REPORT FROM ISAS  10/20/91  (1153 JST)
L. Acton

Tremendous progress is being made on the reformatting and analysis
software front.  The other night Mons, with Jim's help, thoroughly
reorganized the directory structure in our analysis system in a way
that will probably serve us well.  These guys are simply amazing in
their understanding of what will work well for powerful data analysis.
Such skills are as fundamental as differential equations to modern
experimental science.  Anyone interested in the nitty-gritty can
ask and I'll be happy to send Mons' recent memos on the reorganization.

This whole experience has given me a good "lessons learned" I'd like
to share.  The program plan for an experiment like SXT needs a lot
of very careful thought given to reformatted file structures pre-launch
to assure that the important information needed for analysis is gathered
in such a way as to make the finding of data efficient in a multiple
dimensional search space.  Constructing such a framework and testing
it on simulated or test data in advance is crucial.  Then post-launch
put the creators of the plan together in an intense post-launch
environment with users, like me, and commence the process of finishing
and polishing the plan.  Dumb users uncover all kinds of needs that
are not recognized by the experts.  It is foolish to think that the
pre-launch software plans will be right, no matter how much effort and
$ are spent on them.  There is a certain level of pre-launch work
which is appropriate--beyond this you are wasting $ and getting locked
into something which will not be right.

Finally, the use of a bottom up approach in software module developement
(building basic modules for data access/analysis first and stringing
them together as time progresses, all within a clearly defined framework
of file identifiers, etc) gives access and power at a low level 
immediately.  Last but not least, choose a powerful software system
(in our case IDL) and build the system within it.  The user friendly
and intuitive IDL capabilities give tremendous analysis power right
away.  And, oh yes, for goodness sakes do not skimp on the analysis
hardware--try to have it largely in place and strung together pre-launch
so you are not trying to do everything at once.  Buying equipment on
gov't contract is a very time consuming process.  And when you do
buy, buy POWER.  The most precious resource, in the end, is scientist
time (also the most expensive!).  It is wasteful for people to wait
for the computers to catch up and for people to wait for the availability
of a seat at a workstaton.

Enough already!

Last night I looked at a very interesting flaring active region at
around 1100 UT on 10/17/91.  There was a cluster of loops in a 
"rosette' arrangement around a sunspot.  Flare activity was going
on with large structure amongst the loops and lots of tiny flares
occuppying only a few full resolution (2.5 arcsec) pixels popping
off continually in the dark center of the rosette.  Simply beautiful!

Cheers,

Loren Acton