acceptancecoverage

command
v0.2.14 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Nov 16, 2026 License: MIT Imports: 12 Imported by: 0

Documentation

Overview

Package main derives the acceptance coverage table from the catalogs and the driver, rather than from a list written beside them.

PA-01 of the release acceptance protocol asks for a table naming every required cell, the scenarios it executed, and the evidence that it passed. Every part of that except the evidence is already written down somewhere in this repository: the supported Kubernetes minors in support/kubernetes.json, the suites and their phases in support/e2e-suites.json, the Go phase each phase runs and the engine it is handed in hack/e2e-kind.sh, and the scenarios inside a phase in its declaration in test/e2e/phases.

A table typed out by hand from those sources is one more copy, and the first one to go stale. This reads them.

The acceptance record #242 asks for, with everything the repository can answer already answered.

The record was a blank page, and a blank page is why every requirement reads "Not assessed" whether or not anyone has looked. Most of the candidate's identity is in the tree at a commit: the schema and controller-state contracts, the Ptah build and its runner protocol, the supported minors, the suites and what each covers. Typing those by hand into a record is how they come to disagree with the release they describe.

What a build or a deployment decides is left blank and named, because an unfilled capacity or recovery target leaves acceptance incomplete and a record that hides the blank reads as complete.

No requirement is marked passed here. A closed issue, a merged fix and a green run against a pull request are none of them evidence about a candidate; the issue says so, and this prints dispositions rather than awarding them.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL