OWG Meeting 1-Sep-2010

Agenda

• Announcements
• Review of minutes from 18-Aug-2010 meeting
• Collaboration Meeting agenda
• Plans for EVIO/JEvio version 4 - Carl
• Hall D raw data format - all
• Event building strategy - all
• Coil test update - Elliott, Yi
• Other electronics, trigger and DAQ status reports - Fernando, Chris, Ben, Ed, Alex, Dave, William, Bryan

Time/Location

1:30 PM Wed 1-Sep-2010 CC F326

Next Meeting

1:30 PM Wed 15-Sep-2010 CC F326

Minutes

Present: Elliott W, Carl T, David A, Simon T, Bryan M, others (I neglected to write the names down!)

EVIO V4

Carl described upcoming improvements to the Java and C++ EVIO packages.

• Original format designed for file I/O.
• Simplifying format to unify file, network, memory footprints.
• Adding tag/num dictionary and API that uses it.
• Resolving byte and short array padding ambiguity.
• Java version testing late Sep, then C/C++.
• V4 will be able to read V1-3 files.

Raw data format and event building

Dave A presented an overview of event building in Coda 3. A long discussion ensued on this and related topics.

• Events are read out of front ends in blocks.
• Raw data format coming out of TDC and FADC is well known.
• Blocks contain mixtures of data from multiple events.
• Basic event building collects all blocks into one EVIO record, but doesn't disentangle data from the different events.
• Basic event building is very robust and reliable since user data not looked at, just headers and lengths.
• Hall D needs to have the event data disentangled online for monitoring and filtering.
• From DAQ standpoint, disentangling in general is problematic due to custom data added by users.
• In most case we will not add custom data, and will just have JLab TDC's and FADC's.
• DAQ group will provide example disentangler as a base we can build on.
• Not clear where to do disentangling, in additional event building stage or in a farm. We will eventually test both strategies.
• Disentangling may require detailed knowledge of all crate layouts.
• It is not clear how to handle non-standard modules or custom user data.

Trigger test

• Linux roc available, but Yi needs to use current setup for a while.
• Will switch to Linux when Yi is done.
• May use setup for Richard's test later this fall.