-execution with its historical case. Returns the comparison and what was observed of the historical case’s retention.
- with_
comparison - A copy of
livecarrying the comparison and the observed retention. The historical sources andliveitself are left as they were.