{-# LANGUAGE DeriveDataTypeable #-}
{-# LANGUAGE TypeFamilies #-}
{-# LANGUAGE ConstraintKinds #-}
{-# LANGUAGE FlexibleContexts #-}
{-# LANGUAGE StandaloneDeriving #-}

module PlaceHolder where

import GhcPrelude ()

import Type       ( Type )
import Outputable
import Name
import NameSet
import RdrName
import Var

import Data.Data hiding ( Fixity )

%*                                                                      *
\subsection{Annotating the syntax}
%*                                                                      *

-- NB: These are intentionally open, allowing API consumers (like Haddock)
-- to declare new instances

-- | used as place holder in PostTc and PostRn values
data PlaceHolder = PlaceHolder
  deriving (Data)

placeHolderKind :: PlaceHolder
placeHolderKind = PlaceHolder

placeHolderFixity :: PlaceHolder
placeHolderFixity = PlaceHolder

placeHolderType :: PlaceHolder
placeHolderType = PlaceHolder

placeHolderTypeTc :: Type
placeHolderTypeTc = panic "Evaluated the place holder for a PostTcType"

placeHolderNames :: PlaceHolder
placeHolderNames = PlaceHolder

placeHolderNamesTc :: NameSet
placeHolderNamesTc = emptyNameSet

placeHolderHsWrapper :: PlaceHolder
placeHolderHsWrapper = PlaceHolder


Note [Pass sensitive types]
Since the same AST types are re-used through parsing,renaming and type
checking there are naturally some places in the AST that do not have
any meaningful value prior to the pass they are assigned a value.

Historically these have been filled in with place holder values of the form

  panic "error message"

This has meant the AST is difficult to traverse using standard generic
programming techniques. The problem is addressed by introducing
pass-specific data types, implemented as a pair of open type families,
one for PostTc and one for PostRn. These are then explicitly populated
with a PlaceHolder value when they do not yet have meaning.

In terms of actual usage, we have the following

  PostTc id Kind
  PostTc id Type

  PostRn id Fixity
  PostRn id NameSet

TcId and Var are synonyms for Id

Unfortunately the type checker termination checking conditions fail for the
DataId constraint type based on this, so even though it is safe the
UndecidableInstances pragma is required where this is used.

-- |Follow the @id@, but never beyond Name. This is used in a 'HsMatchContext',
-- for printing messages related to a 'Match'
type family NameOrRdrName id where
  NameOrRdrName Id      = Name
  NameOrRdrName Name    = Name
  NameOrRdrName RdrName = RdrName