Extensible, type-safe error handling in Haskell
dea88bce — Thomas Letan 6 years ago
chore: Improve a bit the documentation of the main module
d473cbc2 — Thomas Letan 6 years ago
chore: Remove useless pragmas in Control.Monad.Chain
6aa486ff — Thomas Letan 6 years ago
feature: Change the semantics of 'finally'


browse  log 



You can also use your local clone with git send-email.


This haskell package started as an experiment to implement a Haskell-flavour error-chain. Its result is ResultT, a parameterised Monad which implements an extensible, type-safe error-handling.

This package is released on GitHub as an open source software, but it is not distributed under any particular licence, and therefore is not (yet) a free software. Hopefully, this will be changed before publishing chain to Hackage.

#In a Nutshell

A typical monadic function which lives inside the ResultT monad will have a type signature which looks like that:

function :: ('[Err1, Err2] :| err, Monad m)
        => a -> b -> ResultT msg err m c

'[Err1, Err2] :| err means function may raise an error of type Err1 or Err2 while it computes a result of type c. The computation is done within the monad m, that is ResultT can be part of a monad stack a la mtl. ResultT is not an alternative to mtl, as Eff can be. It is a more flexible EitherT.

To escape the ResultT package means using the runResultT function, whose type signature is:

runResultT :: ResultT msg '[] m a -> m a

runResultT only accepts empty row of errors ('[]). This obliges you to handle your error. The package provides several functions to that end.


This is still a big “work in progress” project. You can have a look at this blogpost for an —already a bit out-dated— introduction to the ResultT monad.

Service Status
Travis Build Status


This package could not have been written without the works of other talented programmers, including (but not limited to):