Skip to content

BACKLOG.md

Latest commit

 

History

History
39 lines (32 loc) · 4.89 KB

BACKLOG.md

File metadata and controls

39 lines (32 loc) · 4.89 KB
Raw

HeadMatch Backlog (autonomy-processed)

The vornik daemon (project headmatch, mode: backlog) consumes the first - [ ] item each tick, runs it through the backlog-item workflow (analyze → implement → test → review → draft PR), then stamps the line - [x] … (task: <id>) on success or - [!] on failure. - [?] items are proposed and stay inert until an operator promotes them to - [ ] — the per-item human gate. Deposited findings also land as - [?].

Keep each item a single, self-contained line: the whole line becomes the task prompt, and the agent works only that item ("smallest correct fix with tests"). The narrative capabilities + release history live in docs/backlog.md.

Ready

  • TASK-113: Document the confidence-scoring derivation — add docstrings and inline rationale for the magic numbers in pipeline_confidence.py, plus a short note under docs/. No behaviour change; add a test that pins the current scores so the documentation cannot silently drift. (task: task_20260712004844_3bf3db60b5a334dd)
  • TASK-112: Add a coverage CI gate — fail CI when total coverage drops below the current 80% level, wired into the existing GitHub Actions matrix without breaking the Python 3.10–3.13 lanes. (task: task_20260712021844_d6e14940dba6130b)
  • TASK-114: Harden fetch_curve_from_url against SSRF — validate the URL scheme and host against a domain allowlist (the known measurement databases) before fetching, and reject private / loopback / link-local targets. Add tests for allowed and rejected URLs. (task: task_20260712034844_5c385e9f59a31f3c)
  • TASK-110: Standardize the error hierarchy — define a HeadMatchError base class with MeasurementError, ConfigError, and NetworkError subclasses and migrate the existing raise sites to them; keep messages intact and add tests asserting the new types are raised. (task: task_20260712051844_0ef3ce20f1615424)
  • TASK-109: Decompose gui/shell.py — extract GuiState into its own module and break out the Tkinter variable initialization, leaving gui/shell.py a thin composition. No UI behaviour change; keep the existing GUI tests green. (task: task_20260712064844_7720364a5e21b289)
  • TASK-106: Split gui_views.py into real per-view modules (one module per view) with a package __init__ re-export so imports stay stable. No behaviour change; tests stay green. (task: task_20260712112832_f730dad6a32fdbf2)
  • TASK-107: Extract the GUI workflow controllers out of HeadMatchGuiApp into dedicated controller classes, leaving HeadMatchGuiApp as wiring. No behaviour change; tests stay green. (task: task_20260712094844_55b5735c44ba96df)
  • TASK-108: Centralize GUI file-picking and background-task helpers into a shared module and route the existing call sites through it. No behaviour change; tests stay green. (task: task_20260713103345_3e494e46d309d871)

Proposed — promote to - [ ] when ready (some may already be shipped; vet first)

  • [?] TASK-115: Async audio backend — replace time.sleep() synchronization with asyncio-based process management across the audio backends. Large refactor; wants a design pass before it is a single task.
  • [?] Dense GraphicEQ clipping — validate the added 1.5 dB headroom holds in real-world use (may still clip). Needs real measurement data, not a pure code change.
  • [?] macOS real-hardware end-to-end testing — CI passes but the PortAudio path needs manual validation on real hardware (not autonomously doable).
  • [?] Mic-calibration workflow (research) — derive a mic response curve via trusted-data comparison; open questions on reliable databases, ear-canal resonance variation, and per-user tractability.
  • [?] Packaging — macOS .app bundle, Windows .exe (PyInstaller), and Linux AppImage wrappers (currently shipped as raw binaries).
  • [?] Target-editor polish — keyboard shortcuts plus undo/redo.
  • [?] Cache fixed-profile basis responses for repeated GraphicEQ runs.
  • [?] Additional export formats beyond CamillaDSP and Equalizer APO.
  • [?] CamillaDSP live-update via its WebSocket API.
  • [?] Closed-loop EQ refinement (measure → apply → re-measure).
  • [?] [refactor] Update test_gui.py to use WorkflowControllers for basic_search_target and choose_basic_search_match — The tests test_basic_search_with_multiple_matches_requires_choice and test_basic_search_single_match_downloads_and_selects_csv directly call HeadMatchGuiApp methods with a SimpleNamespace mock. After moving these methods to WorkflowControllers and adding delegation in shell.py, these tests now need to use WorkflowControllers instead of calling classmethods on HeadMatchGuiApp. (evidence: tests/test_gui.py:293-306 for multiple matches, tests/test_gui.py:323-344 for single match. These tests use SimpleNamespace without _controllers attribute.; via task_20260712095457_5b6438f138dd1e3c, 2026-07-12)