Expand description
Staged evaluation of a compiled plan (spec 002, 3.7 and 3.10).
start validates the snapshot at a supplied evaluation time and computes
every exact step it can. pending lists semantic requests; supply
accepts one output or runtime failure per request; finish returns the
judgment and its evidence record. Nothing here performs I/O, reads a
clock or calls a backend: the caller (a runtime, or a test) does.
Structs§
- Evaluation
- An evaluation in progress.
- Failure
Context - What the runtime knows about a failed request’s backend (spec 033, 3.9):
who registered it and the determinism its descriptor declares. Recorded
as the
backendcause’s origin and used for transience (3.6.4); plans withoutactionabilityignore it, so their records keep their bytes. - NotReady
finishwas called while requests are pending.- Pending
Acquisition - A single pending page, with the exact identity computed by the core.
- Semantic
Request - One semantic request for a runtime to execute.
Enums§
- Request
Identity Error - Why a request identity was not computed (spec 004, 3.4.2).
- Start
Error - Why evaluation could not start.
- Supplied
- What a runtime reports for a request.
- Supply
Error - Why a supplied output was not accepted.