I am working with some libraries that provide APIs for interacting with specific hardware chips (that makes these drivers?). However, different custom boards or shields will have the chip mapped to different pins meaning the library needs to be modified for each case. Needing to modify the library does not work well with the Arduino IDE Library Manager.

Are there preferred/recommended patterns for exposing this configuration so that the library itself does not need to be modified every time?

Here is an example where it is documented which part needs to be changed to match the pin layout of your board.

  • Many of the normal Arduino libraries already do this - start by familiarizing yourself with that method, even from a user perspective. Aug 20, 2017 at 16:48

4 Answers 4


The method I use is to provide the pins as parameters to the constructor. Those pin numbers are stored in variables to use later in the .begin() function and elsewhere.

Most of the time I use initialization lists to keep things simple. For example:

class Something {
    uint8_t _cs;
    uint8_t _dc;

    Something(uint8_t cs, uint8_t dc) : _cs(cs), _dc(dc) {}
    void begin();

void Something::begin() {
    pinMode(_cs, OUTPUT);
    pinMode(_dc, OUTPUT);

Something mySomething(10, 8);

I would use either of the following two possibilities:

Use (class) variables and set them within the constructor.


  • Always initialized
  • Easy to use (constructor and pin setup at once)

Use a separate (e.g. Init) method.


  • Can be dynamically changed


For pin settings, mostly static circuits are used so the first approach is probably better.

For settings, mostly the second method is better.

If many pins are involved (not likely), use a structure or separate pin settings class.


What I wouldn't advice is macros. When users need to change source code themselves, and new versions are installed, they either have to merge or redo the changes again. The advantages is a bit less (machine)code, a little bit faster probably and a bit less memory usage, but all three aspects are minimal.


In case you'd avoid the C++ constructor stuff that's quite commonly an overkill on Arduino, you could use #define's (object-like macros).

Like so:

#define PIN_ONE 1
#define PIN_TWO 2

The preprocessor will seamlessly replace PIN_ONE with the number 1 and PIN_TWO with 2 assuming those definitions are in the library header .h file. This will most likely takes the least amount of resources compared to the other possible solutions.

Alternative 1

If you have the opportunity to use modern-ish C++, you can use constexpr to define variables. Compared to macros you'll get type checking and it should be just as efficient.

  • The problem is that they need to be in a place where both the .ino file as well as the library source can get to them. This usually means a separate header file with everything that requires. Aug 23, 2017 at 18:10
  • Are you sure? Pretty sure I can do #define switches in .ino's and they are used in libraries, but I might be wrong.
    – Avamander
    Aug 24, 2017 at 13:20
  • 1
    It can work if the code for the library is strictly in the header, but not if it's in a different compilation unit altogether. Aug 24, 2017 at 13:26
  • Yeah, that makes sense, didn't know the exact limitations, added that disclaimer.
    – Avamander
    Aug 24, 2017 at 13:28

depending on your approach.

1) if you just provide the binary + header files, you will have to make the pins variables.

2) if you provide the source code and expect the user to recompile the source code, use macros.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.