[Rivet] FastJet and C++11 again

[Andy Buckley]

Hi FastJetters,

Several months ago I was in touch to ask if it would be possible to 
issue a patch release of FastJet to "fix" the slew of compiler warnings 
that appear when FastJet -- or, crucially, code which includes its 
headers -- is compiled with GCC using the C++11 standard.

These warnings are arguably overzealous, and I agree with the arguments 
made to add a suppression flag in this support request:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59325
But the developers seem to have little sympathy for that argument and so 
we are stuck with extremely verbose compiler output every time one of 
the FastJet uses of auto_ptr is spotted.

The reason I'm bringing this up again is that previously this was just 
an internal annoyance for the developers of a project that I was working 
on, but now we would like to make the switch to mandatory C++11 
compilation of Rivet and feel that a slew of misleading FastJet warnings 
very time a user calls rivet-buildplugin is a blocker for that change.

If I remember correctly you are working on a more major development and 
would rather not spend time on maintaining the current branch. But 
unless that new version is going to appear pretty soon, a bugfix would 
be much appreciated -- have the experiments also reported this as an issue?

Best wishes,
Andy & Rivet

PS. I can't resist a technical comment. There are several possible 
solutions, but on https://gcc.gnu.org/wiki/Cxx11AbiCompatibility
it's made clear that mixing C++ standards in a single project is a very 
bad idea, so maybe a (good?) assumption can be made that user code will 
never compile different units against FastJet with the same symbol being 
declared as an auto_ptr in one and as a unique_ptr in another. If so, 
then a simple "#if C++11 ... #else ..." macro solution would not be 
crazy. Given that the incompatibility is at ABI level, harder-work 
approaches like using the PIMPL idiom to hide the smart pointers from 
the public headers probably wouldn't help. Or you could also decide to 
make C++11 mandatory ;-)

-- 
Dr Andy Buckley, Lecturer / Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow