[Rivet] Getting rid of /REF

Andy Buckley andy.buckley at cern.ch
Tue Nov 17 21:50:08 GMT 2015


Hi Riveters,

Apologies if this is stating the bloody obvious, but I realised this 
evening that the ability to set arbitrary annotations on YODA data 
objects means that we can think about using a nicer way to declare that 
an analysis object is reference data.

This would be lovely, because at present one of the things that makes 
code like rivet-cmphistos so messy is the identification, maybe removal, 
and subsequent tracking of the /REF path prefix. This is manually coded 
in umpteen places. I think we even already turn it into an annotation in 
some places. If we could move to the ref data having the exact same 
paths as the MC output, but with extra IsRef annotations on the objects, 
all this can be swept away.

There may well be external code written around Rivet which depends on 
the /REF path prefix, so I think we would have to make the change 
gradually. Specifically, I think we can add the annotation to our 
bundled ref data in the next 2.x series release and make sure that 
HepData's output contains it. Then we can make the big path change to 
all our ref data in version 3.0.0.

Sound good?

Andy

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


More information about the Rivet mailing list