Crate rustev_surface 

Rustev’s pure command surface (spec 027, 3.1): plan check, compile and show, offline replay, eval, gate, calibrate fit and calibrate qualify, and the result envelope of run, rendered exactly as the CLI renders them (spec 006). Every host (the CLI, the native binding, the WebAssembly package) calls these instead of a copy, so no surface drifts. Inputs and outputs cross a host::Host; this crate performs no I/O, reads no clock and runs no executor.

Modules

adapter

The declarative task adapter, rustev.task-adapter/1 (spec 006, 3.8): how evaluation reads labels and decides whether a proposal is correct, without writing Rust.

calibrate

calibrate fit | qualify (spec 006, 3.10), for every host.

deps

Reading a plan’s dependencies: backend descriptors, rules programs (whose descriptors the host derives, spec 005 3.9) and calibration artifacts.

eval

eval and gate (spec 006, 3.9), and the dataset and bundle handling calibrate fit shares, for every host.

explain

Pure explanation command shared by the bindings (spec 040, 3.15.2).

host

What a host supplies to a command (spec 027, 3.1): its inputs as bytes by label, bounded as the host’s transport is, and a place for the artifacts a command produces. The CLI’s labels are file paths and its host is the file system; a binding’s are names and memory.

out

Command output (spec 006, 3.1.2 and 3.2): one sorted, compact JSON object per command, with library records embedded as their record canonical bytes, and the exit code its status maps to.

plan

plan check | compile | show (spec 006, 3.4), for every host.

replay

Offline replay (spec 006, 3.7), for every host.

run

The result envelope of run (spec 006, 3.5): how a decision’s outcome is printed, whichever host ran it. Running the decision is the host’s (the runtime is not a dependency of this crate); the host maps its runtime’s results onto these.