JCOP FWWG meeting 25 January 2022
INDICO: https://indico.cern.ch/event/1116932/
Participants: Christophe Pires, Luis Miguens Fernandez, Piotr Golonka, Manuel Vazquez Muniz, Stefan Schlenker
Clara Gaspar, Piero Giorgio Verdini, Juan Manuel Pablo Rodriguez, Lukasz Goralczyk, Mariusz Suder, Mateusz Lechman, Wassef Karimeh, Laura Sargsyan, Ben Farnham, Cristina Vazquez, Frank Glege, Peter Bond, LuisGC, Riku-Pekka Silvola, Fernando Varela, Piotr Nikiel, Valeri Falaleev, Paris Moschovakos, Ivan Anatasov, Michael Ludwig
WinCC OA news
- Riku presented the motivation for the phase-out of the CMF distribution for WinCC OA (see slides)
- WinCC OA 3.16 patch P030 and WinCC OA 3.18 patch P006 released - CERN packages now available.
- to access the packages for the 3.18 version, please get in touch with the support team
- Piotr presented the report from the "WinCC OA followup meeting", with the roadmap for the upcoming WinCC OA releases and features (see slides)
- question/comment about the security of the New WebUI: this seems to be more and more serious nowadays.
- Cannot tell at this moment - an initial exchange with ETM on the WebUI topic planned on 3rd of March. Piotr invited whoever might be interested to join (Pierro signalled his interest)
Automated tool to scan for necessary #uses statements in panels and scripts
- Presented by Luis (see the slides)
- Comment from ATLAS: it is quite serious limitation to not recognize string-representations in e.g. callbacks. In particular, it would be of great help if this is supported for dpConnects (majority of cases)
- acknowledged that this could only work if the callback function name is a literal string as the parameter. Lookup of the callback in other libraries would still be very useful.
- evalScript may be much more problematic...
- CMS has a similar kind of tool - maybe an exchange of experience
- It would be nice if the tool would also check the global vars. It is not there yet, but possible to implement as functionality is present in the underlying tools
- It was clarified that the tool will report the lib where the function actually is, not its "parent library". Hence important that the action is reviewed and, for instance, "private" libraries (such as those in the FSM component) would not be used.
- Clarified: the tool is smart enough to track the whole chains of #uses dependencies, hence it reacts to the LoadCtrlLib statements in the project config file. Therefore the projects that load the include_legacy.ctl, all the libraries are loaded - this needs to be disabled to work with the tool and effectively detect missing #uses...
- The tool supports libs but also scripts and panels.
- In the current version the tool was supposed to be used "iteratively" with the code developer inserting the necessary #uses statement following suggestions. This is impracticable for huge code basses of the experiments, of complete sets of repositories
- therefore it is necessary to have an option that would also automatically "fix" the code - at least apply the most obvious fixes. It might need to be a configurable option on how this works
- the feature was in planning, but for this first version we wanted Experiments' feedback before proceeding
-
fwOpcUaTools - update on status
- Mariusz presented the features developed based on the recent feedback (see slides)
- CMS is experiencing problem with the WinCC OPC-UA client - some devices not getting the invalid bit when powered on, some others do. REPORTED after the meeting as ENS-30960
- it is essential that the problem is understood as it may lead to serious effect; it could be related to the use of event filters
- ATLAS suggest the use of the data change filter Value+Status+Timestamp (instead of just Value+Status) in all subscriptions, that they use as a proper default for other fw devices.
- For legacy devices, that does not support it correctly, another subscription could be defined/used...
- needs to be followed up
- New tool suggested to check the validity for output addresses
- the problem is that unlike the input addresses that get an invalid bit on an attempt to readout, one realizes about a wrong output address is only at the moment a value is set to the DPE (log file, and a dedicated DPE probably). This seems to be one of the most important problems to look at
- Mentioned that e.g. in CAEN there are settings and their readback, and this could be used to detect a failed set. Yet OPC-UA provides much better ways to handle this
- a tool should be available to validate the output-addresses against those available in the server (browse the OPC-UA namespace, or query from a driver)
- Is there a tool allowing to reconfigure a part of, say, CAEN devices and move them to a separate driver?
- yes, this was released in the very first version of fwOpcUaTools. NA62 to follow up by opening a ticket if this is not working as expected
Proposal for the new Framework exceptions
Due to running out of time the presentation has been postponed to the next JCOP FWWG Meeting, 8 February
https://indico.cern.ch/event/1121310/
There are minutes attached to this event.
Show them.