[ Home | Contents | Search | Post | Reply | Next | Previous | Up ]
From: Andrei Tsaregorodtsev
Date: 4/08/97
Time: 5:01:23 PM
Remote Name: 137.138.117.88
Remote User: Andrei Tsaregorodtsev
Software group meeting, 8 April
J.Harvey, F.Ranjard, P.Schleper, G.Wilkinson, T.Ruf, H.Dijkstra, A.Tsaregorodtsev, J.Saborido, F.Harris.
Since last meeting some new codes were incorporated into the development version of the simulation program. In particular, these are new definition of the downstream RICH geometry and Silicon barrel detector digitization package. Peter has reviewed the status of the trigger code development. He is maintaining a WEB page with links to all the existing documentation, DDF and CDF files, CMZ library. Thomas will put a link to this page in the LHCb Trigger group WEB page. Muon Level0 trigger code has been delivered. The documentation on the trigger algorithm is to be written (Andrei). Electron Level0 trigger code has been given to Peter. Also documentation is lacking (Ivan). Frank asked about a use of preshower in the trigger algorithm implemented. The answer was that it is implemented in a flexible way so that the preshower can either be switched on or off. Guy has prepared the Hadron Level0 trigger code and is testing it now on large statistics. The code is ready to be given to Peter. The same is with the pile-up veto code. Peter is working on the tracking trigger. It is close to be incorporated soon. Thomas reported on his work on incorporation of his Vertex trigger code. The bank definitions are done. This includes banks for track candidates (2D and 3d tracks) and vertices. The banks are described in the DDF files and filled in. The trigger parameters are to be put into CDF file (together with Andrei) shortly. Hans will work on the Vertex trigger summary bank. David Steele has started work on the new Vertex detector description in GEANT as well as the new digitization procedure. He will prepare new digitization and cluster banks. Hans pointed put the necessity of the trigger codes to be able to work starting from the "DST" tapes. Florence presented her ideas about the LHCb software configuration. Now we have 15 libraries including one library per subdetector which are dependant on each other via using COMMON blocks defined in other libraries. Most of the libraries contain both SIMULATION and RECONSTRUCTION codes as well as UTILITIES code. All the libraries are getting the new version with a new release of the whole software. The same is with data files. It will be nice to define now the configuration which will be used on a long term basis. For that it should be as modular as possible, it should be supported by the release management procedures, it should be independant of the tools like code manager, compilers, languages, etc. Florence proposed the following. Use the CVS code manager. It is an emerging standard in the HEP community and is certainly adequate for our needs. Use 4 main libraries: SIMULATION, RECONSTRUCTION, UTILITIES, ANALYSIS. All these libraries except UTILITIES will contain codes related to a separate corresponding programs. The libraries will be managed by librarians (1 to 4) who will take care that the constituent codes are conforming the conventions, documented and tested. The libraries will consist of packages containing codes, include files, documentation. It is important that the include files will be internal for the packages and not allowed to be used in other packages. A set of include files common to all the libraries will be managed by a database librarian. The proposal by Florence triggered a hot discussion. In particular, the organization of the software in libraries in terms of independant programs rather than "subdetector" approach was argued. There was a strong argument that the tasks of the code writing and managing libraries should be splitted and taken over by independant persons. In the end, a consistent and efficient program (i.e RECONSTRUCTION or ANALYSIS) which is used by the end user should be delivered. This makes necessary maintaing the programs as a whole although the code written by the same person will go to different libraries. It is clear that these issues will be discussed seriously in coming days and months. For the time being it was agreed that Florence and Andrei will prepare a CVS repository with the proposed structure and will rearrange code to fit this structure as a pilot project. The transperencies of the presentations by Thomas and Florence are put to the SW group folder in the LHCb secretariat. The next meeting will be on 18th of April, Friday, at 9:00. The place and the detailed agenda will be available shortly.
Andrei Tsaregorodtsev