Hi Pasha, Yes, I agree that this feature should have been documented a bit better. We had hoped to not have to describe this option and to make it automatic by getting the information from CINT if a class has a custom Streamer or not. I am surprised by your statement about the factor 3. Could you send me an example of a class where you obtain this result? I take this opportunity to encourage people to migrate to the automatic Streamers. More details in teh Users Guide. Rene Brun On Tue, 17 Jul 2001, Pasha Murat (630)840-8237@169G wrote: > Gerco Onderwater wrote: > > > > Hi Rene, > > > > As a matter of fact, we just figured out that splitlevel 0 will do the > > job. A value -1, as you suggested, produces the same results as kTRUE > > (which is indeed incorrect). > > > > On Tue, 17 Jul 2001, Rene Brun wrote: > > > > > Hi Gerco, > > > > > > TPolyLine3D is one of the few ROOT classes still with a hand-written Streamer. > > > Classes with a hand-written Streamer cannot be used in split mode in a branch. > > > You can force the branch to not use the split mode and call the Streamer > > > function by using split=-1. We expect this to be automatic in the next release. > > > So, in your example, change the line: > > > > > > Dear ROOTers, > > I think that the fact that the meaning of 'split' parameter has changed > in 3.01 needs more advertizing. Much more. Everybody using customized Streamers > should be warned - this change may affect people in a severe way. A brief comment > in the release notes is not enough. > We'are using hand-written Streamers and I just spent few hours of > debugging to figure that after switching to ROOT 3.01/06 we were supposed to > change split=0 to split=-1 in the depths of the code, luckily this was caught > during the testing. > -best, Pasha > > P.S. In case of no compression customization of Streamers can significantly improve > the I/O performance (I saw improvements by a factor of 3), so it seems to be a > very reasonable option to consider. >
This archive was generated by hypermail 2b29 : Tue Jan 01 2002 - 17:50:52 MET