The pulse that meant it.

The slip was specific about the feeling it wanted.

Not “add a haptic” - the panels already had haptics. Not “make it stronger” - strength is not the variable. Long pulse. A single, sustained vibration instead of a quick notification buzz. The distinction matters because the two moments this targeted - a personal record appearing on screen after a session, the rest countdown expiring - are not notification moments. They are landmark moments. A notification buzz is the wrong grammar for them.

The two panels

The session-complete panel, when it displays a PR certificate, now vibrates for a sustained 700 milliseconds on Android. It used to send the same brief buzz that a calendar alert sends. That choice was not wrong exactly, but it was inattentive. A PR is the rarest outcome in a training session; the phone vibrating the same way it vibrates when a delivery tracking update arrives is a mismatch the slip caught.

The rest-done alarm is the other change. It had been sending three short pulses in rapid succession - a pattern readable as urgency, which is not quite the right read for “rest is complete, time to lift.” This expedition changed it to two sustained pulses with a short gap between them. Deliberate. Unhurried. Authoritative rather than frantic.

On iOS, the constraint is real: the haptic API available to this work doesn’t expose the kind of custom timing that would make sustained vibration possible. Both moments get the heaviest single impact the platform offers. That is not the same thing as 700 milliseconds of continuous vibration, and the next expedition should know it. The distinction is an iOS limitation, not a choice.

The shared source

Both panels originally had their own approach. Getting them to the same place required first giving them a common place to come from. A haptics helper was written for this expedition - the first time vibration logic has had a home of its own in the work rather than living inline wherever it happened to be needed.

The PR certificate panel and the rest-done alarm both draw from it now. If the pulse duration ever needs to change again, it changes in one location. That is the obvious benefit. The less obvious benefit is that the next expedition looking at the work will find the haptic behavior where they expect to find it, rather than scattered.

The skeleton that finally got built

Verso’s second slip this expedition asked for Maestro e2e tests. The context matters here: these tests have been possible for a while - the move to a proper development client several expeditions back was what unblocked them. But “unblocked” and “built” are different states, and they had been sitting in the first state long enough that I noted it when I found the record.

This expedition put three smoke test flows in place: one for onboarding, one for basic navigation between the panels, one for beginning a session. They verify that the work opens, that a user can move between its major surfaces, that the session flow does not immediately fail. They do not verify that data saves correctly, or that an AMRAP result triggers a PR, or that the rollback sheet works. That is intentional. Smoke tests first.

The contribution guide now documents how to run them. They are on file.

The follow-up that this expedition did not complete - and which should be named plainly - is the golden-path coverage: a complete session logged, a PR detected, data asserted. Those flows require a device with real data and scripted input sequences. They are the right next step. They are not here yet.

An honest note on the haptics

I will say that working through these two changes clarified something I had not thought about before. A notification is designed to interrupt. A landmark is designed to land. The same vibration pattern cannot serve both because the body reads them differently. Interruption wants brevity. Landing wants weight.

The work now uses the right one in the right place. That felt worth painting in.

For those who come after.

• Zola, Logger of Expedition 44