Re: [GHC] #12596: can't find interface-file declaration

15 Sep 2016


      #12596: can't find interface-file declaration
-------------------------------------+-------------------------------------
        Reporter:  mwotton           |                Owner:
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.0.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

@@ -3,1 +3,1 @@
- ```
+ {{{
@@ -12,1 +12,3 @@
- ```
+ }}}
+
+ The code is

New description:

 full repro at https://github.com/mwotton/liftwoes/issues/1

 {{{
 /home/mark/projects/liftwoes/src/Lib.hs:14:11: error:
     • Can't find interface-file declaration for variable
 Data.Text.Internal.pack
         Probable cause: bug in .hi-boot file, or inconsistent .hi file
         Use -ddump-if-trace to get an idea of which file caused the error
     • In the first argument of ‘(:)’, namely
         ‘Data.Text.Internal.pack ((:) 'A' [])’
       In the first argument of ‘HS.fromList’, namely
 }}}

 The code is

 {{{#!hs
 {-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE TemplateHaskell   #-}
 module Lib where

 import           Data.Data
 import qualified Data.Set                   as HS
 import qualified Data.Text                  as T
 import qualified Data.Text.IO               as T
 import           Data.Time
 -- import           Instances.TH.Lift
 import           Instances
 import           Language.Haskell.TH.Syntax

 table = $(do r <- runIO (HS.fromList . T.lines  <$> T.readFile
 "/usr/share/dict/words")
              [|r|] )

 someFunc = do
   print $ HS.member "foo" table
 }}}

--

Comment (by simonpj):

 Yes, `bytemyapp` on the [https://github.com/mwotton/liftwoes/issues/1
 github repo] is right.  Here's a shorter example
 {{{
 {-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE TemplateHaskell   #-}
 module T12596 where

 import qualified Data.Set                   as HS
 import qualified Data.Text                  as T
 import           Language.Haskell.TH.Syntax

 table = $(do let r2 :: T.Text
                  r2 = "he" :: T.Text
              liftData r2 )
 }}}
 * `liftData` is defined in `Language.Haskell.TH.Syntax`:
 {{{
 liftData :: Data a => a -> Q Exp
 liftData = dataToExpQ (const Nothing)
 }}}

 * So we need `Data Text`.  The `Data` class is intended to describe
 algebraic data types, but the `text` package cleverly uses it to make
 `Text` behave like an algebraic data type, even though it isnt'.

 * Using the `toConstr` method of `instance Data Text` (in module
 `Data.Text`), it pretends that `Text` has a data constructor called `pack
 :: String -> Text`.  This function is defined in `Data.Text`.

 * But `toConstr :: Data a => a -> Constr` and `Constr` is a record giving
 only the ''string name'' of the constructor.
 {{{
 data Constr = Constr    { conrep    :: ConstrRep
                         , constring :: String
                         , confields :: [String] -- for AlgRep only
                         , confixity :: Fixity   -- for AlgRep only
                         , datatype  :: DataType }
 }}}
   It was really only intended for 'show'.

 * But here it's being used `Language.Haskell.TH.Syntax.dataToQa` to make a
 Template Haskell `Name` for `pack`.   It uses `mkNameG_v` passing the
 string for the "data constructor" (`pack`) but the module for the type
 constructor.  That is, `dataToQa` assumes that the data constructor is
 defined in the same module as the type constructor, which is usually
 reasonable.  Moreover, it has no other choice because `Constr` simply
 doesn't record the full, original `Name` of the data constructor.

 So that's the story.  It took me some while to puzzle it out!  And it's
 clearly unhelpful to users.

 Meanwhile, what can we do?

 * Refactor the `text` package so that the function mentioned in
 `Data.Text.packConstr` is actually defined in the same module as the data
 type `Text` itself.

 * (Better.)  Make `Data.Data.Constr` contain a Template Haskell `Name` for
 the data constructor, rather than just a `String`.  Doing that would mean
 moving `Language.Haskell.TH.Syntax.Name` (and associated types and
 functions) to package `base`.  This would make sense to me; class `Data`
 is really about meta-programming.  The `constring` record selector could
 be come an ordinary function, so almost all programs would work fine.

 Any opinions.?

--
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/12596#comment:4
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler

    