[Rivet] Warning: change to analysis loader!

[Andy Buckley]

Frank Siegert wrote:
> And one more, sorry:
> 
> When running Rivet now, I always get two warnings from ZEUS_2001_S4815815. 
> The first one was easy to fix, just syntax in the .info file.
> But the second one makes me wonder, because it comes from the constructor 
> of that analysis: Do all constructors get called now, irrespective of 
> whether the analysis is asked for? In that case, shouldn't we slim down 
> the constructors and move the projection initialisation to the init 
> method?

I'm glad you're thinking that: I've been thinking exactly the same 
thing! Yes, I agree, but one step at a time: I've identified a bug in 
the projection handler relating to addresses of destructed analyses 
being reallocated, and I want to get that sorted before shifting 
anything around any more. I'm hoping that this will be the last 
iteration of the infrastructure, barring the histograms (on which there 
will be some effort soon, but spontaneous volunteers who hate the 
current histogram classes are more than welcome...)

Andy

-- 
Dr Andy Buckley
Particle Physics Experiment Group, University of Edinburgh