mirror of
https://github.com/kuhyx/testsAndMisc.git
synced 2026-11-10 20:07:04 +01:00
Krzysztof kuhy Rudnicki
6832805e5d
Some checks are pending
dependency freshness / dependencies on newest stable (push) Waiting to run
md-naming / markdown naming convention (push) Waiting to run
Pre-commit checks / pre-commit (push) Waiting to run
Python tests / test (push) Waiting to run
Shell tests / shell-tests (push) Waiting to run
Shell tests / arch-tests (push) Waiting to run
Android runs one instrumentation per package: a second `am instrument`
force-stops the first ("stop com.kuhy.a11ydump due to start instr"), and
so does `install -r`. The live two-session run on the Pixel caught two
sessions' dumps killing each other ("Process crashed"). dump_display now
runs install + instrument under a per-serial flock; a dump takes well
under a second, so queueing costs nothing noticeable.
After the fix the same 300 s run had 0 failures across 659 actions, and
the real screen and both apps stayed where they were.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WpA6wW2DKiovQADTdjWJoH
37 lines
1.8 KiB
JSON
37 lines
1.8 KiB
JSON
{
|
|
"intent": "Two sessions reading their own virtual displays at the same moment must not kill each other's dump.",
|
|
"scope": [
|
|
"python_pkg/android_ui/_a11y_helper.py (lock), driver.py (passes the serial), tests, README"
|
|
],
|
|
"changes": [
|
|
"dump_display runs install + am instrument under a per-phone flock at ~/.cache/android_ui/a11ydump-<serial>.lock",
|
|
"AndroidUi passes its serial so different phones do not queue behind each other"
|
|
],
|
|
"verification": [
|
|
{
|
|
"command": "live two-session run on the Pixel 6a before the fix (fake owners claude:fakeA / claude:fakeB, 300 s)",
|
|
"result": "fail",
|
|
"evidence": "logcat: 'stop com.kuhy.a11ydump due to start instr' then 'Crash of app com.kuhy.a11ydump running instrumentation'; android_ui: 'a11y helper returned no tree: Process crashed'"
|
|
},
|
|
{
|
|
"command": "same run after the fix: /tmp/two_session/run.sh, DURATION=300",
|
|
"result": "pass",
|
|
"evidence": "A 219 element taps on Calculator (display 14), B 440 dp taps + dumps on Clock (display 13), 0 failures, empty stderr; 99 samples: display 0 stayed on Signal, 14 on Calculator, 13 on Clock; all five alarms still off"
|
|
},
|
|
{
|
|
"command": "raw adb from a Claude Bash call while claude:fakeA held pkg:__screen__",
|
|
"result": "pass",
|
|
"evidence": "`adb shell input -d 0 tap` blocked after 40 s naming claude:fakeA; `input keyevent 0` without -d blocked at once; `input -d 13 tap` refused as claude_fakeB's display"
|
|
},
|
|
{
|
|
"command": "pytest python_pkg/android_ui python_pkg/phone_guard --cov --cov-branch",
|
|
"result": "pass",
|
|
"evidence": "153 passed; every module 100% line and branch"
|
|
}
|
|
],
|
|
"risks": [
|
|
"The lock is per host: sessions on another machine sharing the phone over adb are not serialised."
|
|
],
|
|
"rollback": ["git revert the commit"]
|
|
}
|