[GHC] #10762: On Windows, out-of-codepage characters can cause GHC build to fail

[GHC]

#10762: On Windows, out-of-codepage characters can cause GHC build to fail
---------------------------------+-----------------------------------------
        Reporter:  snoyberg      |                   Owner:
            Type:  bug           |                  Status:  new
        Priority:  normal        |               Milestone:
       Component:  Compiler      |                 Version:  7.10.2
      Resolution:                |                Keywords:
Operating System:  Windows       |            Architecture:  x86_64 (amd64)
 Type of failure:  None/Unknown  |               Test Case:
      Blocked By:                |                Blocking:
 Related Tickets:                |  Differential Revisions:
---------------------------------+-----------------------------------------

Comment (by snoyberg):

 I think that all file output from GHC should be UTF-8, regardless of OS,
 and regardless of locale settings. I won't make too many arguments around
 console output, except that even on Windows, there should be some
 environment variable or command line switch to force UTF-8 output. As it
 stands, the only way to capture output reliably from GHC on Windows is to
 change the codepage for the entire console.

 And speaking of which: I won't profess to be a Windows expert myself, but
 from my rather painful research this morning: a console codepage is not
 something which is merely inherited by subprocesses (the way an
 environment variable is), but by other processes in the same console.
 Unless I'm misunderstanding something, the workaround we're using in stack
 now could have negative consequences for things like running stack/GHC
 from inside a text editor, where the text editor may suddenly have its
 code page changed on it.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10762#comment:2>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler