JCOP Framework Working Group
13 November 2018

NEWS:

• WinCC OA patches: 3.15P016 expected this week; 3.16 FeaturePack1 expected end of November; no further details w.r.t.  their content yet. 3.16FeaturePack2 around April, with patches every 6 weeks in a meantime.
• Framework plans: work on the next version, 8.3.0 are  well advanced (testing phase); beta expected next week; release at the end of November; (no comments to this planning signaled)

Report on the status of NextGeneration Archiver: 

• presentation slides are in indico
• Clara asked what if one wants to switch the technology and still keep the continuity of archive. Use case is considered and will be provided (even though may be pretty primitive in the first version); will be improved with some "intelligent" algorithms; option to specify the backend to be queried will be available
• questions about performance comparison between RDB+directQueries and NGA+directQueries. No performance difference expected, as schema does not change. 
• In general, interest on comparison testing for other technologies to be evaluated
• Question about partitioning in Influx: yes, it is present in form of sharding plus in addition the compressions that may be calculated automatically by the database
• Efficient testing and deployment would only be possible with involvement of experiments, as it was in the past.  

Report on the status of NextGen Alarm Screen: 

• presentation slides are in indico
• demonstration of the current development snapshot of the Alarm Screen for current alarms.
• Discussion on having "filter sets" and use of tabs - may be confused. danger of not seeing the whole filter; maybe a list of includes and includes? 
  • there are two cases: preparing the filter and then displaying them. Once it is prepared, there is no interest in modifying them yet still (s)he should know what are  the filters that are effective
• Usefulness of exclude-filters acknowledged
• How to extend the filter widget? this is EWO, which is relatively easy extendable, yet requires C++ at least at the moment; we consider having an auto "flow layout", that would allow to specify additional widgets without C++ code
• how would the performance of "live filtering" compare with Javascript? Cannot really provide answer, because no comparison test were done. The past javascript/webview prototypes were made with a hybrid of CTRL, which might have impacted their performance
• change in approach to having alarmscreen in historical mode already in the first preview is important for LHCb, which usually have the "live" alarm screen empty, yet would be keen on having powerful filtering on historical alarms.
• Memory consumption? depends on use cases, names of dp names, etc; it is being checked and considered
  • hybrid approach seems adequate, e.g. not to connect to all alarms with too low priority that would never be displayed
• copy of projects? ATLAS could deliver DPL files
• usability of filters should be specified/discussed before the final implementations; ATLAS would use lots of them

AOB: