Worked Example: The Greenhouse Log

One complete program, start to finish — the proof the previous chapters have been pointing at. examples/greenhouse.py analyses a week of invented greenhouse readings: three temperatures a day plus an evening soil-moisture percentage. The gardener's rule: water in the evening when the day's average temperature passed 24 degrees or the soil moisture fell below 30%.

The program keeps its data as a list of tuples, computes each day's average with a small function, applies the rule with a second one, prints a table, and finishes with the week's extremes. Nothing exotic: loops, format strings, min, max, sum, a list comprehension. Run under MicroPython:

$ micropython fixtures/greenhouse.py
Day  Average  Water?
Mon   20.83   no
Tue   22.67   no
Wed   23.83   YES
Thu   21.83   no
Fri   20.17   no
Sat   24.50   YES
Sun   22.17   no

Week's coolest reading: 16.5
Week's warmest reading: 29.0
Week's average:         22.29
Evenings to water:      2

Wednesday earns its YES on dry soil (29%) despite an average under the heat line; Saturday earns it on both counts. And now the referee:

$ diff <(micropython fixtures/greenhouse.py) <(python3 fixtures/greenhouse.py)
[exit 0]

No output, exit 0: the two interpreters' answers are byte-identical. This is the fact to build your mental model on. The parts of Python that describe a problem — data in tuples and lists, decisions in functions, results formatted for a human — are the same language in both implementations. A program like this one, written on a desk machine under either interpreter, moves to the other unchanged.

What would not survive the move to an actual greenhouse controller is everything around the program: where the readings come from (a sensor on a pin, not a typed list), where the table goes (there may be no screen), and how big the log may grow before chapter 6's heap says no. The logic transfers; the plumbing is chapter 11. Keeping those two apart — logic you can test here, plumbing you must test there — is the whole working method this guide is teaching.

Check: change DRY_SOIL in the fixture to 35 and re-run under both interpreters. Thursday (33%) should join the watering list in both, and the diff should still find nothing.