L1 Calo Software Video conference at RAL/Heidelberg on 9 December 1999.
-----------------------------------------------------------------------

This is a brief summary of what we discussed during this first attempt
at a video conference between the UK and Heidelberg software groups.
informal meeting.

Present: Murrough, Kithsiri, Scott, Tara, Bruce; Cornelius, Volker.



HDMC Developments at Heidelberg
-------------------------------

Many improvements have been made, mostly by Volker, since version 0.1
was released around the time of the Stockholm meeting.

- The online documentation has been converted to proper HTML and is now
  available also via the web.
  [NB the UK group has found the existing online help very useful!]
- A simple scripting facility has been added (also documented). This
  can open GUIs and perform actions on Parts. 
- Memory and Register Parts are now IOFrames (before there were related
  IOFRame subclasses which could be connected to Memories and Registers).
- The histograms are improved: no longer automatically cleared when new
  data is entered (ie can build up plot of a succession of register
  reads).
- IOFrames can be simply trigger via GUI buttons.
- Improvements to the Part Manager and now a separate GUI Manager class.
- Clean up of some GUI code: common elements now in superclasses
- Improvements to the Assembler Part
- the PartCollection class is now debugged and works "within its
  existing limitations"
  
  
HDMC experience in the UK
-------------------------

Progress towards getting HDMC working in the UK has been difficult,
mainly because the 68K version of LynxOS 2.4 doesnt have any STL at
all (the PPC version of 2.4 merely has big problems with the STL).
An old C version of the busserver has been modified for the MVME167
boards and works, but as this doesnt use the same Bus Parts as the
rest of HDMC, and changes to the protocol would have to be implemented
twice and risk getting out of step.


HDMC Discussion
---------------

In the UK we may try either or both of LynxOS 2.5 (available at RAL
and QMW, but not Birmingham?) or Linux on our 68K processors.

However we agreed that for C++ code development we should only use
platforms which support the full C++ standard and provide a working
Standard Template Library. It will not be worth our time keeping old
platforms going for too long.

The busserver protocol has not been changed recently, but might need
changes if bus error handling is implemented in HDMC (which is
desirable).

It was also agreed that a compact GUI for modules was also needed.
This could have some "whole module" functions, but with displays of
registers etc via an upgraded PartCollection class. A compact version
of the Register class is also required. Since the latter is a complex
class it would be easiest if that work was done by Heidelberg - but
they are now busy with hardware and have less time for software work
until some time next year at least. So maybe a UK person should try
and contribute to this development.

Network VME access measurements
-------------------------------

Measurements of network access to VME between Linux and the MVME167
have been made at RAL. Adding block mode to the UK diagnostics gives
a large factor improvement for reading large memories in the DSS.
No equivalent measurements have been made in Heidelberg for the HDMC
busserver (but it was previously reported that performance is fine for
the user).


TTCvi Software
--------------

Kithsiri described his work on high level software for the TTCvi. He
has now implemented the classes described (though subsequently slightly
modified) in the note he distributed earlier. His next step is to
connect these to a hardware implementation - initially using the UK
diagnostics, later to HDMC (if he has time before we lose him back to
Sri Lanka).

He has been using a software tool called "Code Crusader" which he
demonstrated to the UK group after the meeting. This adds a GUI to
manage a C++ project, helps with Makefiles, draws class diagrams,
and also has a "Code Medic" product which is a GUI front end to gdb.
It seems as nice and light addition to the basic editor,make,compiler.
http://www.newplanetsoftware.com/jcc


Configuration Database
----------------------

We briefly discussed the list of configuration data which Eric had
circulated to the AT1SOFT list on monday. Cornelius thought there were
some missing items in the PreProcesor, eg control of the rate monitor
and histogramming. The PPrAsic has 32 registers to set, but some are
derivable from other settings - the dependency of some settings on others
needs to be thought out. He will email Eric with his comments later.

At some point we need to think how the Module & Crates envisaged in
the DAQ OO database could be integrated with the Part hierarchy of HDMC.
Eg taking a Modules subParts from HDMC definitions, but grouping of
Modules from the database. Certainly this should be considered before
any further work on the HDMC configuration. It also needs more work
on understanding the DAQ -1 proposed database schema.


Prototype DAQ system
--------------------

Murrough has made a very basic start on a framework to make realistic
use of DAQ -1 components (so far only really simple tests have been
made on individual components). This might be developed into a DAQ system
for our prototype full slice tests - if nothing suitable emerges from
the DAQ group in time.

An attempt was made to install this on the RAL Linux system after
the meeting and to demonstrate it - but various problems, eg lack of
java, were not quite overcome before the end of the day. Startup of the
DAQ -1 at QMW was however shown successfully over the network.



Future Workplan
---------------

Cornelius & Volker have to concentrate on hardware for the next month
or two.
Scott will try to add the TTCvi and DSS more completely in HDMC
Kithisiri hopes to finish his work on higher level TTCvi software
Bruce will look into LynxOS/Linux upgrades to our MVME167 systems
Tara is interested in learning more about HDMC internals....
Murrough will do a little more on DAQ -1 based framework before
taking a long and well deserved holiday :)


Next meeting
------------

No date was fixed, but another video conference in mid/late January
seems sensible.