4.3.3/4.4 Regression: the return of PR30285

[Harald Anlauf]

Hi,

I just learned that my old PR30285 returned. :-(
Somewhere in the 4.3.3 series between 20081221 and 20090115
the module writing mechanism was modified, giving me huge
module files for the project I am working on.  The same problem
occurs on current trunk.

If you compile the testcase gfcbug56.f90 attached to

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30285

and look at the mod*.mod files, you will find that the problem lies
in the treatment of variables in COMMON blocks.
If you work with OpenMPI, you can easily get module files which
are a factor of 100 (one hundred!) larger than they should be.
This is a show-stopper for me.

Should I reopen the above PR or open a new one?

Cheers,
Harald

-- 
Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger