Accessing TObj retrieved with FindObject gives sometm. error

ROOT
1

Hi,

(using: /afs/cern.ch/sw/lcg/external/root/5.14.00b/slc3_ia32_gcc323/root/)

I experience some bug in my pyroot-code… and I could reduce it to a small working (=actually not working, because producing an error) example.

At my computer the following subroutine runs fine about two or three times, but then it failes.

def test():
    running = 0
    canv = ROOT.TCanvas()
    for i in xrange( 50 ):
        ch = ROOT.TChain()
        ch.Add('rfio:///castor/cern.ch/grid/atlas/ctb2004/birks/ctb_Run2343_pid211_piplus_E50GeV_Evt10000_bA1._bk0.0486_QGSP_BERT-*.reco.root/TB/tree' )
        print 'cycle: ',i
        ch.Draw( 'Sum$(cl_e_topo)>>htemp()' )
        h = ROOT.gROOT.FindObject( 'htemp' )
        h.SetName( 'newname%s'%(running) )
        running += 1
        h.Draw()
        print 'getname: ',h.GetName()

I think, that the piece of code I use is correct and does only allowed things. Therefore I find it strange, that it works 2.5 times but not more.

thanks for your help,
Peter

2

Peter,

although I’m clueless about what your snippet of code is supposed to do, and whether it does what you think it should do, I can see about half a dozen ways how the internal bookkeeping and refcounting will go haywire.

A simple fix is to name your chain:[code]for i in xrange( 50 ):
ch = ROOT.TChain(‘mychain’)

etc. …[/code]

although that still wouldn’t make me happy (I’d prefer a ‘del ch’ at the end of the loop block as well, such that the ctor of the new chain object is called after the dtor of the previous, same-named, instance), it does make the script run. :slight_smile:

HTH,
Wim[/i]