Hi Dan, On 5 Mar 2003, Dan Krop wrote: > Hi, > Thank you for the solution and the fast response time. I would like to > point out 2 things: > > 1. The script which created the analysis class actually was, > { > > TChain chain("ntuple"); > chain.Add("ntup1.root"); // Ntuple with floats > chain.Add("ntup2.root"); // Ntuple with doubles > > ntuple.MakeClass("Analysis"); > } > > So, the reason floats was declared in the Analysis.h file was because > the 1st file had floats. It would be desirable to have TChain exit with > an error or at least warn the user when there is an incompatibility > between data types in its TFiles. This would imply looking at all the files in the TChain. This could be an option when calling MakeClass. > > 2. When I change the decalaration of branch types to doubles, the > analysis loop does not correctly see the floats in the 1st file. There > does not seem to be any automatic casting here. The automatic promotion or change of type is supported only with objects not for the code generated by MakeClass. Rene Brun > > Thanks again, > Dan > > On Tue, 2003-03-04 at 13:58, Rene Brun wrote: > > Hi Dan, > > > > I have investigated your problem and found the solution. > > Your first file has a TNtuple (with float variables). Your second file > > has a TNtupleD (with double variables). > > You have generated the header file via TTree::MakeClass on the TNtuple > > file. Then you use it on a TChain combining the two files. > > This is wrong. You should have done the inverse. > ><snip> > > Rene Brun > > > > On 4 > > Mar 2003, Dan > > Krop wrote: > > > > > Hi, > > > I tried 3.05/02 and the behavior is the same. I've made an even > > > simpler example which requires no external shared libs. It is on the > > > same website: > > > > > > http://dustbunny.physics.indiana.edu/~dkrop/root_problems > > > > > > Note that the location of the ntuple files must be correctly declared in > > > the header file before testing. I would be interested if anybody can > > > reproduce the behavior I have seen. > > > > > > Thanks again, > > > Dan > > > >
This archive was generated by hypermail 2b29 : Thu Jan 01 2004 - 17:50:09 MET