Pomodoro intervals as the engine runs them

The pattern is the widely used one: a fixed focus interval, a short break after each interval, and a longer break after every 4th. Three lengths and two counts describe a session, plus a switch for whether the next phase starts itself. Every one of them is a default you can change before you press start on the focus timer.

Settings are normalised before they run. Each length has a floor and a ceiling, and so do the counts. Asked for a focus interval of 0 seconds, the engine raises it to 5 seconds; asked for a long break of 10 hours, it holds it at 4 hours. A cadence of 1 is raised to 2, a session goal of 99 is held at 24, and a focus length that is not a number at all falls back to the default, 25 minutes.

The ordering rule is short. Every break is followed by focus. Focus is followed by a long break when the number of focus intervals completed so far is a multiple of the cadence, and by a short break otherwise. The session finishes the moment its last focus interval ends, so no break is scheduled after it.

The table below is not a typed-out plan; it is the engine's own event log. A default session was started at a fixed instant that is a constant in this page, then rolled forward in one call to an instant 24 hours later. The engine crossed every boundary the elapsed time demanded, stopped when the session finished, and reported each phase it completed with the instant that phase ended, measured from the start.

Read down the kind column and the cadence is visible. Phase 8 is the only long break, because the session finishes on phase 15, its 8th focus interval. Asked for the first 16 phases regardless of any goal, the sequence puts long breaks at phases 8 and 16; the second of those would have followed the 8th focus interval, and the session ends there instead. The last entry in the log ends 4 hours 5 minutes after the start, which is exactly what the session estimate reports.

The line under the settings — a full session of 8 intervals takes about 4 hours 5 minutes including breaks — is the sum of the phase lengths in the log above. Counted from that log, the default session is 8 focus intervals, 6 short breaks and 1 long break: 3 hours 20 minutes of focus and 45 minutes of breaks, 4 hours 5 minutes in all.

Two user-set configurations show how the shape changes. The first is the one the tool's own questions use as an example: focus intervals of 50 minutes, short breaks of 10 minutes, long breaks of 25 minutes, the long break every 3 intervals and a goal of 6. Rolled forward the same way, its session runs Focus, Short break, Focus, Short break, Focus, Long break, Focus, Short break, Focus, Short break, Focus, and the estimate is 6 hours 5 minutes. The second keeps the default lengths but takes the long break every 2 intervals with a goal of 4: Focus, Short break, Focus, Long break, Focus, Short break, Focus, estimated at 2 hours 5 minutes.

The figure is an estimate rather than a promise because it adds phase lengths and nothing else. Time spent paused is not in it, and neither is the time a queued phase spends waiting for you when automatic advance is off. The copy preset link button writes the same settings into a query string — ?work=50m&break=10m&long=25m&every=3&goal=6&auto=1 for the first configuration and ?work=25m&break=5m&long=15m&every=2&goal=4&auto=1 for the second — so a configuration can be handed to someone else as a link. The guide to shareable timer links covers how those tokens are read.

Every timer built on the timing engine — the countdown, the focus timer, the interval timer and the speech timer — stores an epoch anchor and derives its display from the difference between now and that anchor; the guide to how the timers stay accurate takes that rule on its own. The focus timer adds one rule of its own. When a phase ends, the next phase is chained from the previous phase's end instant, not from the moment the tick noticed. A hundred consecutive phases therefore end exactly where arithmetic says they should, with no accumulated tick latency.

Here is the rule run through the engine. A default session started at the fixed instant stores the end of its first focus interval, 25:00 after the start. Suppose the tick that notices that boundary arrives 3 seconds late. The engine reports one event — Focus finished at 25:00, with Short break next — and the short break it starts is set to end 30 minutes after the start: the boundary plus 5 minutes. A timer that restarted the break from the tick would instead end it 30 minutes 3 seconds after the start, and that 3 seconds would be carried into every phase that followed. The readout immediately after the late tick is 04:57, because 3 seconds of the break has already gone; the display says so rather than pretending the break began when the tick ran.

Rolling a session forward crosses as many phase boundaries as the elapsed time demands, and the engine states the consequence plainly: the result is identical whether the roll happens on a regular tick or after the tab has been in the background for twenty minutes. Two roll-forwards from the same fixed start make the point.

• Rolled forward 1 hour 1 minute in one call, the engine crosses 4 boundaries — Focus ended at 25:00; Short break ended at 30:00; Focus ended at 55:00; Short break ended at 01:00:00 — and lands in Focus with 2 focus intervals complete and 24:00 on the readout.
• Rolled forward 3 hours, it crosses 10 boundaries, including 1 long break, and lands in Focus with 5 complete and 05:00 remaining.

While the tab is in the background its title carries the same clock string and the phase label, so the first case would show as 24:00 · Focus in the tab strip. What the anchors guarantee is narrow and exact: whenever a tick does run, the display is correct for that instant. They do not make the page run while it is closed. The tab has to stay open for a boundary to be noticed as it happens, and phones and laptops that fall asleep can delay or suppress the chime — described here, not simulated. Boundaries that pass in a gap are listed by the engine in order, and the timer catches up to the last of them on the first tick that runs; a reload resumes the session where the clock says it should be, including any phase boundaries that passed while the page was closed.

With automatic advance off, the boundary is still found by the same roll-forward, but the next phase is queued rather than started: it is stored as paused, at its full length, and the session is marked as waiting. From the fixed start with advance off, a tick 1 minute after the end of the first focus interval leaves the session in Short break, waiting to be started, with 05:00 queued. The 1 minute that passed before the tick is not deducted from the break, because the break has not begun. Pressing start 4 minutes later resumes a paused run in the ordinary way: a fresh end instant is set 5 minutes after the press, and chaining continues from that new end.

Skip phase abandons the current phase and queues the next one the same way, paused at full length. Skipping the first focus interval of a default session queues Short break at 05:00 with 1 focus interval counted as complete. A skipped focus interval counts, which is what moves the long-break cadence along: in the second configuration above, with the long break every 2 intervals, 3 skips from the start queue Long break with 2 intervals complete.

Finishing is the one boundary the engine does not chain past. One second before the 8th focus interval of the default session ends, the phase is Focus and the readout is 00:01. Rolled past that end, the session is finished with 8 of 8 intervals complete, the run is idle, and the event that reports it names no next phase. Later ticks leave a finished session alone: rolled a further 1 hour on, the engine reports no further events.

The stage shows three things from the engine: the phase label — Focus, Short break or Long break — the clock readout for that phase, and the count of focus intervals against the goal. The readout always belongs to the current phase, never to the session: during a break it counts the break down, and it starts again from the full focus length when the next interval begins. Seconds are rounded up, so a fresh focus interval reads 25:00, 400 ms into the run it still reads 25:00, and only once a whole second has gone does it read 24:59. Each boundary comes with a visible change of phase and a written announcement as well as the chime, which can be muted; nothing depends on hearing it.

• “The tick was late, so the break started late.” It did not. The break is chained from the boundary, and the readout shows how much of it has already gone — 04:57 after a tick 3 seconds late, as above.
• “The session should end with a break.” It ends on its last focus interval. The readout reaches 00:00 on phase 15 of the default session and the run goes idle; no trailing break is scheduled or counted in the estimate.
• “The estimate is how long I will be here.” It is the sum of the phase lengths, 4 hours 5 minutes for the defaults. Pauses, and the waits between phases when automatic advance is off, are added on top.
• “Pausing only pauses this phase.” A pause keeps the balance of the current phase, and a resume sets a fresh end instant for it; every later boundary is chained from that new end, so the whole remainder of the session moves by the length of the pause.

Running state is saved to this browser's storage as epochs, never as “seconds left”, so a reload picks the session up exactly where the wall clock says it should be: the phase, the cycle count and the remaining time are all restored, and the boundaries that passed while the page was closed are crossed on the way. The saved envelope records when it was written, and one older than 12 hours is discarded, so the timer opens idle rather than resuming something stale.

The link the timer offers for its own defaults is ?work=25m&break=5m&long=15m&every=4&goal=8&auto=1. It is parsed entirely in your browser, and anything in it that cannot be understood is ignored so the default stands; a mangled link still opens a working timer.

“Pomodoro Technique” is a trademark. WRLD CLOCK is an independent tool that counts focus intervals and breaks; it is not affiliated with, endorsed by, or certified by the trademark holder, and no course, book or certification is implied. The attribution in full is in the note at the foot of the focus timer page.