[evla-sw-discuss] Higher level software

[Barry Clark]

e2e have expressed some concern that they have to start designing 
an Observing Tool (a JObserve-and-more thingee) with no idea as to
what it is going to have to put out in the way of control scripts for
the real time system.  I am going to start thinking about that, and
will be sending messages to this exploder bearing on these questions.
For many recipients these will elicit an "I'll think about that when
the time comes."  But things are likely to become increasingly solidified
as we go along, so I encourage people to think about things now, and
raise points for debate at the earliest possible times.

I am thinking about offering initial thoughts in four areas, which may
proceed as quasi-independent threads.

1.  Design philosophy.  My message of August 28 can be considered to be
the first of this thread of the architecture of the observing layer 
software.  I will soon send a message on a proposed architecture in the MIB.

2.  Language choice.  We should obviously consider, for the control 
script syntax, a.) Python, b.) Glish, c.) XML, d.) home-brew.  These
interact in various and interesting ways with the choice of implementation
language: a.)C++ or b.)Java.  And, of course there may be other choices
for both of these.

3.  Control script primatives.  (What are the methods to set up the LO
chain, for instance.  Or how do we want to handle antennas of different
types.

4.  Subarraying.  May be part of the item above, but I consider it a 
particularly difficult one, because it pervades the entire tool chain
that stands between the astronomer and the observation, and second, 
because it is as yet unclear how the desire to subarray can be best
expressed in a way understood by both man and machine.