Minutes of the Offline Meeting
July 7th, 1998
Present: M.Cattaneo, P.Binko, P.Teixira-Dias, A.Flavell, E. van Herwijnen, P.Colrain, M.Frank, G.Wilkinson, D.Liko, T.Ruf, G.Gracia, J.Saborido, R.Forty
Minutes: M.Cattaneo
Test beam software (T.Ruf)
Pattern Recognition for the RICH and inclusion in event model (D.Liko)
Status of SICB C++ classes (P.Binko)
Thomas gave an overview of the use of ROOT in the vertex detector test beam (see slides.ppt, .ps). The choice to use ROOT and C++ was made only 3 weeks before the start of data-taking, it is impressive how fast progress was made. Thomas stressed the importance of getting help from OO experts (in his case Fons Raedemakers) in designing the class structure; he sees this project as a way of learning C++ and OO programming. A new iteration of the software is under study, beginning with a URD which P.Maley is assembling.
The structure of the test beam data could be the basis for the event and geometry models of the LHCb simulation and reconstruction. Pavel will study how this could be implemented in his event model.
It was felt that, at the present time, ROOT is the only tool that allows interactive data analysis by physicists using C++
Dietrich gave an update of the talk given by Niko on 25th June. There has been discussion with Pavel and with the RICH group on the details of the event model. This will continue at meetings of the RICH software group, to which Pavel will be invited.
Thomas asked whether OO is helping to speed up the pattern recognition. It is to premature to answer this, but this project will allow a direct comparison of the Fortran and C++ implementations of the same standalone algorithm.
Markus objected that the "Detector" class which can "Simulate" and "Reconstruct" opens up the risk of producing a new monolithic set of software (c.f. SICB) which can do everything but takes forever to link and to load. Dietrich explained that the choice was dictated by the need to implement a standalone program that does both, but agreed that this choice can be reviewed at a later stage.
Pavel briefly summarised the status of his and Iain Last's work. The generation of "XYZ_objects" (i.e. classes with a one to one correspondence with the SICB Fortran banks) from the SICB .ddf files is complete. This part of the project is complete in the sense that, once generated, the classes do not change if the .ddf does not change. There are in addition 8 classes that provide the glue to SICB. This can all be linked together and allows access to the C++ equivalent of the SICB Fortran banks, including the references. The program can be steered in the same way as SICB, so allows generation of new data as well as access to existing SICB output data.
Pavel is now working on the the OO event model layer which will hide the SICB classes from OO analysis application (i.e. the LHCb transient event model). He will present the status next week. He is also working with Florence to put Iain's code in a public area under CVS.
We will try to have a brief meeting every week. Next week Pavel will present progress on the event model and its interface to the SICB classes.