Brent Yorgey
On Mon, Feb 14, 2011 at 02:10:13PM -0500, Britt Anderson wrote:
I am looking at code from a tutorial that has the following type signature for a function:
getScreen :: MonadReader AppConfig m => m Surface
AppConfig is a data structure that was defined previously. My question is how am I to interpret this type signature? Is it any m belonging to the MonadReader AppConfig class?
Yes. And there will be an instance for MonadReader AppConfig m exactly when m carries around an implicit "read-only" AppConfig value, available for querying with the 'ask' and 'asks' functions.
Since I can't see any instance of the MonadReader Class being defined for MonadReader AppConfig I am puzzled.
Typically what you will see is some concrete monad later being defined in terms of ReaderT AppConfig, which will automatically give it an appropriate MonadReader instance, allowing users of the monad to query the AppConfig.
I would appreciate some clarification or a pointer to a prior discussion of this issue. I have tried to look on my own.
I recommend reading Cale Gibbard's article "How to use monad transformers"... except at the moment I can't seem to find it anywhere. =(
-Brent
This looks like my code, just to add this is this requires Flexible Contexts extension (for which I believe became part of the Haskell 2010 standard if I remember correctly) I didn't need to write it this way, I could have used the specific monad transformer stack that I use. The only reason to use a typeclass is for more generic code, it makes the code decoupled from a specific monad stack (any monad type for that matter, any which implements MonadReader AppConfig). Also note that at the time I didn't know much about first-class record libraries like fclabels which could have reduce repeative boiler-plate code I wrote in later tutorials.