This was the first time during an AIPS++/CASA test I was able to
reduce a dataset end to end.  Though this is partly due to the fact
that I had much more time dedicated to it than in previous attempts,
it also tells me that the software has matured and especially that
the task-based approach is a promising new development.  The CASA
group deserves to be commended for this.

On the other hand, there remains a lot to be done, which is not sur-
prising as this was an alpha release.  The system suffers from not
having been exposed to this particular type of testing nearly enough.
It seems that at every little side exploration of parameter space new
problems were found.  I therefore strongly suggest that tests of this
kind be continued, either officially as during the past week, or by
everyone individually, in his/her own time.  But in my mind, the
former is clearly superior:

       testers have set this time aside a long time ago which makes
       it easier to ignore/postpone other demands on one's time
       it allows daily progress meetings which I found highly useful
       the CASA people know that during the week they need to be
       'on call'

I still have trouble maneuvering freely in this new framework, and
the discovery of new bugs slows down any momentum, even though the
CASA group has been very responsive.  During one of our daily meet-
ings Crystal pointed out that she has been working for days on a
reduction that would have taken her 30 minutes in AIPS.  That, in my
opinion, was not an exaggeration.

Some of this is due to old habits.  I simply don't like having to
type execfile('taskname.last') instead of tget taskname.  I will be
the first one to admit that I am heavily conditioned by 25 years of
AIPS exposure, and that everything that is non-AIPS feels 'not-right'.
It takes longer practice in CASA than just a week to separate depart-
ures from AIPS from true shortcomings in the new system.

Finally, a word about the Cookbook.  It contains a wealth of inform-
ation, and I have consulted it many times.  What makes it confusing
in my opinion is the addition of a lot of principles of interferometry.
It attempts to explain everything!  I think this detracts from what a
cookbook should be like: a rather dry summary of what to do and when.
Explanation of the underlying principles should be in a completely
separate document, and enough of those exist already.