CERN QMUL
L1Calo Software Meeting at Mainz on 28 June 2001
The agenda was:
| Software overview and discussion | Murrough (30') |
| Readout | Bruce (20') |
| Testing, simulation and monitoring | Steve (30') |
| Proposed changes to HDMC | Murrough (15') |
| HDMC experience and plans for CPM testing | Gilles (10') |
| Documents, software process and quality assurance | Murrough (10') |
| Recent and imminent meetings | Murrough (5') |
Software overview
Murrough presented a summary of the software project. Recent discussions in the UK had focussed around revisiting our old unfinished requirements document. We aim to rewrite this with a new software context diagram and more formal, UML style, decomposition of the project into well described packages. The draft diagrams provoked some discussion at the meeting which was continued at a smaller gathering afterwards. Further contributions by email or phone were encouraged.A new, and still daunting, list of outstanding software tasks was also shown. The effort to implement them is still limited: Bill has unfortunately had to retire due to ill health, but Oliver is about to start work.
Readout
Bruce gave a talk on our possible readout strategies. We have a variety of requirements, from test bench to test beam. At least with test data, the readout may only need to be a pointer to which event in a predefined set was in error. However it was felt that in other cases we do need to read out the actual event fragments. Although a hardware PC based ROS doesnt meet all our requirements, now that (software) components of the ROS are becoming available, it makes sense to use them - where appropriate.There are still outstanding questions related to the ROS which we should discuss with the relevant people at the imminent Trigger/DAQ week at CERN.
Testing, Simulation and Monitoring
Steve discussed the development of a simulation framework. This is loosely inspired by VHDLs concepts of Entities and Ports. The functionality of a complex module, converting inputs to outputs is broken down into a hierarchy of composite or individual processes. Given the framework, the user only has to write appropriate processes for a specific module. The aim is to initialise all the connections from the DAQ configuration database and also to integrate the various standalone test vector generators. Monitoring is still a lower priority item...A worry was expressed that this work was a luxury and might also overlap with offline simulation work. However it was felt that this more detailed, hardware oriented simulation was required for testing the modules. The talk at the software meeting was concentrating on implementation details, eg class diagrams. More detail on the rationale for this work would be given later in the main meeting.
Proposed changes to HDMC
Murrough summarised a document he had written, after discussions with Bruce and others, on the changes needed in HDMC for it to provide the "hardware access library" services required by our other packages. Although this was first drafted a few months ago, little has yet been implemented and some changes are becoming urgent.HDMC experience and plans for CPM testing
Gilles reported briefly on plans to use HDMC to do initial testing of the CPM at Birmingahm. However he has not yet been able to install the latest version due to problems installing the required Qwt library. This should be sorted out soon.Documents, software process and quality assurance
Murrough showed a list of several draft documents which ought to be finished, and a few which ought to be written! We need to establish a simple procedure for reviews and an agreed author list.As part of the TDAQ moves to a more formal software process, we will need follow these developments. It will probably be good for us - those involved in recent discussions around the requirements document, software context diagram and package diagram have certainly found them useful.
Recent and imminent meetings
Murrough then reported on the recent level1 software video conference between CERN, Israel, Japan and the UK. This gave an insight into the status and plans at each site. It also touched on the issue of software process and QA. It was generally agreed we should encourage as much commonality across the whole of level1 as reasonably possible.AOB
Later in the joint meeting we discussed the unsatisfactory layout of the joint meetings. It would be more sensible to have software overviews first in the main meeting with more detailed discussions of software issues coming afterwards for those who were interested - though the view was expressed that as software people sat through lots of hardware related talks it would be good if more hardware people took at least a general interest in the software side of the project.We also touched on our intermediate meetings. We should revive video conferencing between RAL and Heidelberg - at least for well focussed topics.
Another discussion topic was that of software setups in home labs. Will we really be able to maintain a full, up to date, DAQ library environment at home labs? For some tests this will probably be necessary, but we should also try to make sure that some simpler, HDMC style diagnostic setups can be run without too much additional software infrastructure.
Next meeting
The next meeting will be at RAL around 7-8 November.Meanwhile, Oliver felt that a UK-HD video conference in about one month would be appropriate.
Last updated on 20-Oct-2001 by Murrough Landon