“If typeclass a, then a is also an instance of b by this definition.”

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

“If typeclass a, then a is also an instance of b by this definition.”

Saurabh Nanda
Hi,

At the outset I would like to mention that I do not completely understand how type-class instance resolution works, and why it works that way. I have tried reading up, but I don't completely understand WHY it works that way. So, please feel free to ask me to RTFM if what I'm about to say doesn't make any sense, although TFM doesn't make sense to me!

From a pragmatic standpoint is there any way to express the following idea in Haskell (or a future version of Haskell), **with minimal amount of boilerplate**

    If a type has an instance of type-class `a`, then here's how you can get an instance of type-class `b`

Without the ability to express this idea, you end up with a **lot** of type-class boilerplate, (even if you use GHC Generics) eg:

    data State = data State = Inactive | Incomplete | Deleted | Trial | Unpaid | Paid
      deriving (Eq, Show, Generic)

   instance AppEnum State where
       toString = gEnumToString
       fromString = gEnumFromString

   -- You are still forced to do the following for **every single type** even though, conceptually, this can be driven by a rule

    instance ToJSON State where
      toJSON s = toJSON $ toString x

    instance FromJSON State where
      parseJSON (Aeson.String s) = fromString s
      parseJSON _ = fail "unexpected type"

    instance FromFIeld State where
      fromField _ mBS = case mBS of 
        Nothing -> fail "not expecting a NULL/Nothing"
        (Just bs) -> pure $ fromString bs

    instance ToField State where
      toField s = toField $ toString s

I tried talking to people on IRC and also trawling StackOverflow [1] and my take-away is essentially, that Haskell cannot express this idea. However, my submission is, that it is a very useful idea to express and probably the next version of the language should do something to make the lives of programmers a little easier.


-- Saurabh.


_______________________________________________
Haskell-prime mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-prime
Reply | Threaded
Open this post in threaded view
|

Re: “If typeclass a, then a is also an instance of b by this definition.”

Mike Izbicki
I've also wanted this feature in the past.  I think the
DefaultSignatures language extension is the closest thing.  There's a
recent answer to your stackoverflow question that explains how it
works:

https://stackoverflow.com/questions/3213490/how-do-i-write-if-typeclass-a-then-a-is-also-an-instance-of-b-by-this-definit/44456791#44456791

On Fri, Jun 9, 2017 at 3:15 AM, Saurabh Nanda <[hidden email]> wrote:

> Hi,
>
> At the outset I would like to mention that I do not completely understand
> how type-class instance resolution works, and why it works that way. I have
> tried reading up, but I don't completely understand WHY it works that way.
> So, please feel free to ask me to RTFM if what I'm about to say doesn't make
> any sense, although TFM doesn't make sense to me!
>
> From a pragmatic standpoint is there any way to express the following idea
> in Haskell (or a future version of Haskell), **with minimal amount of
> boilerplate**
>
>     If a type has an instance of type-class `a`, then here's how you can get
> an instance of type-class `b`
>
> Without the ability to express this idea, you end up with a **lot** of
> type-class boilerplate, (even if you use GHC Generics) eg:
>
>     data State = data State = Inactive | Incomplete | Deleted | Trial |
> Unpaid | Paid
>       deriving (Eq, Show, Generic)
>
>    instance AppEnum State where
>        toString = gEnumToString
>        fromString = gEnumFromString
>
>    -- You are still forced to do the following for **every single type**
> even though, conceptually, this can be driven by a rule
>
>     instance ToJSON State where
>       toJSON s = toJSON $ toString x
>
>     instance FromJSON State where
>       parseJSON (Aeson.String s) = fromString s
>       parseJSON _ = fail "unexpected type"
>
>     instance FromFIeld State where
>       fromField _ mBS = case mBS of
>         Nothing -> fail "not expecting a NULL/Nothing"
>         (Just bs) -> pure $ fromString bs
>
>     instance ToField State where
>       toField s = toField $ toString s
>
> I tried talking to people on IRC and also trawling StackOverflow [1] and my
> take-away is essentially, that Haskell cannot express this idea. However, my
> submission is, that it is a very useful idea to express and probably the
> next version of the language should do something to make the lives of
> programmers a little easier.
>
> [1]
> https://stackoverflow.com/questions/3213490/how-do-i-write-if-typeclass-a-then-a-is-also-an-instance-of-b-by-this-definit
>
> -- Saurabh.
>
>
> _______________________________________________
> Haskell-prime mailing list
> [hidden email]
> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-prime
>
_______________________________________________
Haskell-prime mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-prime
Reply | Threaded
Open this post in threaded view
|

Re: “If typeclass a, then a is also an instance of b by this definition.”

Joachim Breitner-2
In reply to this post by Saurabh Nanda
Hi,

Am Freitag, den 09.06.2017, 15:45 +0530 schrieb Saurabh Nanda:
>     If a type has an instance of type-class `a`, then here's how you
>     can get an instance of type-class `b`

this has been proposed as “default superclass instances”, see
https://ghc.haskell.org/trac/ghc/wiki/DefaultSuperclassInstances

I am not sure what the status is, and how actively it is pushed for. It
is certainly compelling.

Joachim
--
Joachim Breitner
  [hidden email]
  http://www.joachim-breitner.de/

_______________________________________________
Haskell-prime mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-prime

signature.asc (849 bytes) Download Attachment