[Haskell] Latest Haskell Platform for Windows (2010.1.0.0) does not seem to include C++ support

[Jeremy O'Donoghue]

Hi list(s),

(Apologies for cross-posting, but likely to be of interest to both
lists)

As lack of C++ support on latest Haskell Platform prevents wxHaskell
cabal install for Windows, and may well be causing issues for others, I
present a vile hack which will get you some level of working C++
support. This has been verified to work on a Windows 7 machine.

You will need a clean MinGW 3.4.5 installation (I used the current
Windows MinGW installer to get this, but many of you probably have MinGW
lying around anyway) as a source for the files below:

Copy cc1plus.exe from a MinGW 3.4.5 install into c:\Program Files
(x86)\Haskell Platform\2010.1.0.0\mingw\libexec\gcc\mingw32\3.4.5
Copy libstdc++.a from a MinGW 3.4.5 install into c:\Program Files
(x86)\Haskell Platform\2010.1.0.0\mingw\lib
Copy the entire contents of include\c++ directory from a MinGW 3.4.5
install into c:\Program Files (x86)\Haskell
Platform\2010.1.0.0\mingw\include\c++

At this point you have a sufficient C++ environment to cabal install wx
and get a working wxHaskell installation. There are a number of files
missing from a complete MinGW C++ installation, so I don't guarantee
this as working for all C++ code.

On Windows 7, you will need to do your 'cabal install wx' in a command
window running as Administrator. 'cabal install wx --user' does not work
in a normal shell. I'll look into this later.

Regards
Jeremy

On Tue, 04 May 2010 22:18 +0100, "Neil Mitchell" <ndmitchell at gmail.com>
wrote:
> Hi
> 
> >> On checking, Haskell Platform\2010.1.0.0\mingw\libexec\gcc\mingw32\3.4.5
> >> does not contain cc1plus.exe. Previous versions of the platform have
> >> done so. Is this an accidental change or a deliberate policy decision?
> >
> > ghc-6.12.1 on Windows did not include the mingw C++ compiler.  This was a
> > mistake.  It is included once more in ghc-6.12.2.
> 
> ghc-6.12.1 didn't include cc1plus.exe or libstdc++.a. The lack of
> libstdc++.a caused a failure in something I was building, so I raised
> a bug and it's now included in 6.12.2. I didn't explicitly compile any
> C++ code, so it's possible that still doesn't work.
> 
> It wasn't a policy decision, merely an accident, and I'm sure if you
> alert people it will be rectified (since otherwise it's a regression).
> 
> Thanks, Neil
> 
-- 
  Jeremy O'Donoghue
  jeremy.odonoghue at gmail.com