From efomalon@cv3.cv.nrao.edu Thu Jan 17 08:55:52 2002
Date: Wed, 16 Jan 2002 20:27:16 -0500 (EST)

                SUMMARY:  NAUG TELECON  Jan 16, 2002

                            Ed Fomalont

Participants:  Crystal Brogan, Claire Chandler, Mark Claussen, 
   Kristy Dyer, Ed Fomalont, John Hibbard (CV), Ralph Marson,
   Steve Myers, Frazer Owen, Greg Taylor, Jim Ulvestad

---Summary of the AIPS++ targets for V1.7.  Nearly all of the NAUG
suggestions were incorporated into the targets for the next four(?)
months.  One additional piece of software that would be useful for
D->A configuration in the next few month is to apply the phase
correction for antenna position errors.  The anticipated large AIPS++
effort on the viewer prompted discussion, led by Crystal and John,
about the future of the viewer.  Will its present 'feel' remain as is,
with improvements, or is there a chance of a redesign?  This topic
should be discussed in the Jan 30 meeting with the AIPS++ group, and
collaboration of the implementors (some of them in Australia) with
experienced data graphics users are suggested over the next few
months.

---NAUG target area of testing.  The list of NAUG/AIPS++ target areas
were briefly discussed.  Some additional testing should await recent
code improvements, especially in the flagging and calibration areas.
Claire discussed the results of applying the VLA telescope gains at
the high frequency, based on scripts developed by Steve and George.
The AIPS/AIPS++ comparison agreed to about 1%, but the application of
the gains was somewhat painful.  Incorporating these corrections into
the VLAFILLER and measurement set is on the AIPS++ target list.  Ed
summarized the results of the imaging efficiency tests, which were
sent to the AIPS++ group the previous week.  Feedback from the AIPS++
group and others suggested that the relatively poor AIPS++ performance
has been a serious concern and the group will begin developing
aips-like DDT timing tests.  Athol already reported finding a bug in
the FITS to ms tool which improved its speed considerably.  Frazer
suggested that Eric Greisen might have good suggestions to improve the
execution times since AIPS must also face similar computational
problems.  Kumar and Ed are working on the test u-v data set to try to
improve the execution efficiency, and to implement the potential
squint-handling ability of imager.

---Documentation and Scripts.  Little of the resources of AIPS++
development for V1.7 will be devoted to documentation, and there was
much discussion about how the NAUG might be effective in improving
this situation.  There is a basic difference in opinion between the
NAUG and the AIPS++ group about the organization of documentation that
is most useful to the synthesis user, and this difference may be
limiting any significant development or changes.  With additional
discussion, the NAUG felt that the implementation of a script library
for the AIPS++ system would be a good approach to illustrate the power
of AIPS++, to clarify the reduction of VLA data, and to include
important documentation within the script.

     The following plan is, hereby, suggested:

(1) The NAUG (represented by Steve Myers and Ed Fomalont) and the
AIPS++ group should decide on the basic organizational form of the
scripts.  A few of the topics could be: What to name the scripts; an
introductory paragraph describing what the script does; a convenient
place to input all user parameters and a consistent use of parameter
names through all of the scripts; how to include useful information in
the script about parameters as needed; some built in error handling
(egs does an input file exist?); useful on-line output during
execution; the ability to concatenate scripts with a minimum of
change.  Since we expect these scripts to migrate into the Recipes
section of the AIPS++ documentation, they should be easily convertible
into the suitable forms.

(2) The NAUG and AIPS++ group should email suggestions for scripts to
efomalon@nrao.edu, with an optional copy to aips2-naug.  If you know
of a similar script already available (before standardization), please
indicate where, or include it in the email.  Useful scripts do not
have to be only 'nominal' reduction steps.  For example is a script
with input <source name> and output fieldid# (and visa/versa) useful
to have?  Should there be scripts which show the user how to
manipulate the tabular data, use the pgplot package, balance their
check book?

(3) The NAUG will have the main responsibility of collecting and
revising, or writing, if necessary, the scripts which are selected.
They will be put in 'standard' form and tested.  Consultation with
the AIPS++ group will clearly be needed.

(4) The script will be put in an appropriate place, probably in Steve
Myers' web site, for further testing by the NAUG and others.
Improvement in the documentation within each script and ease of
usability should also be considered.

(5) Just before an AIPS++ release, the set of verified scripts should
migrate into the AIPS++ package.

    There are other issues which should be considered, but probably
can wait until implementation of the above steps.  For example, how
can NAUG responsibility for the scripts be transferred to the AIPS++
group for further maintenance? how can the scripts be interfaced with
the existing or future AIPS++ documentation?