[GHC] #834: bogus gcc option used when compiling ghc-6.4.2

[GHC]

#834: bogus gcc option used when compiling ghc-6.4.2
-------------------------------------+--------------------------------------
  Reporter:  mvanier at cs.caltech.edu  |          Owner:         
      Type:  bug                     |         Status:  closed 
  Priority:  normal                  |      Milestone:         
 Component:  Compiler                |        Version:  6.4.2  
  Severity:  trivial                 |     Resolution:  invalid
  Keywords:                          |             Os:  Linux  
Difficulty:  Easy (1 hr)             |   Architecture:  x86    
-------------------------------------+--------------------------------------
Changes (by dcoutts at gentoo.org):

  * resolution:  => invalid
  * status:  new => closed

Comment:

 This bug is only occurs on Gentoo because of options used in older
 versions of Gentoo's ghc-bin ebuild.

 See Gentoo bug [http://bugs.gentoo.org/show_bug.cgi?id=129891#c2 #129891].

 It is for reasons like this that we recommend Gentoo users to report bugs
 in Gentoo's bugzilla before bothering upstream developers. :-)

 The real cause of this bug is that supporting a wide range of versions of
 gcc is rather tricky. We had tried to support the 'hardened' gcc versions.
 'Hardened' gcc automatically builds with -pie -fstack-protector but these
 options cause gcc to produce output that ghc does not understand (ie it
 breaks every via-C compile). So we tried using -nopie and -fno-stack-
 protector. This made everything work for vanilla gcc-3.x and hardened
 gcc-3.x. Then it turned out that gcc-4.x didn't support the same options.
 So we made the ebuild only add the options if they were supported. Sadly
 that did not work for the -nopie option since gcc doesn't reject it with
 an non-zero return code, it just prints a warning. This has subsequently
 been fixed but it explains why you get the warnings - because the version
 of the ghc ebuild you started with had that issue when used with gcc-4.x.

 So the solution is to do just what you have done, to rebuild ghc using the
 latest version of the ebuild. The version you build should not have the
 problem when building anything else.

 So I'm closing this bug as invalid, because it's not an upstream bug.

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/834>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler