How do you deal with function visibility and unit testing in Haskell?
If you export every function in a module so that the unit tests have access to them, you risk other people calling functions that should not be in the public API.
I thought of using {-# LANGUAGE CPP #-} and then surrounding the exports with an #ifdef:
{-# LANGUAGE CPP #-}
module SomeModule
#ifndef TESTING
( export1
, export2
)
#endif
where
Is there a better way?
The usual convention is to split your module into public and private parts, i.e.
module SomeModule.Internal where
-- ... exports all private methods
and then the public API
module SomeModule where (export1, export2)
import SomeModule.Internal
Then you can import SomeModule.Internal in tests and other places where its crucial to get access to the internal implementation.
The idea is that the users of your library never accidentally call the private API, but they can use it if the know what they are doing (debugging etc.). This greatly increases the usability of you library compared to forcibly hiding the private API.
For testing you normally split the application in the cabal project file, between a library, the production executable, and a test-suite executable that tests the library functions, so the test assertion functions are kept apart.
For external function visibility you split the library modules between the "exposed-modules" section and the "other-modules" section.