Software work - summary report

From: videbaek (videbaek@sgs1.hirg.bnl.gov)
Date: Sun Sep 03 2000 - 11:35:44 EDT

  • Next message: I. G. Bearden: "Re: Software work - summary report"

    To: brahms-dev-l
    
    From: FV
    
    Following Kris's stay at BNL last week he, Konstantin and I had a summarizing discussion on items of general software interest. Most are related to general issues and not the specific analysis efforts that are in progress. For some of the overall issue I refer to my talk at last Monday's local meeting , which will be posted shortly on the web (Tomorrow Monday).
    
      a.. The Brat and related applications should be updated regularly (daily). Konstantin will implement this. The afs /opt/brahms will be updated for a "current" version. The pro and new will be handled 'manually' and will correspond to tagged versions. If a daily update fails e-mail with offending errors will be send to the dev-l list. Both Linux and Solaris version will be handled this way.
      b.. Bugs should be reported to the dev-l list. Will not use the bug-report suite.
      c.. There are some incompatibilities with the new Root version 2.25. The default setups should be with root 2.24
      d.. Though not of highest priority it is important that follow-up takes place in the area of data monitoring.
      e.. Histogram selections should be reviewed by detector people and people who had been on shift.
      f.. Methods should be implemented such that the Ntuple , Supermon etc. will be reset and started, and generates automatic storage at the end (possible web-based)
      a.. Of high priority is the fact the event sizes are very large. Lots of this is caused by MTP1 data (50-60kB), but other contributors are a) keeping raw unpacked data b) on suppression of empty TDC channels etc.
      b.. Data objects should be finalized - this in particular to global detector rdo data, tracking data and PID data. This should be done through a consensus process.
      c.. We had a discussion on eventnodes vs events. Analysis modules should always deal with nodes while i/o deals with events. This could possibly be changed and the overall event header information made available by the EventManager.
      d.. Kris is making good progress in the re-organization of the brat directories. The main effort is making new Makefiles and checking these out. The new structures are updated via scripts, thus the switchover once a decision day is made can be made smootly.
    
    
    ------------------------------
    Flemming Videbaek
    Physics Department
    Brookhaven National Laboratory
    tlf: 631-344-4106
    fax: 631-344-1334 
    videbaek@bnl.gov
     
    



    This archive was generated by hypermail 2b29 : Sun Sep 03 2000 - 11:16:48 EDT