Require call to `pinMode()` before read/write

[jgfoster]

digitalRead() and digitalWrite() should fail if we haven't made a call to pinMode(). Also, calling digitalRead() on an OUTPUT pin should be an error.

[per1234]

digitalRead() [...] should fail if we haven't made a call to pinMode()

The pin is in INPUT mode by default, so digitalRead() without prior pinMode() call is valid and common.
https://www.arduino.cc/en/Tutorial/Foundations/DigitalPins#properties-of-pins-configured-as-input

digitalWrite() should fail if we haven't made a call to pinMode()

digitalWrite() on a pin in INPUT mode sets the pull-up resistor. This less intuitive approach was the only way to do it via the Arduino API prior to the introduction of pinMode() in Arduino IDE 1.0.1, so there is still some legacy code that does this.
https://www.arduino.cc/reference/en/language/functions/digital-io/pinmode/#_description

Also, calling digitalRead() on an OUTPUT pin should be an error.

digitalRead() can be used to determine the state of an OUTPUT pin. Some people use this to avoid the need for a shadow variable for tracking the pin state:

void setup() {
  pinMode(13, OUTPUT);
}
 
void loop() {
  digitalWrite(13, !digitalRead(13));  // toggle state
  delay(1000);
}

---

You could certainly argue (as I do) that the latter two things are not good code style. I don't know whether the scope of this project covers code style linting. It is interesting to think of a tool specific to enforcement of good Arduino code style (avoiding overlap of functionality with the existing C++ code style and formatting tools).

[jgfoster]

Thanks for the feedback. So does digitalWrite() work without calling pinMode(pin, OUTPUT)? It seemed that I needed to add that to get things to run.

[per1234]

does digitalWrite() work without calling pinMode(pin, OUTPUT)?

If the pin is in the default INPUT mode, then digitalWrite() will enable or disable the pull-up resistor:

void setup() {}
 void loop() {
  digitalWrite(13, HIGH);  // enable internal pull-up resistor
  delay(1000);
  digitalWrite(13, LOW);  // disable internal pull-up resistor
  delay(1000);
}

[jgfoster]

Perhaps I need to go study the "pull-up resistor" idea. In any case, we had code that wasn't properly controlling a switch and when I added pinMode(PIN, OUTPUT); then things worked. Or so I thought; maybe it was something else I changed.

[ianfixes]

Did we get to a consensus on whether we are failing to mock the behavior of actual hardware?

[jgfoster]

I think @per1234 makes a good case that the existing behavior is legal and legitimate code, though, as he says, it might be good coding style to be more explicit and it is a fair question as to whether this tool should help find potential bugs. In any case, there is probably enough else to work on so I'll close this. Thanks to all!