Re: [GHC] #14106: Out of scope errors appear after type errors caused by them

18 Aug 2017

      #14106: Out of scope errors appear after type errors caused by them
-------------------------------------+-------------------------------------
        Reporter:  EyalLotem         |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.0.2
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Poor/confusing    |  Unknown/Multiple
  error message                      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------
Description changed by simonpj:

Old description:
...
Example:
import Control.Lens (_Just, (&))
  main = Just 5 & _Just .~ 100 & print
.~ is out of scope, the error output:
ghc-error.hs:2:17-21: error: …
    • Couldn't match type ‘Maybe’ with ‘p0 a0’
      Expected type: Maybe (f0 b0) -> p0 (Maybe a0) (f0 (Maybe b0))
        Actual type: p0 a0 (f0 b0) -> p0 (Maybe a0) (f0 (Maybe b0))
    • In the second argument of ‘(&)’, namely ‘_Just’
      In the first argument of ‘(.~)’, namely ‘Just 5 & _Just’
      In the expression: (.~) Just 5 & _Just 100 & print
ghc-error.hs:2:23-24: error: …
    • Variable not in scope:
        (.~) :: p0 (Maybe a0) (f0 (Maybe b0)) -> IO () -> t
    • Perhaps you meant ‘.’ (imported from Prelude)
      Perhaps you want to add ‘.~’ to the import list in the import of
      ‘Control.Lens’ (/home/eyal/devel/test/ghc-error.hs:1:1-32).
Compilation failed.
In larger examples, the out of scope error can be buried deep down.
In the case of operators - the fixity is unknown so it can even cause the
parse to go wrong - and very weird type errors to result from that.
Out of scope errors should be put BEFORE any type errors that might be
caused by them.
New description:

 Example:
 {{{
   import Control.Lens (_Just, (&))
   main = Just 5 & _Just .~ 100 & print
 }}}
 .~ is out of scope, the error output:
 {{{
 ghc-error.hs:2:17-21: error: …
     • Couldn't match type ‘Maybe’ with ‘p0 a0’
       Expected type: Maybe (f0 b0) -> p0 (Maybe a0) (f0 (Maybe b0))
         Actual type: p0 a0 (f0 b0) -> p0 (Maybe a0) (f0 (Maybe b0))
     • In the second argument of ‘(&)’, namely ‘_Just’
       In the first argument of ‘(.~)’, namely ‘Just 5 & _Just’
       In the expression: (.~) Just 5 & _Just 100 & print
 ghc-error.hs:2:23-24: error: …
     • Variable not in scope:
         (.~) :: p0 (Maybe a0) (f0 (Maybe b0)) -> IO () -> t
     • Perhaps you meant ‘.’ (imported from Prelude)
       Perhaps you want to add ‘.~’ to the import list in the import of
       ‘Control.Lens’ (/home/eyal/devel/test/ghc-error.hs:1:1-32).
 Compilation failed.
 }}}
 In larger examples, the out of scope error can be buried deep down.

 In the case of operators - the fixity is unknown so it can even cause the
 parse to go wrong - and very weird type errors to result from that.

 Out of scope errors should be put BEFORE any type errors that might be
 caused by them.

--

-- 
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/14106#comment:1
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler