implementation-test-pipeline

# Implementation Test Pipeline Own the transition from the selected Go command and raw streams to bounded, occurrence-aware results and durable child status. Keep presentation and coverage algorithms with their owners. Parsed events may explain execution; they cannot rewrite its outcome. ## Choose the affected boundary | Task | Required detail | | --- | --- | | Child argv, capture, pipes, cancellation, process-tree cleanup, helper synchronization | Read [execution and cancellation](references/execution.md). | | Framing, union decoding, budgets, repeated tests, benchmark assembly, metadata or durations | Read [protocol and occurrence results](references/protocol-results.md). | | Durable status, run/report exit policy, external evidence import or coverage-policy projection | Read [bound status and offline reporting](references/status-and-offline.md). | Read multiple references when a change crosses those boundaries. Inspect the linked implementation and actual assertions before describing conformance. ## Preserve the lifecycle 1. Accept executable, workdir, environment policy, and argv separately. Prepare secure raw destinations and cancellation ownership before starting. 2. Persist stdout and stderr before bounded parsing or display. Keep raw bytes complete when downstream output, entry, or normalized-string budgets fill. 3. Aggregate by package, test, and occurrence. Keep metadata tied to source sequence and owning occurrence; keep malformed, missing, or contradictory semantic evidence explicit. 4. Drain and reap owned processes and streams. Finalize open state as incomplete; retain measured, estimated, and absent time distinctly. 5. Bind closed evidence files to authoritative status. Offline rendering must validate those bindings before it can claim success. 6. Preserve every contributing failure, then apply the documented exit precedence. Cancellation remains authoritative through report publication. ## Verification Select owner tests for the changed boundary, app tests for composition, and controlled CLI fixtures for end-to-end behavior. Native host tests are required to qualify OS-specific signaling and reaping. Preserve raw output alongside expected normalized state when reproducing malformed or fragmented records. Do not turn a documentation-only correction into a process execution audit. ## Acceptance scenarios - A passing run streams raw JSONL, produces measured results, and returns zero. - A normal `go test` failure still generates reports and returns the child's nonzero status. - A compile failure represented by text or `BuildEvent` appears in normalized build failures even without a completed test. - Two occurrences of the same package and test name have different ordinals and independent outcome/duration/output. - A malformed or oversized record remains in raw evidence, yields a bounded diagnostic, and leaves affected active state incomplete. - Output beyond the per-occurrence report bound is marked as truncated while the raw JSONL remains byte-for-byte complete. - A helper publishes a complete PID record atomically; partial observations cannot trigger parsing, leaked goroutines, or reused-PID signaling. - Cancellation terminates descendants, drains both streams, reaps the child, and returns cancellation even if rendering later fails. - A renderer failure after a child test failure is reported but does not replace the child failure status. - Duplicate status members or mismatched evidence bindings cannot establish offline success, even when every parsed test passed. - A policy-free no-coverage rerender removes only the coverage binding; a recorded coverage policy rejects that option before bundle mutation. - Entry/string exhaustion causes explicit incomplete evidence. Ordinary report output clipping reports truncation without inventing a semantic failure.