[Rivet] Rivet on lxplus at CERN - clash with stdio ?

[Barra at stud.uni-heidelberg.de]

Quoting Andy Buckley <andy.buckley at cern.ch>:

> On 14/05/13 11:45, Hendrik Hoeth wrote:
>> Hi,
>>
>> Thus spake Antonin Maire (amaire at physi.uni-heidelberg.de):
>>
>>> Since last week, when one tries to connect to lxplus, one does not
>>> manage to compile any analysis.cc (home made like official ones
>>> already in the sources). [...] Maybe we have missed something obvious.
>>
>> I guess the "obvious" change was that on 6 May lxplus switched to SL6
>> (http://cern.ch/go/F7kM). That means the system compiler is not gcc-4.1
>> anymore, but gcc-4.4, which is newer than the LCG compiler (4.3).
>> Apparently that made /usr/include/stdio.h incompatible. Now don't ask me
>> why the compiler tries to access that include file in the first place ...
>
> If I remember correctly, that setup script that you are sourcing is from
> my public directory and uses the SLC5 (not 6) builds of Rivet, GCC, etc.
>
> To make it work, either keep logging in to lxplus5, or copy and update
> that script in my public area. All it does is set up gcc43 and source
> the rivet setup script in the LCG area: for SLC6 you probably want the
> x86_64-slc6-gcc46-opt platform or similar.
>

Thanks for your fast answer.

I have edited the setupRivetProf.sh file in your public directory,  
like you said. In addition to your point of changing the platform, I  
had to change the path
as following:
/afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt64d...
to
/afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt64a...

My question is now: How does this effect me, running rivet or agile?

For your information, in the MCGenerators_lcgmt64a directory I  
couldn't find a version to run Professor on slc6. However this does  
not effect me.

Have a nice day
    Marc and Antonin, ALICE in Heidelberg