Hoi Ric, I realized one drawback (for b->s gamma, not us): Filtering on the tagbits precludes dumping all true signal events. This was a point specifically raised by Henning. Is there the possibility to filter on the truth block? Of course we can skip the tagfilter business and just save diskspace by dumping events with either reco'ed lepton or photon true b -> s gamma true b->u l nu (?) I am not 100% familiar with the terminology of "PID weighting", but even PID killing is fine as long as we have data/MC PidTables. Another drawback is brem recovery if the cut-off is high. But I think the diskspace gains would be high enough even when requiring a lepton with p*>0.5GeV (i.e. effectively the same cutoff as now). Cheers, --U. Riccardo Faccini writes: > Hi Everybody, > regardless of the availability or not of the hbook disk space, I think we > need to seriously consider the option of saving only events with an > energetic lepton or photon. I would also consider applying a purity cut > (looser than the final one) at the ntuple generation level. > > I would say that, assuming we use PID weighting for PID systematics, the > semileptonic and b->s gamma analyses would not be affected. > > While I test the filter to do this, does anybody see any drawback on such > a decision? > ciao > ric >