[Haskell-cafe] hoisting as well as lifting and returning

Matthew Pocock matthew.pocock at ncl.ac.uk
Sun Mar 23 14:56:00 EDT 2008


Hi,

Happy Easter!

I've been using monad transformers for the first time in anger this week. They 
certainly do the job. However, there's an operation that I keep defining over 
and over again. It is sort of like lift and return in that it's glue to get 
operations and values into the monad from outside it. Here are two examples.

hoistList :: (Monad m) => [a] -> ListT m a
hoistList = foldr (mplus . return) mzero

hoistMaybe :: (Monad m) => Maybe a -> MaybeT m a
hoistMaybe = maybe mzero return

The general idea is that you have some legacy, non-transform operation in some 
monad MonadFoo, and you are writing an operation in MonadFooT. You want to 
get the monadFoo value into MonadFooT. So, you say something like:

do vInMFT <- hoist vInMF

Is this a common enough operation on a monad transformer and it's 'raw' monad 
to warrant a class of its own? I have 'hoist' methods for several monad 
transformers, including RamdomMonadT, although I've probably defined them in 
ways that are overly eager or eat stack/heap, so I'd prefer knowledgable 
people to write them :)

There must be other operations that link the base monad with its transformed 
version, other than 'hoist' - the runFooT method for example.

Perhaps 'hoist' already exists and I'm re-inventing the wheel?

Matthew


More information about the Haskell-Cafe mailing list