[Rivet] segmentation fault with pythia8

[Andy Buckley]

Lars Sonnenschein wrote:
> I'm afraid I don't get that far with the head version.
> For some reason my specified AGILE_GEN_PATH variable gets ignored
> and from the LCG generators I get error messages like:
> 
> [lxbuild021] ~/scratch0/cedar/AgileCommit > rivetgun -g Pythia6:413 -a 
> CDF_2007_S7057202 --beam1 PROTON --beam2 ANTIPROTON --mom1 980. --mom2 
> 980. -P ../rivetgun_params/fpythia-TuneApthatmin30.params
> RivetGun.Main: INFO  Requested generator = Pythia6:413
> rivetgun: symbol lookup error: 
> /afs/cern.ch/sw/lcg/external/MCGenerators/pythia6/413/slc4_ia32_gcc34/lib/libpythia6.so: 
> undefined symbol: s_cmp

I've never seen that error... anyone else? Google says:

http://www.google.co.uk/search?q=s_cmp

The first entry mentions f2c, so I suspect that's to blame. Can you look
into this, and also try building it with gfortran rather than g77?

> (Same for Herwig. Pythia8 does not show up as available, but that's 
> because I configured with my private Pythia8 installation, I assume)

The AGILe loader insists on a Genser-type path structure for loading
generators, so if you want it to work, you'll have to install your
private copy into a private Genser area, and list that extra area in
your AGILE_GEN_PATH.

Andy

-- 
Dr Andy Buckley
Institute for Particle Physics Phenomenology
Durham University
0191 3343798 | 0191 3732613 | www.insectnation.org