Level 1 Calo Software Meeting at Heidelberg on Thursday 2 November 2000.
------------------------------------------------------------------------

The agenda was:

- Experience with new VME PC in Mainz..............Juergen (10')
- HDMC progress/plans in Heidelberg................Cornelius (15')
- Recent HDMC/DAQ work in the UK...................Bruce (30')
- DAQ -1 integration plans and timescales..........Murrough (10')
- Software required for the slice tests............Murrough (30')
- Preparations for November TDAQ Workshop..........Murrough (10')

Experience with new VME PC in Mainz [Juergen]
---------------------------------------------

Juergen discussed the new Concurrent Technologies VME PC which
they have at Mainz.
His slides may be found here under Talks.

The Concurrent board was chosen after a brief market survey. VMIC
is too expensive and Eltec didnt have a Pentium III board at the time.
The standard SuSE distribution was installed, with a patch for Xfree.
The Ethernet isnt recognised correctly and must be manually selected.

Two VME drivers are available:
- one from www.vmelinux.org. This is recommended by Concurrent
  and is known to handle bus errors, but is not yet incorporated
  into HDMC.
- also, the Hannapel driver used already in the existing Eltec board
  at Mainz. This is already in HDMC and is reported to have better
  performance than the vmelinux one, but it doesnt handle bus errors
  at all.

HDMC progess/plans in Heidelberg [Cornelius]
--------------------------------------------

Cornelius described recent improvements he has made to HDMC.
His slides may be found via here.

The main change is that the long existing GuiCollection has finally
been implemented successfully using some classes from the KDE library
(thought KDE itself is not required). Many kinds of GUI can now be
grouped together in a completely flexible way.

There is now a facility for a local Makefile. This can be used to
keep preferred site options and fixes without updating the main
Makefile in CVS with trivial and possibly conflicting choices.
HDMC is now always built as a shared library due to problems on HP.
Cornelius would like to move to using autoconf and allied tools to
support shared libraries on various platforms. ATLAS has still not
chosen the replacement for SRT, but as any replacement is likely to
use autoconf we may go ahead anyway.

Cornelius demonstrated the new GuiCollection and the attractions
of the new Qt Designer tool for creating new GUIs.


Recent HDMC/DAQ work in the UK [Bruce]
--------------------------------------

Bruce presented his work on using HDMC as a DAQ toolkit.
His slides are here.

Bruce listed a number of facilities which are required to build DAQ
functions, from RTTI, containers and databases to VME access etc.
HDMC provides most of these, but not always in an ideal way (for the
DAQ) as it is optimised with GUIs in mind.

Bruce has extended the HDMC Parts file to support additional
subgroupings of a Modules parts according to function. This is
particularly useful for the DSS which has a complex intermingled
set of registers and memories, not organised with OO software
in mind. Creating these subgroupings as Module subclasses allows
reuse of the ModuleView GUI to provide simpler functional views
of parts of the DSS.

These Module subclasses implemented a few DAQ-related methods
for initialisation and readout. This then allows a simple list
based implementation of a DAQ "producer" program.

His wish list for the future includes extension of the grouping
scheme and perhaps implementation of higher level remote access
- ie calls to Modules as opposed to low level bus accesses.



DAQ -1 Integration Plans and Timescales,
Software Required for the Slice Tests;
Preparations for November TDAQ Workshop [Murrough]
--------------------------------------------------

Murrough gave an overview of recent developments and plans of DAQ -1.
He also outlined the range of software we will require for the slice
tests and ended by mentioning the imminent TDAQ workshop.
His slides are here.

Most interest was roused by the announced "Readout Crate in a PC"
solution: using Slink-PCI cards and software ROBs running in the
PC. This may be a useful alternative to either the "DetDAQ" which
(when written) or to continued private developments of our own on the
lines of Bruces work above.

The DetDAQ solution should be able to read out RODs and create event
fragments - at least in a single crate. Event building between crates
may required additional development.

The list of software we need for the slice tests is quite long:
- run controllers able to initialise and load all our modules
- readout code (in one framework or another)
- various monitoring and analysis programs
- programs to create suites of test vectors
- calibration and test programs

The effort available for this tasklist remains small. We will
have to decide which are the most essential and urgent ones.

The November TDAQ workshop has two sessions concerned with level 1.
One is on dataflow issues: mainly progress towards various integration
tests. The other is to discuss our requirements on and use of the Online
software and to provide our feedback on our experience so far. We need
to quickly write a short document summarising our expected use of the
Online software (listing the various Use Cases).