-
cargo test --workspace --all-features --locked,cargo clippy --workspace --all-targets --all-features --locked -- -D warningsandcargo fmt --check(per package) pass - New or changed behaviour / the fixed bug has a unit test:
- No existing test removed, ignored or loosened
-
#[ignore]tests of the changed area run by hand:
- Device, iOS version, host (Linux / macOS):
-
tests/device/run.pydefault steps pass on this branch (exit code 0) - The same steps on
mainor the latest release, same device, for comparison; no step clearly slower, or why: - Steps exercising this change:
- Longer run (WDA start-up/recovery, reconnecting, unlocking, live picture, timers):
- Not needed: the change doesn't touch code that talks to a phone — why:
run.py output
this branch:
main / latest release:
- All cases still hold; device cases in the changed area re-run:
- A newly fixed problem is added as a row:
- Device, iOS version, host (Linux / macOS):
- Bug fix: reproduced on
main/ the release, gone with this branch, same device and steps - Feature: exercised end to end (MCP tool, CLI command or Viewer action), including the error and edge cases it handles
- Untested conditions (device kind, iOS version, orientation, host OS), if any:
- Not applicable: no bug fix or feature (docs, CI, refactoring without behaviour change) — why:
Steps and results (before / after, output, log lines, screenshots)
- MCP tool names, parameters and default outputs stay compatible (docs/decisions.md D10)
- New or changed dependencies are pinned to exact versions; vendor changes recorded in THIRD_PARTY.md
- User-visible changes noted in CHANGELOG.md (and CHANGELOG.zh-CN.md if you can) under Unreleased
- No private data in fixtures, screenshots, logs or the outputs above