[Rivet] segmentation fault with pythia8

[Holger Schulz]

Andy Buckley wrote:
> Holger Schulz wrote:
>
>   
>> But if I switch on a process like
>>             rivetgun -g Pythia8:108 -n 1000 -P lep1.params -p 
>> "WeakSingleBoson:ffbar2gmZ=on"
>> there is a segmentation fault at the end after finalize() is called.
>>
>> here is some gdb output:
>>     
>
>   
>>         #1  0x9d21165b in AGILe::CCPythia::~CCPythia$delete () from 
>> /mnt/eeh/hschulz/Software/lib/libAGILeCCPythia.so
>>         #2  0xafee85aa in AGILe::Loader::destroyGen () from 
>> /users/eeh/hschulz/Software/lib/libAGILe.so.0
>>         #3  0x08053dba in Rivet::generate ()
>>         #4  0x080500ef in main ()
>>     
>
> Looks to me like a memory corruption, but I'm not sure how: there isn't
> even a pointer in that class that could get corrupted, nor an explicit
> "delete" call!
>
>   
>> Has anyone observed something similar, or is my installation  broken?
>>     
>
> I haven't tried yet ... Lars, do you see this? Any suggestions?
>
>   
>> Furthermore, it seems as if prof-tune doesn't handle statements like -p 
>> "WeakSingleBoson:ffbar2gmZ=on"
>> properly, since the output in out.log then gives lines as such:
>>
>>              PYTHIA Abort from Pythia::next: not properly initialized so 
>> cannot generate events             
>>              Event generation aborted prematurely
>> Reading these parameter settings from a file with -P doesn't work 
>> either. On the other hand,
>> the parameter values that shall be varied are handed over correctly.
>>     
>
> I'm assuming that this is a legacy of ":" being a valid param definition
> character. I'll look into it. Again, Lars, had you noticed this being a
> problem?
>
>   
Ok, now all parameters are handed over correctly, the events are being 
generated, but
still there is that segmentation fault at the end, so that sadly no aida 
files are being saved.
The gdb output is exactly the same as above :(

Holger




> Andy
>
>