Skip to content

v0.6.8

Choose a tag to compare

View all tags
@bbingz bbingz released this 07 May 07:44
· 112 commits to main since this release
v0.6.8
d81a239

polycli v0.6.8

Adds background-worker run-ledger plumbing and a read-only terminal TUI inspector on top of v0.6.7. No provider runtime semantics or upstream CLI behavior changed.

What changed

Background-job ledger plumbing (Q6 Spec 2)

  • The parent process persists a top-level runContext (runId / command / hostSurface / argv / jobId / provider / kind / model / defaultModel / logFile) into the per-job config when --run-id (or POLYCLI_RUN_ID) is in scope. After spawning the worker, the parent writes one job_started event.
  • _job-worker reads runContext and writes attempt_started before runProviderPromptStreaming. After the job write succeeds, the worker writes attempt_result (status completed / failed) plus provider_decision (adopted on ok, failed reason=<kind>_failed on not-ok). Worker-observed cancellation produces attempt_result status=cancelled + provider_decision status=cancelled reason=job_cancelled.
  • New shared writer recordRunEventForContext(workspaceRoot, runContext, base); existing recordRunEvent delegates via buildCurrentRunContext(). Worker code never mutates the parent-side RUN_CONTEXT global.
  • createRunLedgerEvent schema gains nullable pid / durationMs slots; foreground events round-trip with the existing fields and add null defaults for the new ones.
  • Tests: 3 new background integration tests (success with --run-id, failed cmd ask without full-prompt leakage, explicit POLYCLI_HOST_SURFACE=codex-skill propagation).
  • Killed-worker (kill -9 after provider returns but before the ledger write) perfect recovery is open ledger-side hardening (reaper or scan-on-read step), not a TUI gate.

Read-only TUI inspector (Q6 Spec 3)

  • New terminal-only polycli tui command, routed through the existing @bbingz/polycli wrapper. It does not run, retry, cancel, or mutate provider jobs; it only renders persisted run-ledger / debug data.
  • Pure view-model layer (packages/polycli-terminal/lib/tui/view-model.mjs) with classifyProviderStates, formatReproductionCommand, truncateMiddle, applyKey, buildTuiModel, renderTuiFrame — testable without a real TTY.
  • Real navigation: up/down/k/j move run selection (clamped at the ends), enter opens the selected run's detail view, b returns to list, tab cycles the focused pane (runs→providers→events→repro), ? toggles a Help line, r refreshes from the ledger, q / Ctrl-C quit. Footer only names keys actually wired.
  • Selection change reloads the selected run's debug show + debug explain so the matrix / timeline / detail panes always reflect the active run, not the initial one. r refresh preserves the current selection unless the run disappears from the index.
  • Raw-mode safety: interactive() enters raw mode under a try / finally guard with an idempotent restoreRawMode (also hooked into SIGINT and exit). Initial-load failure and refresh-load failure both restore raw mode before exiting; refresh failures show an error frame and stay in the loop until q.
  • --history <count> validates a non-negative integer and slices the runs index in the TUI (companion debug runs contract unchanged). --run-id <id> selects an explicit run.
  • TUI files ship in the @bbingz/polycli tarball: bin/polycli-tui.mjs, lib/tui/view-model.mjs. New packaging assertion in scripts/tests/open-source-packaging.test.mjs.
  • Jobs with started or attempt_started but no terminal attempt_result / provider_decision render as unfinished / unknown. Killed-worker perfect recovery is the open ledger-side follow-up; it is not a TUI gate.
  • Smoke-only --script-keys "<k1,k2,...>" test hook drives the runtime through key sequences (down, down,enter, etc.) so selection-change-reloads-detail and other key transitions are testable without a real TTY.

Run-ledger debug examples docs

  • docs/polycli-v1-public-surface.md adds a "Run ledger debug examples" section that walks through the original Q6 narrative end-to-end:
    • cmd health passed, but two ask attempts failed → not adopted.
    • pi health failed → skipped before any prompt-bearing work.
  • Examples reference event-schema slots (provider_decision, health_result, reason: ask_failed / health_failed) and polycli debug runs / show / explain, not invented live provider output.

Manual smoke (v0.6.8 prep)

Automated smoke against the real node packages/polycli-terminal/bin/polycli.mjs tui ... binary (via --smoke and --script-keys) on 2026-05-07:

  • ✅ TUI renders a frame from a real ledger fixture (view:list pane:runs, run list, provider matrix, footer).
  • ✅ down reloads run-b's detail (matrix flips from qwen adopted to pi skipped health_failed).
  • ✅ down,enter enters detail view and surfaces run-b's explanation block (view:detail, explanation, pi skipped (health_failed)).
  • ✅ down,enter,b returns to view:list.
  • ✅ ? toggles the Help: line on; the persistent footer stays.
  • ✅ tab,tab cycles focused pane to events.
  • ✅ --history=1 renders only one run.
  • ✅ Non-TTY interactive entry exits 1 with the requires an interactive TTY error and never enters raw mode.

User-side TTY verification still required before tagging / publishing v0.6.8 (these paths are interactive-only and cannot be exercised from a non-TTY harness):

  • polycli tui opens in a real terminal and q quits without leaving the terminal in raw mode.
  • r refresh in interactive mode re-reads the ledger and re-renders without crashing.
  • Visible cursor / colour state restored after q.

Verification targets

  • node --test scripts/tests/terminal-tui.test.mjs scripts/tests/open-source-packaging.test.mjs
  • npm test
  • npm run release:check

Publish notes

This release adds no new npm package. Same 6 release artifacts as v0.6.7:

  • GitHub release v0.6.8
  • npm @bbingz/polycli-opencode@0.6.8
  • npm @bbingz/polycli@0.6.8
  • Utility packages stay on independent v1.x cadence (@bbingz/polycli-utils@1.0.1, @bbingz/polycli-timing@1.0.1).

See docs/release.md for the full sequence.

Assets 2
Loading