From bglenden@colobus.CV.NRAO.EDU Wed Apr  1 22:48:30 1992
Status: RO
X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil]
	["2208" "Wed" "1" "April" "92" "22:48:22" "EST" "Brian Glendenning" "bglenden@colobus.CV.NRAO.EDU " nil "55" "My summary of the day's confinement:" "^From:" nil nil "4"])
Return-Path: <bglenden@colobus.CV.NRAO.EDU>
Received: from cv3.cv.nrao.edu by colobus.cv.nrao.edu (4.1/DDN-DLB/1.5)
	id AA09506; Wed, 1 Apr 92 22:48:29 EST
Received: from colobus.cv.nrao.edu by cv3.cv.nrao.edu (4.1/DDN-DLB/1.12)
	id AA26877; Wed, 1 Apr 92 22:48:26 EST
Received: by colobus.cv.nrao.edu (4.1/DDN-DLB/1.5)
	id AA09500; Wed, 1 Apr 92 22:48:22 EST
Message-Id: <9204020348.AA09500@colobus.cv.nrao.edu>
From: bglenden@colobus.CV.NRAO.EDU (Brian Glendenning)
To: aips2-cv@colobus.CV.NRAO.EDU
Subject: My summary of the day's confinement:
Date: Wed, 1 Apr 92 22:48:22 EST

What did we learn today?

The exact role of the associator is still fuzzy. We fudged the role of
the associator by saying essentially that the associator was an
omnipotent user who gave "main" exactly what it needed. What comes in
via associations and what comes in via parameters needs to be
clarified. (It should also be possible to associate parameters with
datasets, to make things even more complicated).

The telescope is not needed - it is subsumed by the associator.

The IntImagingModel and IntTelescopeModel seem to hold up for the
spectral line self-calibration problem, however the details still need
to be worked out in detail (e.g. if you give a continuum image what
yegs get predicted, how do you allow for spectral index, etc).

When exactly does the IntTelescopeModel get created, and what kinds
are there and how and when do we want to associate them with the
IntYegSets?

We probably want "virtual" IntYetgSets that mimic the interface of
IntYegSets and create the Yegs on demand. Internally an imaging model
will need to be consulted.

Does the solver for polarization look like the solver we looked at in
more detail? What do we do with Q+U? What do we attach them to? Is
this (logically) history or does it need to be attached more directly
somewhere?

Nonisoplaniticity seems to require that the telescope and imaging
models be combined? How? Multiple inheritance or via some sort of
pointer/association binding? We need to think about this a bit more.

What next? It seems to me that we're just about at the point where we
need to make detailed study of a few things, which we should be able
to do in smallish groups (1-3?). E.g. my guess about logical groups is
the following:

IntYegSets         |
IntImagingModel    | To do a complicated spectral/multi IF and pol'n problem
IntTelescopeModel  | (Think about mosaicing/nonisoplaniticity)

SDYegSet           |
SDImagingModel     | Similar (but maybe smaller scope)
SDTelescopeModel   |

FITS/Blob

Association/Parameters

The interferometer group might well want to subdivide between the
imaging model and the telescope model and maybe stick the intyegset
onto the FITS/Blob group. Anyway, we'll discuss this in the morning.

Brian