[Rivet] new analysis

[Frank Siegert]

Hi Sparsh,

Sparsh Navin, Thursday 17 June 2010:
> While writing an analysis for ALICE, Anton Karneyeu and I came across
> something different to what is says on the page
> http://projects.hepforge.org/rivet/trac/wiki/WritingAnAnalysis.

I'm afraid this webpage is also very much out of date, apologies for 
that. I think the best option is to use the plugindemo directory 
distributed with Rivet as an example. It does contain a Makefile as 
well, but there is even a more convenient option to compile your 
analysis like:

  rivet-buildplugin RivetMyAnalysis.so MY_ANALYSIS.cc (see below)

> Right at the bottom, in the "Compiling and linking" section, it says
> that you need to have the word "Rivet" in the name of the library or
> it will not get loaded at run time. However, we noticed that the
> following the example there and in the section called "Running", my
> analysis didn't get listed until I changed the name of the library
> libRivetMyAnalysis.so to *begin* with the word "Rivet". Naming it
> libRivetMyAnalysis.so allowed it to be compiled but didn't list with
> "rivet --list-analyses".

Yes, that's indeed one of the changes for which the above page is out of 
date. Rivet*.so is correct since 1.2.0.

> Also g++ -o libRivetMyAnalysis.so -shared
> -I/path/to/installed/rivet/include MyAnalysis.cc didn't compile it.
> I had to use the attached Makefile.

Your Rivet installation directory is probably different from your HepMC 
and FastJet installation directory, that's why you need the extra -I 
flags. This should to some extent be automatic using the current Rivet 
trunk (I think not in release 1.2.1 yet) and its rivet-buildplugin 
script like:

  rivet-buildplugin RivetMyAnalysis.so ALICE_2010_S9999999.cc

But all of this should be documented on the page above, of course. I 
have created ticket #407 so that we don't forget about this again.

  http://projects.hepforge.org/rivet/trac/ticket/407

Thanks for pointing this out and please contact us right away if you 
have any other questions or find something that is documented as poorly!

Frank