Hi Volker, We are currently making several mods in the I/O and Trees to make the system "thread-safe" (many thanks to Mathieu de Naurois for his help in this process). With these changes, we are removing several side-effects due to the global state. I introduced a new version in CVS this morning that (very likely) solves the problem you are reporting. I know that importing/testing a new version takes time, but it would be very nice if you could confirm this with your use case. You sent me a few mails about one month ago. I did not forget. Your points are on my todo list. Concerning the "advertizing" notice, I will see with Fons what is the problem. Rene Brun Volker Hejny wrote: > > Hi all, > > I'd just fought against some strange error in one of our > applications. The sympton was, that a tree->Draw() command > only gave an segmentation fault. On further investigation I > found, that when opening the file, root complained about > the missing dictionary only for the data classes, but not > for the underlying base classes as it was before. Then, > when reconstructing the streamer info of the data class, > the segfault happened. > > The reason for this was the following: After creation of the > tree there was (for some other reason) a directory->cd() to > another place. Then the branches have been created and so on. > After removing this command browsing the tree did work again. > > Since after creating the tree, the tree was only used by its > pointer, so I didn't expect that the current directory has to > the file at some point. For what action on trees it has to > be necessary to be in the file directory? > > I use the current cvs version (Saturday). > > Best regards, > Volker > > P.S.: Using the subject 'trees, branches and the current directory' > I got the error 'we do not accept advertising' from your > mailserver ... > > -- > Dr. Volker Hejny Tel: 02461/616853 ** > Institut f. Kernphysik Fax: 02461/613930 ** > ---------------------------------------------------------------- ** ** --- > Forschungszentrum Juelich GmbH, D-52425 Juelich **
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:50:42 MET