28 lines
1.9 KiB
Markdown
28 lines
1.9 KiB
Markdown
# Real device test plan / 单次实机测试计划
|
||
|
||
**Not executed. Explicit owner authorization is required before any write.**
|
||
只读核对完成后询问：“真实设备只读验收完成，是否允许进行一次真实图片上传与显示测试？”
|
||
|
||
1. GET Frame Flow device (driver POST /device); record current imageId AND coverId,
|
||
synced, geometry, model, rotation, timezone, firmware, heartbeat and battery.
|
||
2. POST /listImages for metadata only. Record count, current image, sorted IDs,
|
||
duplicates, name/date/dimensions anomalies. No download-all action.
|
||
3. Stop on discrepancy. Missing model fields must be reported as unavailable.
|
||
4. Once explicitly authorized, locally render **one** 1200×1600 test image containing
|
||
PIGWAY FRAME FLOW TEST and Shanghai timestamp, no private content.
|
||
5. Record an operation ID and previous image; one upload only, no write retries.
|
||
6. Poll device read-only no more often than 60 seconds, respecting the 20/min budget.
|
||
Confirm imageId and synced. If sent but response lost, mark UNKNOWN, block writes,
|
||
reconcile gallery/device; never upload again to guess the outcome.
|
||
7. Keep the test image visible at most 1–3 minutes, then restore the recorded artwork
|
||
with a separate tracked switch operation. Query once more to confirm restored
|
||
imageId and synced. Record coverId after restoration too.
|
||
8. Delete only that known test image after restoration is positively confirmed AND
|
||
it is not a base/previous/override/current reference. If ambiguous, leave it and
|
||
report. Never delete art or retry a timed-out delete.
|
||
9. Disable real writes again. Report each operation ID/result and remaining test image.
|
||
|
||
The API cannot set coverId. Restoration guarantees a requested image switch when
|
||
confirmed, not arbitrary cover metadata restoration. Physical screen refresh has
|
||
no documented pixel telemetry; visual owner observation supplements cloud synced.
|