[ps2]

Settings gets a Current Setup row under Statistics showing how long the setup has been unchanged and what the last change was, e.g. "5 days · Last change: Carb Ratios". It opens a screen with the time since the last change, its date, what changed, and a note that about 2 weeks with the same setup is a good amount of time to see a change in outcomes.

This is the first step toward comparing your results before and after a change, and later toward sharing that data with the developers. Both of those build on this date.

Settings	Current Setup

What counts as a change

The last change is found in the settings history Loop already stores (90 days), by comparing consecutive records:

• basal rates, insulin sensitivities, carb ratios, correction range, pre-meal range, glucose safety limit and delivery limits
• insulin type and model
• dosing strategy
• turning an algorithm experiment on or off

Not counted: running a preset, turning closed loop on or off, alerts and notifications, devices, the glucose unit and the time zone.

Details:

• Schedules are compared by their values. A value saved in the other glucose unit counts as the same within 2%, which covers the rounding to 0.1 mmol/L.
• Changes within an hour of the latest one are reported together, as one edit.
• With no change in the history, the screen says the setup has had no change recorded since the oldest record, and the time reads "at least …".

Experiments

Experiment changes come from the experiments recorded on each dosing decision, which builds since LoopKit/LoopWorkspace#534 store. That covers changes made before this build, and it uses the experiments actually in effect, so Glucose Based Partial Application turns off with automatic bolus. Decisions from older builds, which recorded nothing, are skipped. The experiment transitions are cached in UserDefaults, so each decision is read once. The 2-week note also appears on the Algorithm Experiments screen.

SetupHistory lists every change from both sources, oldest first, each with the date of the change before it. The Current Setup row uses the latest; #2619 uses the rest.

Needs LoopKit/LoopKit#669 (getDosingDecisions(start:end:as:)).

Testing

• 12 SetupChangeTests: change detection, grouping, ignored fields, units, the pump's missing insulin type, experiment changes from dosing decisions, grouping with settings changes, and the cache reading each decision once. All pass.
• Built with the LoopWorkspace scheme and checked in the simulator: turning Guard + Runway off updated the row to "1 minute · Last change: Guard + Runway turned off".
• A therapy settings change was not checked in the UI; the tests cover that comparison.

[ps2]

Added c0a7651: the stored insulin type comes from the pump's status, which has none while the pump is being replaced or hasn't loaded, so the history flipped between a type and none and showed up as "Insulin Type" changes. The last known type is now carried across those gaps, and only a change from one type to another counts. Found by decoding a simulator's settings history.