[dpdev69]

Requested by Boss

What and why

Before: the Build workflow on main has been red since #19. ControllerTests.testLinkUpAndWirelessStatusTogetherReconfigureOnce failed with 8 reporting writes instead of 4, although nothing in #19 touched UprightKit.

After: the test only measures what it is about, and the suite passes.

The test checks that a wake burst (the receiver's 0x41 link-up plus the mouse's 0x1D4B reconnect) leads to one reconfiguration. But the retry poll armed when the link dropped kept running with 0.1 to 0.2 s delays. On a slow runner a poll could fire after the mouse woke, reconfigure it on its own (4 writes), and then the debounced link-up reconfiguration ran as well (4 more). The fix parks the retry poll in this test, as the neighbouring wake tests already do, so the burst is the only thing that can reconfigure. No sleep was lengthened and nothing is skipped.

How I tested it

• swift test in UprightKit passes twice in a row: 395 tests, 0 failures each run
• make app builds (no app or project changes)
• Simulator only (no hardware). Reproduced the race on the old test by delaying the wake burst 0.15 s so the retry poll fires first: it fails (6 writes instead of 4). With the fix the same forced race passes 5 of 5. Plain runs under full CPU load passed 40 of 40, which is why CI only catches it sometimes.
• macOS version: 27.0.1

Checklist

• Code outside Platform/ uses Foundation only (test-only change)
• Every test method in a @MainActor test class is async
• Added a line under Unreleased in CHANGELOG.md (test-only, no user-facing change)

🤖 Generated with Claude Code