[Rivet] Rivet 1.5.0

[Andy Buckley]

On 07/03/11 15:54, Frank Siegert wrote:
>> Second, when the analysis is run with any of the energies 900, 2360 and
>> 7000, it gives the error
>>
>> Rivet.Analysis.Handler: WARN  Analysis 'ALICE_2010_S8625980' is
>> incompatible with the provided beams: removing
>>
>> and seg faults
>>
>> ./runRivet.sh: line 114: 25323 Segmentation fault      $rivetExecString
> 
> Here is a follow-up question: Since the beam info restructuring I've
> also seen this *segfault* when running an incompatible analysis. Can
> somebody reproduce this? I looked at the code and it's using smart
> pointers, which I don't know good enough to debug them.

This is my personal nightmare. Something is wrong in the memory
management -- remember the segfaults that we got when we *fixed* the
smart-pointer comparison operator used in a std::set? -- and needs a
proper review, possibly with the assistance of CERN's Coverity (static
code analysis tool) license. LPCC/Genser might be able to help us with
that, as they did with the validation testing.

Fortunately this doesn't affect physics results, otherwise I would have
fixed it by now, but we do get segfaults when analyses/projections are
removed. It's a blocker bug for 2.0.0: while we've got Rivet jacked up
for an episode of Pimp My Analysis System, we may as well make the motor
run smoothly as well ;)

Andy

-- 
Dr Andy Buckley
SUPA Advanced Research Fellow
Particle Physics Experiment Group, University of Edinburgh

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.