Expand description
Run-owned assertion map lifecycle and execution-backed, agent-assessed score.
Structs§
Constants§
Functions§
- affected_
tests - Which of a run’s tests the current checkout’s changes could have reached, from what each test executed and what has changed since, declaration by declaration. A test is affected by a change in code it ran, in its test file, or in a file it ran code in whose declarations changed shape; not by a change confined to code it never ran; not by comments or blank lines. A test that did not pass is listed as affected regardless: it has a result to establish. What this cannot see is a file the run never captured – a file added since – and the run’s dependencies and configuration, which the working-tree check answers for.
- assertion
- Read authored flows and their assessment from the same map snapshot.
- assess
- assess_
summary - The assessment’s summary and everything but its per-statement and per-line
lists.
runs latestshows five fields of the assessment and used to build all of it to get them: every measured statement, each carrying the source of the node around it. On a 300-file project that was 300,000 entries, a 177 MB cache, and 4 GB held to print a handful of counts. Every count is computed the same way here as in the full assessment; only the lists are skipped. - coverage
- executions
- Each test’s execution, placed in the declarations of the run’s own manifest: for every probe a test fired, the innermost unit holding it. Also which units hold a probe at all, per file, since only a change confined to such units can be said to have missed a test. Sources are needed to place JavaScript columns; without them there is no record.
- load
- load_
inputs - load_
manifest - report
- Cache derived assessments separately from immutable coverage evidence. Every query still verifies current source, map and managed-state identities.
- report_
summary - The report
runs latestshows: every count, none of the per-statement and per-line lists. Seeassess_summary.