Re: fluka2011.2.4 (respin of fluka2011.2), gfortran version as well

From: Joseph Comfort <Joseph.Comfort_at_asu.edu>
Date: Mon, 4 Jul 2011 10:07:21 -0700 (MST)

Hi Alfredo,

There is another little problem with ffluka.sh, pertaining to SuSE.
The GFVER variable in line 85 is not obtained correctly for gfortran
because 'gfortran --version' yields
GNU Fortran (SUSE Linux) 4.5.1 20101208 [gcc-4_5-branch revision 167585]

'g77 --version' works because '(SUSE Linux)' is '(GCC)' instead.

A general solution may be difficult. Perhaps 'gfortran -dumpversion' is
sufficient. It yields the single line '4.5' in my case. However,
'g77 -dumpversion' gives the full single liine as in 'g77 --version'.
(Note that in line 18, there is no difference in results between
'--version -dumpversion' and '--version', for either gfortran for g77,
so the line works in both cases.)

In one of my scripts, I compile (in-line) a short code and do a
'cpp -dM' on the file to find the version number that was used. It seems
to work.

One other small change in lflukac would be helpful. The setting of
X11PATH would be better done after lllib is set up. Then the relevant
line would read
    X11PATH=/usr/${lllib}
Most libraries are in .../lib64 (except that my cernlibs are in .../lib).

(Also, my cernlib is in a different location and does not include either
nsl or X11, so other adjustments have to be made. But I always had to do
these in the past, so not an issue.)

Thank you,
Joe

P.S. Tests so far are looking good. Am I correct that the new 32-bit and
64-bit versions are unlikely to have the same random number sequence, and
therefore will not have the same numbers in detail?

On Mon, 4 Jul 2011, Alfredo Ferrari wrote:

> Thanks Joe
>
> the MACHNAME variable was an attempt to make the script Mac OS compatible,
> clearly it made compatible between Fedora and Mac Os, but it broke SUSE...
> I'll try to devise a better way. Thanks for the remark!
>
> Alfredo
>
>
> +----------------------------------------------------------------------+
> | Alfredo Ferrari || Tel.: +41.22.76.76119 |
> | CERN-EN/STI || Fax.: +41.22.76.69474 |
> | 1211 Geneva 23 || e-mail: Alfredo.Ferrari_at_cern.ch |
> | Switzerland || |
> +----------------------------------------------------------------------+
>
> On Sun, 3 Jul 2011, Joseph Comfort wrote:
>
>> Alfredo,
>>
>> Very good news about the 64-bit version. Not having it has been very
>> awkward.
>>
>> Can you please elaborate on the co-existence of both the 32-bit and 64-bit
>> versions? Separate directories, or one? Does the user choose through
>> the environment variables?
>>
>> To test, I installed the 64-bit version in a clean /usr/local/fluka. I
>> also set FLUFOR to 'gfortran'. I could not build the executable;
>> libflukahp.a is not compatible.
>>
>> The problem was that $MACHNAME, used in ffluka.sh, is set to
>> 'x86_64-suse-linux' in SuSE. Instead, I uncommented the line
>> 'CPUARCH=`uname -m` and commented out the next line. This may be a better
>> default feature of the script.
>>
>> I'll start to test the code.
>>
>> Thank you,
>> Joe Comfort
>>
>
Received on Mon Jul 04 2011 - 22:36:16 CEST

This archive was generated by hypermail 2.2.0 : Mon Jul 04 2011 - 22:36:26 CEST