[GHC] #1527: parsing error message that could be improved

[GHC]

#1527: parsing error message that could be improved
-------------------------+--------------------------------------------------
    Reporter:  guest     |        Owner:         
        Type:  bug       |       Status:  new    
    Priority:  lowest    |    Milestone:         
   Component:  Compiler  |      Version:  6.6.1  
    Severity:  minor     |   Resolution:         
    Keywords:  parser    |   Difficulty:  Unknown
          Os:  Unknown   |     Testcase:         
Architecture:  Unknown   |  
-------------------------+--------------------------------------------------
Old description:

> This is just barely a bug.  The following code uses do notation and gets
> a left arrow in the wrong direction:
>
> tcRho (TupleE exps) (Check exp_ty)
>    = do { tys <- unTupleType exp_ty
>         ; checkTuple (text "term" <+> ppr (TupleE exps)) exps tys
>         ; exps' -> mapM (uncurry checkRho) (zip exps tys)
>         ; return $ ftuple exps'
>         }
>
> The error message is: TxPgm.hs:323:44: Parse error in pattern
>
> The suggestion of 'pattern' is surprising here, and so is the position of
> the error message.  I know parsing is a black art, and I don't know what
> technology GHC is using, but if this message could be improved that would
> be a Good Thing.
>
> Norman

New description:

 This is just barely a bug.  The following code uses do notation and gets a
 left arrow in the wrong direction:

 {{{
 tcRho (TupleE exps) (Check exp_ty)
    = do { tys <- unTupleType exp_ty
         ; checkTuple (text "term" <+> ppr (TupleE exps)) exps tys
         ; exps' -> mapM (uncurry checkRho) (zip exps tys)
         ; return $ ftuple exps'
         }
 }}}

 The error message is: TxPgm.hs:323:44: Parse error in pattern

 The suggestion of 'pattern' is surprising here, and so is the position of
 the error message.  I know parsing is a black art, and I don't know what
 technology GHC is using, but if this message could be improved that would
 be a Good Thing.

 Norman

Comment (by Isaac Dupree):

 Well, exps' is supposed to be a pattern there, so mentioning "pattern"
 makes sense to me... given the possibilities, it might be more accurate to
 say "parse error in pattern or expression" ("... or let", to be
 comprehensive about do-notation possibilities).

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/1527>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
-------------- next part --------------
_______________________________________________
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs at haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs