Did you try to use transformers instead of mtl? I am just in doubt about it in my work.

I've found that Conduit use it. So you remove extra dependency.
I hope that transformers have more readable messages (without FD) and you can control what is "lift" more clear.

I have no experience although.



2012/3/2 Alexander V Vershilov <alexander.vershilov@gmail.com>
I've found a solution, I should not use lift for runWriterT, and should
explicilty lift all computation of level I need, i.e. (lift.lift) for ask
and lift for register.

Thu, Mar 01, 2012 at 02:19:29PM +0400, Dmitry Olshansky wrote
> >> If I'm running register outside runWriterT everything will work.
>
> Maybe just
>
> > lift $ register $ print "freed2"
>
> or I didn't catch something?
>
>
>
> 2012/3/1 Alexander V Vershilov <alexander.vershilov@gmail.com>
>
>     Hello.
>
>     I'm trying to add monad stack into network-conduit, and everything
>     works except some details [1].
>
>     I've run runReaderT $ runTCPServer (wrapper around runResourceT) and
>     inside conduit I want to run writer to gather results of inner computation.
>     In inner computation I want to use IO, data from outter stack (ReaderT)
>     so I'm running {-1-}:
>
>      (k,t) <- lift $ runWriterT $ ask >>= \x -> tell [x]     {- 1 -}
>
>     and {-2-}
>
>      (k,t) <- lift $ runWriterT $ do {- 2 -}
>                        x <- ask
>                        liftIO $ print $x+1
>                        tell [x]
>
>     and that will work (except I've thought I should not lift runWriterT, but
>     calling functions inside.
>
>     And finally in computation that will run once I want to register cleaning
>     function (for example register $ putStrLn "cleaned") ({-3-})
>
>      (k,t) <- lift $ runWriterT $ do {- 3 -}
>                        x <- ask
>                        liftIO $ print $x+1
>                        register $ print "freed2"
>                        tell [x]
>
>     but I've got type error. If I'm running register outside runWriterT
>     everything
>     will work.
>
>     I would apperated if there will be any suggestions how to make this code
>     better or use register in internal computation (runWriterT)
>
>     [1] https://gist.github.com/1941151
>     --
>     Best regards,
>      Alexander V Vershilov
>
>     _______________________________________________
>     Haskell-Cafe mailing list
>     Haskell-Cafe@haskell.org
>     http://www.haskell.org/mailman/listinfo/haskell-cafe
>
>
>

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe