Hi Jens Ivar After a discussion with Anders it became clear that maybe we want to do something more violent less hack like. His suggestion and one that Flemming has proposed before was to make a new class called BrTPCDetectorHit (derived from BrDetectorHit) where we could have information like pointers to all the substructures and probably row should fit in here if we want to make the BrDetectorHit more general, so it can be used for everything, Tof, DC, RHIC etc. As a first approach we want to make all this info //! but it is very possible that some would want to make an option for writing everything to disk (Sort of a debug option) and customize a streamer function. That way you could do the tracking and then write the cluster info as a sort of reduced datafile for debugging. What do you other guys think ? Cheers Peter :-) --------------------------------- )-: |Peter H L Christiansen aka PAN @ NBI | |EMAIL : pchristi@nbi.dk | |OFFICE : Tb1 @ NBI | |PHONE : 353 25269 | |SNAIL : Hans Tavsensgade 35, 4th | |PHONE : 35 349336 | :-D --------------------------------- \-: On Fri, 17 Nov 2000, Peter H. L. Christiansen wrote: > Hi Jens Ivar > > I don't know if we reached a conclusion. > > Maybe the nicest thing would be to have an ID number, that way you could > go whatever direction you wanted between BrTPCCluster, BrTPCHitCluster and > BrDetectorHit. > > Root has some stuff for finding objects in a container, called > FindObject(). The default is to use the default IsEqual() for comapring > objects which compares the address of the objects. > You can overwrite this and get some different behaviour by trying to cast > the pointer to the different classes and then compare ID. I have used the > IsEqual and FindObject in a very simple way in > BrTPCSequence/BrTPCClusterFinder::FindClusters. > > Maybe one could also use operator overloading of the ==, I have no feel > for this, but maybe someone has an idea. > > Anyway, I don't think anyone would think it is a bad idea if you > implement something;) > > A few thoughts > Peter > > :-) --------------------------------- )-: > |Peter H L Christiansen aka PAN @ NBI | > |EMAIL : pchristi@nbi.dk | > |OFFICE : Tb1 @ NBI | > |PHONE : 353 25269 | > |SNAIL : Hans Tavsensgade 35, 4th | > |PHONE : 35 349336 | > :-D --------------------------------- \-: > > On Thu, 16 Nov 2000, Jens Ivar Jordre wrote: > > > Morning BRATers of the world. > > > > I just wonder, did we reach any conclusion about how to match instances of > > BrDeterctorHit and BrTPCHitCluster? Nobody answered my last mail > > suggesting to assign the address of BrTPCCluster instance to fID member of > > its BrDetectorHit descendant. Does that mean that we agree upon this, or > > is someone still looking for a better solution? Has anyone implemented > > something to be committed to CVS? Just trying to avoid CVS conflicts. :) > > > > JI > > > > --- > > Jens Ivar Jřrdre, Dep. of Phys., Allégt. 55, N-5007 BERGEN, NORWAY > > room 521, e-mail: JensIvar.Jordre@fi.uib.no, phone: (+47) 55 58 27 92 > > > > > >
This archive was generated by hypermail 2b29 : Fri Nov 17 2000 - 09:41:45 EST