Don't enforce compatibility with `base`

[adithyaov]

trafficstars

Being compatible with base is NOT required. This is not the responsibility of the library. Enforcing compatibility also gives an incorrect idea on the dependency graph. We use functions in base, not the unicode primitives. Hence it would be incorrect to constraint base depending on the unicode version.

The best case solution is to let base depend unicode-data and friends. Although, this is not a straightforward task.

The compatibility with the base primitives is a domain specific problem. The overhead of the check should be pushed to the user. The user may choose to force compatibility depending on how unicode is being used. The following snippet checks for compatibility and fails accordingly:

{-# LANGUAGE TemplateHaskell #-}
{-# LANGUAGE QuasiQuotes #-}

import qualified Unicode.Char as UnicodeData (unicodeVersion)
import qualified GHC.Unicode as Base (unicodeVersion)

unicodeCompatibility :: ()
unicodeCompatibility =
    $(if Base.unicodeVersion == UnicodeData.unicodeVersion
      then [|()|]
      else error $ "Incompatible unicode versions: "
                ++ "base: " ++ show Base.unicodeVersion ++ " "
                ++ "unicode-data: " ++ show UnicodeData.unicodeVersion)