I have a method that creates a data file after talking to a digital board:
CreateDataFile(IFileAccess boardFileAccess, IMeasurer boardMeasurer)
Here boardFileAccess and boardMeasurer are the same instance of a Board object that implements both IFileAccess and IMeasurer. IMeasurer is used in this case for a single method that will set one pin on the board active to make a simple measurement. The data from this measurement is then stored locally on the board using IFileAccess. Board is located in a separate project.
I've come to the conclusion that CreateDataFile is doing one thing by making a quick measurement and then storing the data, and doing both in the same method is more intuitive for someone else using this code then having to make a measurement and write to a file as separate method calls.
To me, it seems awkward to pass the same object to a method twice. I've considered making a local interface IDataFileCreator that will extend IFileAccess and IMeasurer and then have an implementation containing a Board instance that will just call the required Board methods. Considering that the same board object would always be used for measurement and file writing, is it a bad practice to pass the same object to a method twice? If so, is using a local interface and implementation an appropriate solution?
Best Answer
No, this is perfectly fine. It merely means that the API is over-engineered with regards to your current application.
But that doesn't prove that there will never a use case in which the data source and the measurer are different. The point of an API is to offer the application programmer possibilities, not all of which will be used. You should not artificially restrict what API users can do unless it complicates the API so that the net understandability goes down.