In my first post I mentioned that Cabal is capable of building executables, but so far we’ve only looked at libraries. Hackage also contains a plethora of tools for the working Haskell programmer, and today we will look at HLint.
HLint, created by Neil Mitchell, is a tool that takes Haskell
.hs files as input, and analyses them to produce hints on better ways to write code - be that more performent, more succinct or more idiomatic. Because of this, HLint is frequently suggested to beginners - but it’s still a tool I frequently use! Though maybe I’m really still just a beginner…
It would be a rather long article if I were to enumerate all of the hints HLint can give you; instead I will concentrate on a few that have helped further my Haskell knowledge.
To the beginner, Haskell seems full of operators and combinators, and has a somewhat alien way of calling functions. This threw me for a while - “how on earth am I meant to remember the precedance for function application?!” and “why can’t I drop the brackets here?” were amongst the questions I found myself asking. HLint helped clear a lot of this confusion for me: instead, I started liberally using parentheses and let HLint guide me to write more idiomatic Haskell. For example, we could start with the following code.
[(10 + 20), 9] == (map (\x -> x * 3) [10, 3])
HLint, after a few passes, reduces this code to:
[10 + 20, 9] == map (* 3) [10, 3]
Which is certainly much cleaner! By repeating this exercise, it all started to click a litle bit more.
HLint is also aware of common programming patterns too, which has been another way to strengthen my knowledge, especially in developing intuition for how all the pieces fit together. All the various monad combinators seemed a little mythical to me at first, and I would often write code such as:
getLine >>= return . map toUpper
But this can be simplified by the
liftM (map toUpper) getLine
Why not indeed - thanks HLint!
HLint also allows you to extend it with your own hints - a feature that I think can be over looked, but is nonetheless very useful, and likely essential for large team who wish to write consistent code. I personally tend to prefer the use of infix
<$> - so ideally I’d like
fmap x foo to produce a hint to change it to
x <$> foo.
warn = fmap f x ==> f <$> x
And now if we try:
fmap (map toUpper) getLine
HLint rightfully suggests:
Warning: Use <$> Found: fmap (map toUpper) getLine Why not: map toUpper <$> getLine
Simple! HLint features literally hundreds of hints out the box, which are already enough to get going with. If your participating in 24 Pull Requests this year, HLint can be a great way to make pull requests on code that is otherwise too difficult for you - after all, which maintainer doesn’t want pretty, consistent code?
You can contact me via email at firstname.lastname@example.org or tweet to me @acid2. I share almost all of my work at GitHub. This post is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.