L1Calo Software Video Conference on 6 September 2001

Actions from last meeting

Almost all actions from the last meeting were completed.

Actions from this meeting (and one carried over) are summarised at the end.

Reports of recent work: Bruce

Bruce has installed RedHat 7.0 on one of Concurrent CPUs destined for Birmingham. CERN is expected to certify 7.1 or 7.2, but all provide gcc 2.96 so we should not have problems with the next Online software release (scheduled for RedHat 7.1 in October). He still wants to investigate the better bus error handling in the new Concurrent VME driver.

As part of his work in the CPUs and continuing ROD tests, he has made a few more updates in the HDMC repository. He has also stored his ROD test program in the RAL CVS repository.

Reports of recent work: Steve

Changes to HDMC: Oliver

It is important that we have this nice interactive tool for debugging, but HDMC can not easily provide batch type behaviour required for the DAQ environment.

Olivers view met with a favourable response and he was encouraged to make more detailed suggestions for the proposed changes. Some detailed requests were aired, mainly covered in part of Bruces talk.

Documentation and comments in the code are still on the to do list.

Requirements and Software Organisation documents

On the requirements document, Oliver suggested that we should expand the brief summary of "software areas" into a more formal and lengthy discussion of use cases. He also suggested that we needed a software view of the hardware, perhaps as an appendix. Here it was felt that more detailed "user guide" to the hardware might be better as separate document(s).

On the software organisation, Oliver was strongly in favour of us agreeing a mutually acceptable software process. So far we have waited for proposals from the TDAQ "Connect Forum". This likely to propose that for a typical package, we should write and review (a) requirements, (b) design, (c) test plan and (d) user guide. Oliver offered to make some suggestions for the path he would like us to follow.

Discussion of software organisation and packages

NB after we assigned some names to packages at our last meeting, Murrough updated the packages page on the software website accordingly.

Discussion of module services package

Bruce included in his talk some comments on the two practical approaches he has used in implementing code describing modules for the ROD tests. The first worked "bottom up" adding syntax to HDMC parts files to build descriptions of modules. His more recent work was more "top down", instantiating parts outside the HDMC part manager. He suggests merging the more recent work into the first approach.

Murrough also prepared a number of slides setting out his views on the module services package. These are mainly focussed on configuring the system, so there should be a close connection between module services and the configuration database. Our modules classes should be composed of some major subcomponent classes which also handle database objects. Detailed implementation should be done using lower level HDMC parts.

Open questions include how much detail of a modules internal structure should be visible via its methods, details of how to describe collections of test vectors, etc.

We didnt have enough time for discussion, but it was agreed that a coherent approach for all our modules was important and that we need further discussion to achieve a concensus.

Next meeting(s)

Actions from this meeting