suggest adding Read instances for everything about github HOT 7 CLOSED

I've pushed this to master.

from github.

Re-opening because, well, that broke everything.

from github.

I've pushed this to master: derive Read for each data type except Error.

from github.

The Read instances take horribly long to compile with -O (the default with cabal install). Removing them cuts down compilation time of "Github/Data/Definitions.hs" from 1m18s to 17s (with ghc-7.4, with ghc-7.0.4 it was over 3 minutes); -O0 also helps a lot.

@mike-burns @joeyh can you verify this?

As far as I know, Read is not very popular, because derived instances have the reputation to be very slow. There is even something on that in the Report ("The parsing techniques used by the reader are very inefficient; reading a large structure may be quite slow.").

Maybe adding ToJSON instances would be a more useful alternative (would also close #9). We may need some of them anyway, when adding support for POST/PATCH requests.

@joeyh would JSON serialization/deserialization be suitable for your use case, or do you rely on Read for some reason?

from github.

Simon Hengel wrote:

@joeyh would JSON serialization/deserialization be suitable for your use case, or do you rely on Read for some reason?

I'd be happy with JSON serialization.

see shy jo

from github.

It should be possible to automatically derive FromJSON/ToJSON with template Haskell.

module Util where

import qualified Data.Char as Char

convertCamelCase :: String -> String
convertCamelCase = foldr f []
  where
    f x xs
      | Char.isUpper x = '_' : Char.toLower x : xs
      | otherwise      = x : xs

An then the following should work:

$(deriveJSON (drop 5 . convertCamelCase) ''Repo)

from github.

Closing this issue as a duplicate of #9. I've removed the Read instances from master.

from github.