CERN QMUL
L1Calo Software Meeting at RAL on 13 June 2001
Present: Bruce, Eric, Gilles, Murrough, Norman, Steve.Discussion of the project as a whole
We devoted most of this meeting to discussing the state of the software project as a whole.We started with a short discussion about our general approach. We have tended to talk only about details at our meetings and havent paid enough attention to ensuring that we are all still going in the (same agreed) right direction after the various changes of ambition and timescale of the hardware, and most importantly have not checked that all the required work is done.
Another concern has been our documents. We have historically been poor at either producing or especially finishing documents. The few that exist are generally drafts for discussion and almost all have single authors listed, though they may have emerged from wider discussions. As part of a better software process, we should pay more attention to finishing and reviewing documents, which can then be the agreed view of the whole group and recognised as such in the author list. We need to distinguish carefully between agreed and draft documents when talking to external colleagues.
Software Context Diagram
A brainstorming meeting between Bruce, Murrough and Norman a few days earlier had produced a software context diagram and a preliminary decomposition into software "themes" which Murrough had recorded in some slides.In the discussion, both the diagram and the list of "themes" were modified slightly. A new version of the talk will be presented at Mainz. The software themes roughly map onto the main use cases of our software. Later we talked about the division of the software into packages.
We agreed that the results of our discussion, with any further refinements arising from the Mainz meeting, should go into a revamped version of our of requirements document. This was left unfinished in 1999 (with a space for a context diagram!).
L1Calo Software Packages
In the last hour of the meeting we made a first attempt at drawing a set of packages and their dependencies in a UML style format.As with the context diagram there were many areas of detail provoking discussion. The snapshot of this diagram is still partly Murroughs personal view as there were still several unresolved issues including:
- Hardware Access Library (HAL): maybe this needs to be split into layers, eg with higher level packages intermediate between HAL and its various dependent packages such as Run Control etc.
- Calibration and System Testing: in one view these are not true packages but activities which are implement by extra classes falling naturally into various other packages. Eg calibrations may need special test vectors, loading by special run control code with the calibrations derived by special event monitoring programs.
Once a set of packages and their dependency relations are agreed we need to define the scope of each package in more detail and the interfaces between them more clearly.
In all the above, we shouldnt lose sight of the work already carried out on what we have up to now considered the general packages in the software (some of which are much the same as in our new analysis).
Recent Work
Discussion on Steves work on testing and simulation was put off till the Mainz meeting due to lack of time.Documents
We had already discussed documents in general at the start of the meeting. Here we recapped: existing drafts should be reviewed properly at one of our meetings (probably one document per meeting is enough).The first priority should be reworking the old requirements document with our new thinking and diagrams mentioned above.
In addition to the fairly recent series of software notes we now also have the first of a series of technical information notes.
Norman is working on a future such note, describing all the digital data formats passing around the system. It was also suggested we need an overview of the whole system - since we have recently discovered instances of different interpretation of separate specifications.
Talks at Mainz
Murrough showed a draft agenda for the Mainz software meeting. As ever there is some overlap with the agenda for the rest of the meeting which should be resolved with Eric.AOB
None.Actions
New actions from this meeting:- All: consider our new diagrams
- Various: work on documents...
Next meeting
The next meeting is on 28 June at Mainz.Post meeting developments
After a phone conference a couple of days after the meeting, Murrough drew up some words elaborating on the suggested scope of the packages in his view. You can read this here.Last updated on 15-Jun-2001 by Murrough Landon