Changelog

v1.2.3 — 2026-07-31

How hard a set was, in whichever of the two scales you already think in — and the ratings your
old app recorded come across with the rest of your history. Plus: the phone stops locking itself
mid-workout, the rest timer can hand time back as well as take it, and Settings is grouped by
what each thing actually affects.

The screen stays on while you train

• ☀️ Keep screen awake — Settings → During a workout, on by default. Locking, unlocking
  and finding your place again between every set was the single most annoying thing about
  logging on a phone. The screen now stays lit for as long as a workout is running and lets go
  the moment you finish it, so nothing is held while you are not training.
• It survives a tab switch. Browsers release the lock whenever the page stops being visible,
  which is exactly what happens when you glance at a message. The lock is taken again each time
  the app comes back, rather than dying the first time you look away.
• It follows the workout, not the screen you are on. Checking Stats mid-session keeps the
  screen awake.
• Where it isn't available, it says so. iOS grants no wake lock in Low Power Mode, and older
  browsers have no Wake Lock API at all — the first is silent, the second shows the row disabled
  rather than offering a switch that does nothing. Needs HTTPS, like every other modern browser
  capability.

Rest timer: take 15 seconds off, too

• ⏳ A −15s button next to +15s. The timer could only ever be extended or skipped outright;
  now it goes both ways. Taking off more than is left finishes the rest rather than counting
  into the negative — the same thing Skip does.
• Rearranged so three controls fit. The clock and the progress bar take the top row and the
  controls sit underneath: −15 and +15 together in number-line order, Skip pushed to the far
  edge so the button that ends the rest is not next to the one you tap to buy more time. On a
  wide screen it stays on one line. Tap targets are bigger than they were.
• The bar is nearly opaque. The set rows underneath were reading through it and making the
  clock hard to pick out.

Settings, grouped by what it affects

• General (language, units) · During a workout (rest timer, keep screen awake, sounds,
  effort per set) · Notifications · Appearance (theme, body diagram, accent) · Data.
• The old grouping mixed axes: "Units & timer" put a display preference next to two workout
  behaviours, language sat under Appearance, and Load starter plan was buried between the
  backup actions and the destructive reset. Data now reads in the order you would use it — fill
  the plan, bring history over from another app, restore a backup, export one, wipe everything.
• Nothing was removed and no setting changed its meaning.

Effort per set: RIR or RPE (#21)

• 🎯 A third column on a working set, off by default. Settings → Effort per set switches
  it between Off, RIR and RPE. It only appears on weighted rep sets: a plank or a
  treadmill row has nowhere to put it.
• Two names for the same judgement. RIR counts the reps you left in the tank; RPE reads the
  same effort off a 10-point scale, so RPE ≈ 10 − RIR. The setting has an (i) that lays the two
  scales side by side in a conversion table rather than explaining them in a paragraph.
• Each set keeps the scale it was logged with. Switching the setting changes what new sets
  ask for and nothing else — history is never silently rewritten, and a set logged as RIR 2
  still reads back as RIR 2 years later.
• An unrated set stays unrated. Blank and 0 are different things: RIR 0 says the set went to
  failure. So − on an untouched cell leaves it empty, + starts at the bottom of the scale
  and walks up in even steps, and stepping back off the bottom clears the cell again — a mistap
  is always undoable.
• Nothing else reads the value. Progression rules and estimated 1RM are unaffected; the
  rating is yours to look at, not an input to the maths.
• Upgrading keeps the column you had: a profile still carrying the old showRir flag — from
  this device, a sync, or a backup restored later — comes across as RIR.

Import brings your ratings with it

• 📥 The RPE Hevy and Strong export is no longer dropped. An RPE column is read into the
  set, as is an RIR column if a file has one, and the import summary says how many sets
  arrived with a rating — plus where to switch the column on if it's off.
• A blank cell stays unrated rather than becoming 0. A written-out 0 counts as a rating on the
  RIR scale (a set to failure) but not on RPE, which starts at 1 — apps write 0 there to mean
  "nothing here", and reading it as an effort would stamp one on every unrated set in the file.
• Ratings above the scale are capped instead of thrown away, and junk in the column is ignored
  without losing the set.
• Backups already carried both fields and the setting, since a backup is the whole state — there
  are now tests pinning that, so it can't quietly stop being true.