ROOT 6 planning
Present: Marco, Roel ,David M., Marcin, Wim, Benedikt, Muhammad, Peter, John H.
o RooFit: Updated to most recent version in 5, 6 patches and 6.
o Axel+Danilo: fix the bug for the usage of Roo* classes at the prompt for CMSSW.
Peter
o Trying to make a new release, not much done. The release will be done this week.
o Performance measurements and, if needed, will be continued afterwards
o No blocker so far.
o Root5 and 6 will be in parallel on a validation cluster.
Marcin
o First reconstruction job ran successfully (congratulations for the high quality work, Marcin and Atlas colleagues!) No measurement of memory was done.
o I/O error: if a reproducer cannot be isolated, a joint debug session is to be foreseen with Axel, Scott and possibly Philippe
o No other outstanding issue.
Marco
o Try to pick up the new stack against the nightlies. Not much to report
o It was heard that it seems that ROOT6 is not an issue at HLT for LHCb. The new nodes have 1GB/core and LHCb want to overcommit by 20% the farm. It is tight and more tests are being performed at the moment: forking is not trivial and careful measurements is needed.
o RSS is for last year 1.2 GB at HLT and mem less than 2GB
Bill
o 7 tests failing over >250.
o 5 unit tests in the conditions area: Giacomo Govi is taking care of those.
o The other two tests: macro execution. One is obsolete and one is fixed by the latest version of ROOT.
o It seems that in one week one could think of having, with some luck, all of the tests fixed next week.
o Memory: +10/+100 MB RSS. One can improve this with a richer PCH or, with some knowledge of the problem, one could think of removing some parsing even on the CMS side.
o On holiday until Tue next week.
Mohammad
o Problem of rootmap generation solved: rootmaps now there for 5 and 6.
o ROOT 6.02.01 being used: speed and memory comparable. Maybe a report will be presented at some point.
o Bug with Maveriks and Geant3, maybe unrelated to ROOT.