g77/gfortran incompatibilities

[Ed Hill]

On Fri, 2005-12-09 at 17:44 +0100, Patrice Dumas wrote:
> Hello,
> 
> I am currently annoyed by the gfortran/g77 incompatibilities. Indeed gfortran
> has still some regressions with regard with g77, especially some missing 
> intrinsics. Some are listed in the g77 regressions meta-bug here:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19292
> some are not listed (getarg, imag).
> 
> So there is still a need for g77 to build some code (for example the cernlib
> in fedora extras). Now the trouble is that code compiled with g77 and 
> gfortran are not binary compatible. I have been told that compiling using 
> g77 with -fno-f2c leads to compatibility with gfortran. Is it the case?
> 
> If it is the case, I think that all the libraries compiled with g77 in 
> fedora extras (is there any fortran in fedora core?) should be compiled
> with  -fno-f2c with the issue that all the code to be compiled against 
> those libraries should also be compiled with -fno-f2c. Does this make sense?

Hi Patrice,

Thanks for pointing this out!  I was aware of the incompatible name-
mangling between g77 and gfortran but wasn't aware of the -fno-f2c flag.
It does sound like a good idea provided the code generated on either end
really is compatible.  Could one of the Fedora compiler gurus please
step in and say whether this is a good idea or not?

Also, for the netcdf-devel package, I've been providing two sets of
libs: one compiled with g77 and one with gfortran.  I realize that its a
bit of an annoyance for a packager but I was trying to do everything I
could to make it easy for end users who might still be stuck on g77 for
various reasons.

Ed

-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
             Cambridge, MA 02139-4307
emails:  eh3 at mit.edu                ed at eh3.com
URLs:    http://web.mit.edu/eh3/    http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464