Hi Marian, This is not a bug. By definition, one cannot make an instance of an abstract class. TTree::Branch was not correctly handling this case in the early 3.01 versions. This problem was fixed a few weeks ago in the CVS version. Rene Brun On Thu, 6 Sep 2001, Marian Ivanov wrote: > Hi Rooters, > > It seems that there is a problem in the new ROOT versions (3.01/2): > > the TClass::BuildRealData() just return when sub-classing of a > > abstract class is done. So that TClass does not contain any > > information about the layout of data members and as consequence > > the IO is corrupted. Changing the abstract class to non-abstract > > solve the problem, the layout is then built correctly. > > Of course it is just work around and a better solution to enable > > subclassing of abstract class. > > Commenting line 453 (in TClass::BuildRealData()) also solve > > owr problems > // if ((!pointer) && (Property() & kIsAbstract)) return; MI change > > but I didn't make intensive test's of the code. > > Best regards > > Marian > > -- > ***************************************************************** > ** Marian Ivanov * > * * > ** KP1, GSI Internet: miranov@gsi.de * > ** Planckstr. 1 Tel.: 49-6159-712183 * > ** 64220 Darmstadt Fax: 49-6159-712989 * > * * > ** CERN -- European Laboratory for Particle Physics * > * Office: 892 1-B01 Tel.: +41 22 76 76505 * > * * > ** MFF UK, Bratislava * > ** Internet :ivanov@fmph.uniba.sk * > ***************************************************************** > > >
This archive was generated by hypermail 2b29 : Tue Jan 01 2002 - 17:50:59 MET