gfortran link failure in current llvm svn

Chris and Bill,
   I have tested the proposed patch from...

http://lists.cs.uiuc.edu/pipermail/llvmdev/2008-August/016490.html

under i686-apple-darwin9 and it solves the problems building gfortran
from llvm svn. The resulting compiler works fine so can we get that
patch in before 2.4 is release?
           Jack
ps We do have one oddity left in llvm-gfortran from current llvm
svn. I find everytime I compile something with llvm-gfortran that
I get a series of warning messages...

f951: warning: command line option "-Wformat" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wformat-security" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-format-extra-args" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-format-zero-length" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-nonnull" is valid for C/C++/ObjC/ObjC++ but not for Fortran

Is this something we can suppress easily for the 2.4 release?

Chris and Bill,
  I have tested the proposed patch from...

http://lists.cs.uiuc.edu/pipermail/llvmdev/2008-August/016490.html

under i686-apple-darwin9 and it solves the problems building gfortran
from llvm svn. The resulting compiler works fine so can we get that
patch in before 2.4 is release?

I applied it to mainline.

ps We do have one oddity left in llvm-gfortran from current llvm
svn. I find everytime I compile something with llvm-gfortran that
I get a series of warning messages...

f951: warning: command line option "-Wformat" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wformat-security" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-format-extra-args" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-format-zero-length" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-nonnull" is valid for C/C++/ObjC/ObjC++ but not for Fortran

Is this something we can suppress easily for the 2.4 release?

Ick! Those are enabled in the config/i386/darwin.h file. I don't know
how to turn them off for Fortran et al. Anyone else know?

-bw

Sorry, all of this is too late for 2.4.

-Tanya

Actually this does qualify as a regression fix since the gfortran built on darwin
in llvm 2.3 (unless llvm.org is going to claim that no other compilers than c, c++
and objc matter).
          Jack

Chris and Bill,
I have tested the proposed patch from...

http://lists.cs.uiuc.edu/pipermail/llvmdev/2008-August/016490.html

under i686-apple-darwin9 and it solves the problems building gfortran
from llvm svn. The resulting compiler works fine so can we get that
patch in before 2.4 is release?

I applied it to mainline.

ps We do have one oddity left in llvm-gfortran from current llvm
svn. I find everytime I compile something with llvm-gfortran that
I get a series of warning messages...

f951: warning: command line option "-Wformat" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wformat-security" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-format-extra-args" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-format-zero-length" is valid for C/C++/ObjC/ObjC++ but not for Fortran
f951: warning: command line option "-Wno-nonnull" is valid for C/C++/ObjC/ObjC++ but not for Fortran

Is this something we can suppress easily for the 2.4 release?

Ick! Those are enabled in the config/i386/darwin.h file. I don't know
how to turn them off for Fortran et al. Anyone else know?

-bw

  Actually this does qualify as a regression fix since the gfortran built on darwin
in llvm 2.3 (unless llvm.org is going to claim that no other compilers than c, c++
and objc matter).

This does not qualify as a regression because we were not testing gfortran as criteria for previous releases. I've discussed this with Duncan and we will have a plan to incorporate this into the testing process for 2.5. Please see PR2909.

We do plan to support gfortran but its not criteria for this release (or has been for any previous release).

Thanks,
Tanya

Unfortunately, it is too late for the 2.4 release, however, now is a great time to fix it for 2.5. Once it is fixed in mainline, it would be great to get a tester set up to verify that we don't regress on this,

-Chris

Jack,

Try this not so elegant and untested patch.

I'd not expect this to work for Ada, java or Pascal.

It is obvious, I know. I encourage you to provide a better patch.

Devang,
    The patch isn't suppressing the errors.
                Jack