Software phone meeting on 28 September 2006

Present: Bruce, Florian, Frederik, Markus, Murrough, Paolo, Steve.

Status reports

Heidelberg

Florian has been at CERN this week and has tested his GNAM changes as far as possible without TileCal signals. This now includes the DAC, pedestal and DAQ pointer runs with multiple PPMs in two crates.

He again raised the problems reported last time:

• his GNAM library crashes on the second configure step. This seems related to ROOT dictionary classes. Paolo said this was an old problem and will try to find out how (or if) it was resolved in the past.
• it now seems clear that histograms are not published after the endOfRun entry point. This makes is difficult to fill histograms after all events have been processed. Steve said that in future releases the behaviour at the stop transition may be rethought.

Florians next steps are to test with tdaq-01-06-01, complete tests with TileCal signals, try to optimise the speed and find a solution to where to store results (until database work is done).

One immediately useful thing would be to be able to record the L1Calo run type used for a run - at the moment its cannot be found in the ROS data. A four bit field was reserved in our detector event type, but thats probably not enough.

It would be useful if Murrough could get the conditions database interface working so we could store our IS variables keyed to run numbers in the near future. This does have implications for test rigs needing to have some DB infrastructure (though perhaps it could just be the SQLite implementation of the CORAL DB access). People would also need to be careful about using strictly increasing run numbers and not deleting the RunParams IS server backup file (which resets the run number to 1000).

Mainz

Markus now has an lxplus account so can be added to the CVS commit list. He also is happy to move to tdaq-01-06-01 soon, eg when Gilles does, and so presumably before production modules start arriving (next month).

QMUL

Paolo has committed updates to the gnamL1CaloHisto package which fixes the problems in being able to run more than one shared library with HistogrammerBase subclasess at once. He has started looking at adding the extra entry points requested - but needs to understand which libraries to use for the database.

CERN

Bruce has made the module services updates for tdaq-01-06-01. Apart from the new connect/disconnect methods, these include an isDataAvailable method for use by RCD and also methods to return "full statistics" from modules. There are some thin implementations of full statistics classes for each module type in a new statsL1Calo package.

Murrough, with the help of Sergei, has fixed the IS update problem seen in tdaq-01-06-01. Another problem reported by Florian whereby there was no VME readout in RCD is also now understood, but still needs testing.

It was proposed to tag all our packages on friday afternoon as the last versions guaranteed to work with tdaq-01-04-01. Then we will start to move to tdaq-01-06-01 at point 1 and the external test rigs.

Next meeting

As a reminder, the GNAM "experts week" is set for 23 October. The previous week, starting 16 October will be a DCS week where Gilles and Chris will come to set up our DCS at point 1.

Next meeting

The next phone meeting will be on Thursday 12 October at 11:00 CET (10:00 GMT).