Towards a Timeline for Reprocessing
To create the following timeline, I consider
technical and organizational improvements wrt. the previous reprocessing.
Basic strategy
Default mode of operation is the one we used last time.
I.e.: Reprocessing from DST, TMB merging at d0mino.
Any improvement has to be tested before a decision to include it
shall be taken.
Physics
The definition of the programm version, the type of reprocessing
(from DST vs. from RAW) and the order of reprocessing
should be driven by physics needs and timescales. But as they
also involve technical issues they should be made in a timely manner.
The decisions should remain valid in case of moderate delays or
speedups to avoid extra work on the technical side.
(In autuum last minute changes from the physics side )
integrated to a delay of 2-3 weeks.
Such decisions should be coordinated with the technical team.
It would be good to discuss requests
from the physics side (presented by a representative of the phyics teams)
inside the reprocessing meeting where the technical implication can be
discussed and in case of conflicts an agreement can be searched.
Technical
For each new method involved we should have one site to test the new
feature. This site should try to avoid using site specific features
of their environment to assure compatibility with as many sites as possible.
When working that new method is distributed to the other
participating sites and they should test it as well.
Only when both steps succeeded (at sufficiently many/all sites) that
method can be admitted for inclusion in our procedure.
Daniel Wicke, 18-Feb-2003. Last change 18-Feb-2004.
wicke@fnal.gov
