scawful/yaze
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: update ROM test paths and envs

Browse Source
This commit is contained in:
scawful
2025-12-22 14:50:57 -05:00
parent 42ae359abc
commit 319e903f24
22 changed files with 102 additions and 73 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/internal/agents/cli-ux-proposals.md
View File

@@ -72,7 +72,7 @@ Board: docs/internal/agents/coordination-board.md (2025-12-01 ai-infra-architect
- Agent-side test commands (`agent test run/list/status/...`) print ad-hoc logs and lack `OutputFormatter`/metadata, making automation fragile; no JSON status, exit codes, or artifacts paths surfaced.
- Test helper tools (`tools-*/` commands, `tools/test_helpers/*`) mix stdout banners with file emission and manual path requirements; they are not discoverable via TUI or CommandRegistry-driven palettes and do not expose dry-run/plan outputs.
- TUI/editor have no test surface: no panel to run `stable/gui/rom_dependent/experimental` suites, inspect failing cases, or attach ROM paths/presets; quick actions and history are missing.
- Build/preset coupling is implicit—no guided flow to pick `mac-test/mac-ai/mac-dev`, enable ROM/AI flags, or attach `YAZE_TEST_ROM_PATH`; agents/humans can misconfigure and get empty test sets.
- Build/preset coupling is implicit—no guided flow to pick `mac-test/mac-ai/mac-dev`, enable ROM/AI flags, or attach `YAZE_TEST_ROM_VANILLA`; agents/humans can misconfigure and get empty test sets.
### Proposed Improvements
- **Unified test CLI/TUI API**

2
docs/internal/agents/tests-binary-ux-proposals.md
View File

@@ -22,7 +22,7 @@ Board: docs/internal/agents/coordination-board.md (2025-12-01 ai-infra-architect
## Improvement Proposals
- **Manifest & discovery**: Generate a JSON manifest (per build) listing tests, labels, requirements (ROM path, AI runtime), and expected artifacts. Expose via `yaze_test --export-manifest <path>` and `ctest -T mem`-style option. Agents can load it instead of scraping.
- **Structured output**: Add `--output-format {text,json,junit}` to `yaze_test` to emit summaries (pass/fail, duration, seed, artifacts) in one file; default to text for humans, JSON for automation. Wire ctest to collect the JSON and place it in a predictable directory.
- **Requirements gating**: On startup, detect missing ROM/AI/headless support and fail fast with actionable messages and suggested CMake flags/env (e.g., `YAZE_ENABLE_ROM_TESTS`, `YAZE_TEST_ROM_PATH`). Offer a `--mock-rom-ok` mode to downgrade ROM tests when a mock is acceptable.
- **Requirements gating**: On startup, detect missing ROM/AI/headless support and fail fast with actionable messages and suggested CMake flags/env (e.g., `YAZE_ENABLE_ROM_TESTS`, `YAZE_TEST_ROM_VANILLA`). Offer a `--mock-rom-ok` mode to downgrade ROM tests when a mock is acceptable.
- **Filters & subsets**: Provide a unified front-end flag set (`--label stable|gui|rom_dependent|experimental`, `--gtest_filter`, `--list`) that internally routes to gtest/labels so humans/agents don’t guess. Add `--shard <N/M>` for parallel runs.
- **Artifacts & logs**: Standardize artifact output location (`build/artifacts/tests/<run-id>/`) and name failing-test logs/screenshots accordingly. Emit paths in JSON output. Ensure GUI/agent recordings are captured when labels include `gui` or `experimental`.
- **Preset hints**: Print which CMake preset was used to build the binary and whether ROM/AI options are compiled in. Add `--recommend-preset` helper to suggest `mac-test/mac-ai/mac-dev` based on requested labels.