Level 1 Calo Software Meeting at RAL on Wednesday 9 December 1998.
------------------------------------------------------------------

Present: Eric, Murrough, Norman, Reg (pm only), Scott, Tara.


Before the meeting we started by giving Scott a demonstration of our
existing diagnostic and DAQ software. This was also helpful to some
others who werent familiar with all the features of the existing code!


The agenda of the meeting was:

(0) Feedback from Heidelberg meeting?
(1) Feedback from discussions at CERN.......Murrough
(2) Overview of DAQ -1 prototype............Murrough
(3) Report on CVS and SRT...................Tara
(4) Requirements document
    (a) Monitoring..........................Eric
    (a) Test & Diagnostic Software..........Murrough
    (a) Controls............................Norman
    (a) Calibration.........................Tara
(5) Future workplan and timescales


Feedback from Heidelberg Meeting
--------------------------------

Murrough gave a talk at the Heidelberg joint meeting, mentioning that the
software group had been revived (initially) within the UK. Our immediate
plans for a requirements document and to establish some agreed software
process, possibly using some or all of the tools favoured by some of the
rest of ATLAS accepted. Meanwhile it is clear that Heidelberg are currently
pursuing a separate software development for their test systems. We should
try to converge at some point. As part of this we need to reach a concensus
on what a module is, what functions it can have as seen by software. We have
to be open to the different views that exist, but concensus may involve some
compromises.


Feedback from CERN discussions
------------------------------

Murrough spent two days at CERN talking to members of the DAQ group. In
practice most time was spent with Bob Jones concentrating on the Back End
software; with a much shorter discussion with Giuseppe Mornacchi on the
dataflow side, especially ideas for extending dataflow services into the
ROD crates. Some very sketchy
notes of the meetings are available.

The discussions with Bob ranged over hardware platforms, the evolution of
test software, use of run control (with a demonstration) and databases.
Giuseppe explained the two models of how dataflow software might be extended
into the ROD crates. Its not clear which might be best for us.

The DAQ group (for which read Livio in this context) requires us to provide
a workplan setting out how we will initially use the DAQ software before they
give us access to it. Murrough prepared a draft
workplan document for this purpose.
This outlines our timescales and suggests that Murrough will initially look
at databases and Norman will look at the Message and Information Services
(eg in the context of our existing DAQ perhaps?).

After the meeting (Thursday 10 December 1998) this bore fruit: both Norman
and Murrough were granted access to some Back End DAQ components.

The DAQ group are not keen on the idea of us trying to port the software to
our older 68K based Motorola CPUs. In principle if we upgrade to LynxOS 2.5.1
and the associated GNU C++ compiler at least all the Back End software should
compile. However all the software is built within the SRT environment at CERN
and it might not be simple to just extract the code, copy it to the UK and
build it here. Also the dataflow code might require some features of the CES
PPC based CPUs. Nonetheless it was made clear that we might still want to try
doing this in the short term for our ROD prototype tests.


DAQ -1 Prototype Software
-------------------------

Murrough gave a presentation of the DAQ -1 software. This is divided into
two main areas: the dataflow side dealing with the collection of data and
monitoring; and the Back End support services. Murroughs investigations
have concentrated on the Back End which he presented in greater detail.
In lieu of writing it up in the minutes, take a look at the DAQ groups
copious documentation.


Software Tools
--------------

Tara presented some slides on CVS and SRT. CVS is used in Zeus and they
have developed a model for their software management. They use CVS for
managing software versions; they define managers who are allowed to update
the various major projects; lastly they have developed scripts and makefiles
for controlling the build process. The latter aspect is the most difficult
and is the area addressed by SRT.

Key concepts in CVS: all code is kept in the "Repository" which is never
accessed directly. Developers who want to work on some code "check out" the
module(s) they are interested in; make modifications and test them; when they
are happy with the changes they "commit" them back to the Repository. Zeus
maintain three separate CVS trees: the main repository, one for manageers and
the released version for general users. All of them are AFS mounted for access
across the collaboration.

Tara gave an overview of SRT: it uses GNU autoconf tool to create the makefiles
which are used to build the software on different platforms. It seems as if
most of the work goes into creating the templates used by autoconf. After that
the rest should be "easy". However it is difficult to judge just from reading
the manual: we need to talk to someone who has practical experience.
[After the meeting, Bob Jones suggested we start by looking at the DAQ groups
SRT page].

We had hoped to have a presentation about the Software through Pictures (StP)
CASE tool by Steve Fisher. However he was not available so hopefully this will
happen at our next meeting.


Requirements Document
---------------------

At the last meeting Eric, Murrough, Norman and Tara agreed to look at our
requirements in the areas of Monitoring, Test & Diagnostics, Control and
Calibration software respectively. Tara subsequently felt he needed extra
input into the Calibration area, so before the next meeting it will
discussed between Reg, Tara and other RAL based members of the group.

Eric, Norman and Murrough presented their thoughts on the other topics.
Murrough will try to update the draft requirements document with this
extra input. (Again in lieu of typing it up here....at least for now).


Future Workplan and Timescales
------------------------------

For the last 15 minutes of the meeting we tried to set out our next steps
and what should be done by (or at) the next meeting.

We hope that Steve Fisher will tell us about StP then. Meanwhile Tara will
try to put our existing DAQ and/or diagnostic software into CVS. Perhaps
first on an NFS mounted area (accessible to LynxOS) with the intention of
moving to AFS (either RAL or CERN cell) at a later date. We should enter
both the latest and previous version, or else test making modifications
to the latest version as a way of gaining experience of how CVS works.
In our naming conventions we should be consistent with future use of SRT.

Norman will worry about our short term planning for 1999 when the prototype
ROD, its associated data source and sink module (DSS) and the ROI builder
will arrive. It is possible that a version of the DSS will arrive even sooner
than first thought. Scott will look into the plans for backplane tests using
the new cards designed by Richard Staley.

[Scott will also try to get onto the CERN courses: Object Oriented Analysis
and Design; and either a C++ or Java course].

Murrough will try to use the database component of the DAQ -1 software and
send reports to the next meeting via Eric. [NB Murrough will be away from
21 Dec 1998 to 2 Feb 1999 inclusive]. Norman may try to look at the MRS and
IS components and will also look into upgrading to LynxOS 2.5.

Murrough will incorporate new text into the requirements document. Meanwhile
Reg will convene a group to think about calibration.

We also need to think about our data acquisition architecture. Is it really
ideal for our (DAQ) purposes? If not how could we improve it? What facilities
might we need in the RODs etc etc...