Properly recurse into :*:

[dminuoso]

The old FromValue code makes the assumption that product types are always of the shape (a :*: (b :*: c)).
However GHC balances product and sum types (https://hackage.haskell.org/package/base-4.11.0.0/docs/GHC-Generics.html#g:9)

Resolves #173

[Akii]

Can confirm this solves the issue. Thanks for fixing it!

[teh]

Looks good thank you!

[jml]

Can you please add a test that takes advantage of the fix in this code?

[Akii]

data Resource = Resource
    { roleId      :: Text
    , resAge      :: Int32
    , resGender   :: Text
    , resLang     :: Text
    } deriving (Generic, Show)

instance FromValue Resource

Will that do? If the test compiles it "should" work. Can I write the actual test somehow?

[jml]

Yes, please add a test. It should use a resource definition like the one you provided and it should call fromValue on a JSON object and confirm that it's what you expect.

[Shou]

What's the status on this?

[jml]

This needs some kind of executable test before it can be merged. I've asked @dminuoso to do that, but they are a volunteer, so it's up to them when they do it. If you would like to create a new PR that includes the work of this one and adds the necessary tests, that would be very much appreciated.

[dminuoso]

I'll look at it this week.

…

On Mon, Jul 16, 2018, 08:13 Jonathan Lange ***@***.***> wrote: This needs some kind of executable test before it can be merged. I've asked @dminuoso <https://github.com/dminuoso> to do that, but they are a volunteer, so it's up to them when they do it. If you would like to create a new PR that includes the work of this one and adds the necessary tests, that would be very much appreciated. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#178 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ATQbiQOJF8pRyd2DqqggQwq-HyG78XWLks5uHC8XgaJpZM4TJ3HM> .

[EdmundsEcho]

I have FromValue being derived like a charm... Thank you for that! Should I expect this fix to include the ability to derive HasAnnotatedInputType for a record with more than 3 fields? I'm still getting an error.

data MixInputType = MixInputType
  { keyType    :: Text
  , keyName    :: Text
  , textValues :: Maybe [Text]
  , intValues  :: Maybe [Int32]
  , spanValues :: Maybe [SpanInputType]
  } deriving (Show, Generic)

instance FromValue Mix              <<< works using the commit referenced in the previous post
instance HasAnnotatedInputType Mix  <<< generates an error

I can avoid the error if I limit the number of fields to three.

[Gabriella439]

@EdmundsEcho: It looks like HasAnnotatedInputType has the same issue. It needs to be fixed in the same way as in this pull request since it was also written with the assumption that :*: constructors would form a linked list instead of a balanced binary tree

[dminuoso]

Sorry, work stuff kept me too occupied. The issue with writing tests here is that I'd have to write generics for the other direction from scratch.

@Gabriel439 Do you have a testcase that I can use to validate my fix?

[Gabriella439]

@dminuoso: The following example fails to derive the HasAnnotatedInputType because the data type has four fields:

{-# LANGUAGE DeriveAnyClass #-}
{-# LANGUAGE DeriveGeneric  #-}

import GHC.Generics (Generic)
import GraphQL.API (HasAnnotatedInputType)

data Example = Example
    { field0 :: Bool
    , field1 :: Bool
    , field2 :: Bool
    , field3 :: Bool
    } deriving (Generic, HasAnnotatedInputType)

... but if you remove one of the fields then the example works:

{-# LANGUAGE DeriveAnyClass #-}
{-# LANGUAGE DeriveGeneric  #-}

import GHC.Generics (Generic)
import GraphQL.API (HasAnnotatedInputType)

data Example = Example
    { field0 :: Bool
    , field1 :: Bool
    , field2 :: Bool
    } deriving (Generic, HasAnnotatedInputType)

I tested those two examples using graphql-api-0.3.0

[dminuoso]

Yeah it's all addressed by 5747b88