#5700: TH: InlinePs inside InstanceD are handled inconsistently
--------------------------------+-------------------------------------------
Reporter: mikhail.vorozhtsov | Owner:
Type: bug | Status: new
Priority: normal | Component: Template Haskell
Version: 7.3 | Keywords:
Os: Unknown/Multiple | Architecture: Unknown/Multiple
Failure: None/Unknown | Testcase:
Blockedby: | Blocking:
Related: |
--------------------------------+-------------------------------------------
Consider the following modules:
{{{
module Base where
class C a where
inlinable :: a -> ()
}}}
{{{
{-# LANGUAGE TemplateHaskell #-}
module TH where
import Language.Haskell.TH
import Base
mkC :: Name -> Q [Dec]
mkC n = return
[InstanceD [] (AppT (ConT ''C) (ConT n))
[ FunD 'inlinable [Clause [WildP] (NormalB (ConE '())) []]
]
]
}}}
{{{
{-# LANGUAGE TemplateHaskell #-}
module Use where
import TH
data D = D
$(mkC ''D)
}}}
which compile just fine with GHC. Lets throw in an InlineP pragma after
the FunD:
{{{
PragmaD (InlineP 'inlinable (InlineSpec True False Nothing))
}}}
Sadly, this results in an error:
{{{
Use.hs:9:3:
`Base.inlinable' is not a (visible) method of class `Base.C'
}}}
or, with 7.2.1:
{{{
Use.hs:9:3:
The INLINE pragma for `Base.inlinable'
lacks an accompanying binding
}}}
OK, another try with
{{{
PragmaD (InlineP (mkName "inlinable") (InlineSpec True False Nothing))
}}}
Still a no-go in 7.3.20111130 (but accepted by 7.2.1):
{{{
Use.hs:9:3: `inlinable' is not a (visible) method of class `Base.C'
}}}
Finally, I was able to satisfy GHC by importing Base into Use.
I think that names in FunD and PragmaD should be treated uniformly (i.e.
using 'inlinable works/fails in both cases. I'd prefer it works, without
importing Base).
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5700>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
