-fdefer-type-errors

17 Apr 2019

      Hi,

I'm trying to use the -fdefer-type-error flag, discussed in  an ICFP paper from 2012:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.366.841&rep=rep1&type=pdf http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.366.841&rep=rep1&type=pdf 

On page 2, the paper has an example, showing that fdefer-type-errors will allow a the first element to be extracted fro a tuple, even if the second element does not correctly typecheck:

 ghci> let foo = (True, ’a’ && False)
 Warning: Couldn’t match ‘Bool’ with ‘Char’
 ghci> :type foo
 (Bool, Bool)
 ghci> fst foo
 True
However, with GHC 8.2.2, I get an error when trying to do this:
GHCi, version 8.2.2: http://www.haskell.org/ghc/  :? for help
Prelude> let foo = (True, 'a' && False)

<interactive>:1:18: warning: [-Wdeferred-type-errors]
    • Couldn't match expected type ‘Bool’ with actual type ‘Char’
    • In the first argument of ‘(&&)’, namely ‘'a'’
      In the expression: 'a' && False
      In the expression: (True, 'a' && False)
Prelude> :type foo
foo :: (Bool, Bool)
Prelude> fst foo
*** Exception: <interactive>:1:18: error:
    • Couldn't match expected type ‘Bool’ with actual type ‘Char’
    • In the first argument of ‘(&&)’, namely ‘'a'’
      In the expression: 'a' && False
      In the expression: (True, 'a' && False)
(deferred type error)

Does anyone know the reason for the change in the behavior?  Is there any way to get back to the old behavior, i.e. allow the first element of the tuple to still be extracted?  Ideally, I would like to get the errors given by defer-type-errors to be as fine-grained/close-to-the-source as possible.

Thanks,
Bill Hallahan

Bill Hallahan

Tom Ellis

Artem Pelenitsyn

Richard Eisenberg

Bill Hallahan

tags

participants (4)