I am in favor of enduring the short term hardships and fixing the problem and then repopulating the database. I think that we are early enough in the production use of the DB that it is wise to do that. Kris Quoting Flemming Videbaek <videbaek@sgs1.hirg.bnl.gov>: > > Database problems > > As you may have seen Steve Sanders pointed to a problem with the > calibration > DB in so far data was not correctly transferred to his linux ppc. > Christian and I have since looked at it further, and must admit that we > made > a mistake / design flaw back about 1.5 year ago. > Unfortunately only now has it shown up. > There seem to be several possibilities on how to proceed. > > a) Ignore the problem i.e. DB access cannot be done to > calibrations > from linux ppc, solaris etc > b) Modification to code, that will disallow certain operation to > be > invalid on e.g. > linux-ppc and solaris machines, but maintain access to db constants via > the > BrCalibration class only > c) Redesign the table Revision that hold the calibration data, as > well > as the internal > storage. This implies the Cal DB has to be re-populated. It may be > possible > to do this > by extracting the revisions, deleting and then re-creating, the easier > would > be to ask the limited cal done so far to be re-inserted e.g. DO thought > it > was ok for all the TO\F > stuff. > > > The solution a) is clear un-acceptable; though most work is done on > the > intel boxes, a > not insignificant piece is done on platforms with opposite endian data > storage. > > The solution b) does not really appeal to neither Christian nor me, and > is > therefore un- > desirable. The last solution is not convenient,. I have looked in > detail at > it and it is > workable, even though it has some slight messy pieces (same information > both > in > Parameter and Revision, which could lead to problems if people were to > exploit low- > level access). The last solution imposes some hard-ship on the ongoing > analysis, but > could probably be put in during a scheduled stopping and refilling of > the > DB. > > It is obvious something has to be done, and the sooner the better. > > In particular I will like comments back from people who have made > actual > valid > calibrations (don't count pedestals- too easy) for tof, cherenkovs, > tiles, > si . I/we will be > happy to provide you copies of the tougths made so far. > > Flemming > > > ------------------------------------------------------ > Flemming Videbaek > Physics Department > Brookhaven National Laboratory > > tlf: 631-344-4106 > fax 631-344-1334 > e-mail: videbaek@bnl.gov >
This archive was generated by hypermail 2b30 : Fri Nov 09 2001 - 12:32:03 EST