Module python_evidence 

Validation and normalization of the Python runtime’s evidence records.

Each Supercov-hooked interpreter publishes commit-framed JSON records into its own mmap: the process identity, every phase it entered (with the exact test identity that phase stands for), runner outcomes, hits, decision vectors and any measurement limitation the runtime detected. The one-byte commit marker is written last, so records completed before a hard kill remain readable while a torn tail stays inert.

Probes do not write records: they store bytes into their context’s slot, a small mapped file beside the transport, and the runtime harvests a slot into records when a phase ends. A process that dies between harvests leaves its slots behind; layout.json, written once per run, says which obligation or decision vector each byte stands for, and the slot’s bytes are read as the last records of its process. Rust joins those records into the shared frontend protocol; the runtime never computes a verdict.

Structs

PythonFrontendRun

Enums

PythonEvidenceError

Constants

PYTEST_RUNNER

PYTHON_EVIDENCE_VERSION

PYTHON_FRONTEND_VERSION

UNITTEST_RUNNER

Functions

build_python_frontend_run

Join the runtime’s evidence directory with the ahead-of-run manifest into a protocol-conformant frontend run.

python_coverage_model

Python is observed through probes compiled into each measured module as it is imported, from the obligations the plan derived from the source.