[Rivet] Suggestion: change to compare-histos default behaviour

[Andy Buckley]

On 27/07/10 19:41, Frank Siegert wrote:
> Hi Andy,
> 
> I'm not worried about backwards compatibility in this case, but I am 
> worried about analyses where the refdata has a lot of histograms that 
> aren't filled by the MC. In those cases, with the --all option by 
> default, they will be plotted without any MC lines in them. Just think 
> of correlation matrices in some of the aida files. I've been discussing 
> this with David and Hendrik the other day, and we thought the neatest 
> solution would be to (by default?) plot histos if they contain only one 
> MC line, but not if they contain only the data line.
> I think Hendrik wanted to implement this within compare-histos, I have 
> followed in the last days though whether he has yet.

I thought that the implementation already looked like this... I
certainly never intended --all to be quite so literal ;) I certainly now
always use compare-histos with the -R option which automatically looks
up the Rivet ref data entries corresponding to the MC histos provided on
the command line: this reduces the problem: only the ref histos which
match those in the MC files are considered. It shouldn't be hard to
generalise this matching to also exclude REF histos with no MC
counterpart (unless overridden).

The compare-histos code has historically been a real mish-mash, as we
were developing the *ideas* of how best to structure the analyses and
post-processing as (and often after) we developed the code! For some
reason compare-histos was the main battleground for our own lack of
clarity. Anyway, I think it's cleaner now than it's ever been, though,
so it's a good time to make the change.

Thanks for the feedback,
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.