[petsc-dev] Fortran source format

[Satish Balay]

On Mon, 19 Apr 2010, Jed Brown wrote:

> On Sat, 23 Jan 2010 21:32:32 +0100, Jed Brown <jed at 59A2.org> wrote:
> > On Sat, 23 Jan 2010 14:19:41 -0600 (CST), Satish Balay <balay at mcs.anl.gov> wrote:
> > > I should have added - the gfortran I have access don't show this issue.
> > 
> > Yeah, I'm normally using 4.4.2, this just showed up on a test build with
> > a 4.5 snapshot.  If they prioritize this (I imagine it qualifies as
> > "regression fix" and is simple to fix) then it would be resolved in the
> > 4.5 release.
> 
> It wasn't, PETSc compilation and tests are noisy with gfortran-4.5.
> What does it take to check for this and add -Wno-line-truncation to the
> build?  It pains me to do this, but Steve Kargl didn't seem to think
> that allowing libraries to be dialect-agnostic was useful at all and
> Jerry DeLisle's patch wasn't completed/applied for 4.5.0, perhaps it
> will be in 4.5.1 whenever that comes.

The following change should add this flag automatically [have to make
sure it actually works]. If it works - I'll add it to petsc-31.


Satish
------------

diff -r 6e51c85213a3 config/compilerOptions.py
--- a/config/compilerOptions.py	Mon Apr 05 00:44:44 2010 -0500
+++ b/config/compilerOptions.py	Mon Apr 19 13:20:55 2010 -0500
@@ -177,7 +177,7 @@
     flags = []
     if config.setCompilers.Configure.isGNU(compiler):
       if bopt == '':
-        flags.extend(['-Wall', '-Wno-unused-variable'])
+        flags.extend(['-Wall', '-Wno-unused-variable', '-Wno-line-truncation'])
       elif bopt == 'g':
         if self.framework.argDB['with-gcov']:
           flags.extend(['-fprofile-arcs', '-ftest-coverage'])