Scanning list: RSSI for all cars, stable order, spinner, auto-stop by dzid26 · Pull Request #100 · opencartools/TeslaBatteryBLE

dzid26 · 2026-10-06T21:46:05Z

Summary

Part 1 of #98, the scan-list behavior. Owner spec items:

Unpaired cars show their advertised RSSI. The scan advert already carries it, so no connection is needed.
The list sorts by name (known cars first, then discovered ones) instead of by RSSI, so signal jitter can't shuffle the rows.
A single spinner in the status row replaces the Scanning: N Tesla(s) text, and the Stop button is gone. The pull pill now appears only while you're actually pulling.
A scan ends by itself 30 s after the last new car appears; each new car resets the timer.
Unpaired cars leave the list once they stop advertising, and disconnected connections are no longer carried over. List rows no longer show a Pair button; pairing is one tap away in the car view.
The log card is capped at a share of the list height.

Also fixed here: Back from Settings now returns to the cars list instead of reopening the last car. MainScreen left the composition while Settings was open, so its saved list/car selection was lost and the restore-last-car step ran again. A SaveableStateHolder in MainActivity keeps that state. This bug is also why the light-mode scan screenshot kept failing: the capture landed on the car view.

Part 2 (connect once to read the GATT name, cache it and close; reopen on tap; close on back) comes in a follow-up PR.

Type

Testing

./gradlew :core:test (CI)
./gradlew :app:testDebugUnitTest (CI)
./gradlew :app:lintDebug (CI)
Real car (describe what was tested) — not yet

Demo-car screenshots come from the PR screenshots workflow.

Checklist

Protocol changes ship test vectors; expected.txt regenerated, not hand-edited
No VINs, keys, or decrypted payloads in logs
Docs/master-plan updated where relevant
New source files carry SPDX-License-Identifier: AGPL-3.0-only

🤖 Generated with Claude Code

https://claude.ai/code/session_0129RzhbEE2cboMcYyZMuxqX

github-actions · 2026-10-06T21:53:21Z

Screenshots

Captured from the demo build at ccdc0fe.

dzid26 · 2026-10-06T22:03:59Z

Screenshots disagree. - Analyze them.

Part 1 of #98. Unpaired cars show their advertised RSSI (the scan advert already carries it, no connection needed); the list sorts by name so RSSI jitter cannot shuffle rows; a spinner replaces the scanning text and the Stop button; a scan ends by itself 30 s after the last new car; and unpaired cars leave the list once they stop advertising (disconnected connections are no longer carried over). Name resolution and the unpaired connection lifecycle follow in part 2.

Take 02-scanning.png after the car flow, when the list shows the known car, so it matches 02-scanning-dark.png. The early capture showed raw pair addresses and a pull pill; the address wait stays as a gate.

github-actions · 2026-10-06T23:51:06Z

Screenshots

Captured from the demo build at 9114d53.

dzid26 · 2026-10-07T01:41:19Z

analyze screenshot, double spinners, Scanning... label on the spinner still shows up. New discovery - there is unnecessary pair button on Cars list car now.

github-actions · 2026-10-07T01:47:17Z

Screenshots

Captured from the demo build at e329c68.

Owner review of the #100 sheet: the status-row spinner plus the pull pill showed two spinners, the pill kept its Scanning label, and the rows carried a Pair button. The pill now only appears while pulling and takes nullable labels (null on the list, kept on the car view); the list rows drop Pair entirely - pairing stays one tap away in the car view.

github-actions · 2026-10-07T02:01:04Z

Analysis of the sheet at e329c68, item by item:

Double spinners: the new small spinner in the status row plus the pull-to-refresh pill, which hangs around for the whole scan because it follows state.scanning.
"Scanning…" label: that is the pull pill's refreshing label, same cause.
Pair button on the list row: the unpaired S1a2 row still shows it.

Fixed in 76ba99c: the pill now appears only while the finger is pulling and takes a null label on the list (spinner-only; the car view keeps its labels), the status row keeps the single spinner, and the list rows drop Pair entirely - pairing stays one tap away in the car view. New sheet coming from CI.

github-actions · 2026-10-07T02:10:53Z

Screenshots

Captured from the demo build at 76ba99c.

The drag-fraction gate alone did not hold on the phone: the pill lingered after release, so a scan still showed two spinners. Gate it on no scan running as well; the pill now only appears mid-drag before a scan starts.

github-actions · 2026-10-07T02:19:09Z

Screenshots

Captured from the demo build at 2384477.

The spinner replaced the cars/connected string; keep the text as-is and show the spinner at the row's end while scanning or discovering.

github-actions · 2026-10-07T02:29:21Z

Screenshots

Captured from the demo build at f0a46f3.

In landscape the fixed 200dp log content left no room for the cars. The card now takes 30 percent of the available height (at least 96dp) with internal scrolling.

With auto-stop the deferred shot caught a static list. Pull once more before capturing so the spinner shows with the known car, matching the dark pass.

github-actions · 2026-10-07T02:40:54Z

Screenshots

Captured from the demo build at 88d25bd.

wait_for_text Demo Tesla also matches the car view and settings, so a missed BACK captured the wrong screen as 02-scanning.png. Require Cars too, which only the list shows; otherwise leave a scan-known debug dump.

The light-pass Settings step used a bare BACK, which can land on the car view and capture it as 05-settings.png. back_to_list now follows BACK with the car view's own All-cars button when needed, so the step starts from the list.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0129RzhbEE2cboMcYyZMuxqX

MainScreen left the composition while Settings was open, so its saved list/car selection was dropped and Back re-ran the restore-last-car step. A SaveableStateHolder keeps that state across the round trip. This is also why the light-pass scan shot failed: Back from Settings landed on the car view instead of the list. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0129RzhbEE2cboMcYyZMuxqX

github-actions · 2026-10-07T20:42:12Z

Screenshots

Captured from the demo build at 9c2b951.

dzid26 marked this pull request as draft October 6, 2026 22:04

dzid26 added 2 commits October 7, 2026 00:06

Defer the light scan shot until the car is known

9114d53

Take 02-scanning.png after the car flow, when the list shows the known car, so it matches 02-scanning-dark.png. The early capture showed raw pair addresses and a pull pill; the address wait stays as a gate.

dzid26 force-pushed the work/scan-list branch from ccdc0fe to 9114d53 Compare October 6, 2026 23:11

dzid26 marked this pull request as ready for review October 6, 2026 23:44

Merge branch 'main' into work/scan-list

e329c68

dzid26 marked this pull request as draft October 7, 2026 01:49

dzid26 marked this pull request as ready for review October 7, 2026 02:04

dzid26 marked this pull request as draft October 7, 2026 02:10

Hide the pull pill whenever a scan is running

2384477

The drag-fraction gate alone did not hold on the phone: the pill lingered after release, so a scan still showed two spinners. Gate it on no scan running as well; the pill now only appears mid-drag before a scan starts.

dzid26 marked this pull request as ready for review October 7, 2026 02:13

dzid26 marked this pull request as draft October 7, 2026 02:19

Keep the counts text with the spinner docked right

f0a46f3

The spinner replaced the cars/connected string; keep the text as-is and show the spinner at the row's end while scanning or discovering.

dzid26 marked this pull request as ready for review October 7, 2026 02:22

dzid26 marked this pull request as draft October 7, 2026 02:25

dzid26 added 2 commits October 7, 2026 03:33

Cap the log card at a share of the list height

a273374

In landscape the fixed 200dp log content left no room for the cars. The card now takes 30 percent of the available height (at least 96dp) with internal scrolling.

Rescan before the light scan shot

88d25bd

With auto-stop the deferred shot caught a static list. Pull once more before capturing so the spinner shows with the known car, matching the dark pass.

dzid26 marked this pull request as ready for review October 7, 2026 02:34

dzid26 marked this pull request as draft October 7, 2026 14:41

Gate the scan shot on the cars-list title

39df737

wait_for_text Demo Tesla also matches the car view and settings, so a missed BACK captured the wrong screen as 02-scanning.png. Require Cars too, which only the list shows; otherwise leave a scan-known debug dump.

dzid26 marked this pull request as ready for review October 7, 2026 14:46

Return to the list before the light Settings step

6cdf27f

The light-pass Settings step used a bare BACK, which can land on the car view and capture it as 05-settings.png. back_to_list now follows BACK with the car view's own All-cars button when needed, so the step starts from the list.

This was referenced Oct 7, 2026

CI: replace the emulator screenshot pipeline with JVM screenshot tests #125

Closed

Architecture review 2026-10-07: findings index and decision audit #109

Open

claude added 2 commits October 7, 2026 20:29

Merge branch 'main' into work/scan-list

086e092

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0129RzhbEE2cboMcYyZMuxqX

dzid26 merged commit d6974e1 into main Oct 7, 2026
4 checks passed

dzid26 deleted the work/scan-list branch October 7, 2026 20:47

dzid26 mentioned this pull request Oct 7, 2026

Dismiss system ANR dialogs during the emulator screenshot capture #137

Merged

2 of 13 tasks

Navigation Menu

Repository navigation

Scanning list: RSSI for all cars, stable order, spinner, auto-stop - #100

Scanning list: RSSI for all cars, stable order, spinner, auto-stop#100
dzid26 merged 12 commits into
mainfrom
work/scan-list

dzid26 commented Oct 6, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 6, 2026

Uh oh!

dzid26 commented Oct 6, 2026

Uh oh!

github-actions Bot commented Oct 6, 2026

Uh oh!

dzid26 commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

dzid26 commented Oct 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Type

Testing

Checklist

Uh oh!

github-actions Bot commented Oct 6, 2026

Screenshots

Uh oh!

dzid26 commented Oct 6, 2026

Uh oh!

github-actions Bot commented Oct 6, 2026

Screenshots

Uh oh!

dzid26 commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Screenshots

Uh oh!

github-actions Bot commented Oct 7, 2026

Uh oh!

github-actions Bot commented Oct 7, 2026

Screenshots

Uh oh!

github-actions Bot commented Oct 7, 2026

Screenshots

Uh oh!

github-actions Bot commented Oct 7, 2026

Screenshots

Uh oh!

github-actions Bot commented Oct 7, 2026

Screenshots

Uh oh!

github-actions Bot commented Oct 7, 2026

Screenshots

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

dzid26 commented Oct 6, 2026 •

edited

Loading