Proposal: Add Text.Read.maybeRead :: Read a => String -> Maybe a

Conor McBride conor at strictlypositive.org
Fri Feb 8 12:03:38 EST 2008


Hi

Don:
>
> > This function is typically defined once per project. So its
> > about time this safe variant of 'read' made it into the base.
> >
> >     maybeRead :: Read a => String -> Maybe a
> >     maybeRead s = case reads s of
> >         [(x, "")] -> Just x
> >         _         -> Nothing
> >

Jeff:

> Why not allow an arbitrary monad?
>
> readM :: (Monad m, Read a) => String -> String -> m a
> readM errMsg s = case reads s of
>     [(x, "")] -> return x
>     _ -> fail errMsg

My instinct here is to follow James McKinna's observation
(which give rise to views in Epigram): you don't need to
produce elements of an *arbitrary* whatever-it-is when you
can produce elements of the *initial* whatever-it-is. It
makes the construction a bit easier to implement, because
you're working at a concrete type, but no harder to use.

I'd suggest going with the Maybe version, but then add the
relevant initiality principles for Maybe (if they're not
already there). This thing

mayA :: Alternative a => Maybe x -> a x
mayA (Just x) = pure x
mayA Nothing = empty

is a useful little piece of glue, and it has a few friends
which might also help, including, for the moment,

mayM :: Monadplus m => Maybe x -> m x
mayM (Just x) = return x
mayM Nothing = mzero

and (oh all right then)

mayhem :: Monad m => Maybe x -> m x
mayhem (Just x) = return x
mayhem Nothing = fail "I told you so!"

These things are common factors in quite a lot of
unnecessarily abstract operations.

I suggest factoring out the final appeals to initiality,
keeping the actual machinery simple and specific.

All the best

Conor



More information about the Libraries mailing list