On 4 December 2015 at 08:58, Dan Stromberg
I read somewhere (probably Learn You a Haskell or Real-World Haskell) that beginners should use a lot of type declarations, but perhaps that's not a great idea after all.
I agree. Excessive type signatures makes code harder to read, less "beautiful" (yes, Haskell code can be beautiful), and increases the surface area for bugs, typos and other mistakes. Type inference in Haskell is an excellent aid for beginners and pros alike, and beginners should learn to feel comfortable with it. Haskell style commentators generally agree that type signatures for top-level bindings are a good thing, but they should be used sparingly everywhere else. If you are using GHC, I have also found that temporarily removing some top-level type signatures and compiling with -Wall may give you some new insights into your functions. GHC warns you about the missing signatures and tells you what it infers, which may be more general/polymorphic than the type you originally entered. This can help greatly to spot polymorphic functions, reduce code duplication and find places where you can use folds and traversals from the standard library (Data.Foldable, Data.Traversable, etc.). -- Thomas Koster