[Rivet] Fwd: Running RAPGAP with RIVET

[Holger Schulz]

Andy Buckley wrote:
> On 15/04/10 15:11, Tim Martin wrote:
>   
>> I guess so, I've tried other versions of Pythia through changing the...
>>
>> baselibs.push_back(GenLibInfo("libpythia6", "pythia6", "419"));
>>
>> ...line but no luck.
>>     
>
> I've just rewritten some chunks of that code to make AGILe (the SVN
> head) try all versions of Pythia. It shouldn't change the symbol content
> of the libraries though: that bit doesn't change between versions. In
> particular, PYTIME is a dummy routine because different FORTRAN
> implementations provide different extensions for getting date-time
> information.
>
> Okay, here's some examples of where pytime_ is defined: it's undefined
> in Genser's libpythia.so, which is why you need it:
>   
Ah, ok, my libpythia was compiled from the source files I
grapped from the pythia homepage.
> andy at duality:~/proj/hep/agile$ nm
> /home/andy/genserarea/pythia6/422.2/i686-Linux-gcc44-opt/lib/libpythia6.so
> | grep pytime
>          U pytime_
>
> (U="undefined")
>
> but it is defined in libpythia_dummy.so, which is why AGILe works for
> the normal Pythia, which loads the dummies library:
>
> andy at duality:~/proj/hep/agile$ nm
> /home/andy/genserarea/pythia6/422.2/i686-Linux-gcc44-opt/lib/libpythia6_dummy.so
> | grep pytime
> 00000e70 T pytime_
>
> (T="text")
>
> I'm assuming that Charybdis, AlpGen and other generators that piggyback
> on Pythia also supply their own pytime_ dummy, since none of the AGILe
> libraries currently provide (or even know about) it:
>
> andy at duality:~/proj/hep/agile$ for i in
> /home/andy/heplocal/lib/libAGILe*.so; do nm $i | grep pytime; done
> andy at duality:~/proj/hep/agile$
>
> Can you try running
> nm path/to/librapgap.so | grep pytime
> to see if it defines this symbol? Maybe we should be loading
> libpythia_dummies as well... I've added it in commented-out form on line
> 368 of the src/Core/Loader.cc file (in the SVN head), so you could try
> updating from SVN, uncommenting, recompiling and reinstalling and see if
> it helps. Maybe get a different error message, at least ;)
>
>  Also tried your suggestion of piping rapgap
>   
>> straight into Rivet but that sadly did not work either.
>>     
>
> Weird. My only guess is that it isn't writing HepMC... maybe its own
> format, maybe a dump of the HEPEVT common block?
>   
Exactly.
> Andy
>
>