[Rivet] [Rivet-svn] r1768 - in trunk: include/Rivet src

[Frank Siegert]

Andy Buckley, Thursday 20 August 2009:
> The one thing that popped into my mind was that we are effectively
> "static-free" apart from the LWH histograms, since they write to one
> global histo tree. Did you find a way around that? 

Actually I didn't have any problems with it. I simply initialise multiple 
"AnalysisHandler"s with different "basefilename", and it works fine.

> The idea of writing
> multiple histograms for each analysis overlaps quite strongly with the
> multiple-weights implementation that James has been working on, and
> since both use-cases are driven by Sherpa maybe you can have a think
> about how to make this coherent? ;)

I had been thinking about it initially, but what I am doing now is 
different: I still fill one histogram per analysis, but I have multiple 
instances of each analysis (one per AnalysisHandler).
Maybe this could be done in a smarter way once the multiple-weights 
solution is implemented... my weight vector would essentially contain a 
lot of zeros and one weight. But it's not actually needed to achieve what 
I am trying to do.

Frank