Date: Mon, 27 Mar 2006 17:22:22 -0700 (MST)
From: Walter Brisken <wbrisken@aoc.nrao.edu>
Subject: Complex TASKs : an idea


Today we discussed setting parameters for tasks such as imager that have
many parameters, many of which are not always used (based on type of imaging
being done), some of which have values that rarely need changing (wg.
convolution function, uvrange), and some of which can be automatically set
to reasonable values that are reasonable most of the time (cell size, imsize
and wproj parameter for mosaic).  One possible solution is to enter the
input screen for the imager task with additional non-mutually exclusive
qualifiers, such as SIMPLE, EXPERT, WIDEFIELD, MOSAIC, CUBE, DIRTY; ie one
might say

In[23]: task imager:SIMPLE,WIDEFIELD

By doing so only the relevant parameters are displayed to the user, and the
user will know that each must be set to something sensible.  In this
example, SIMPLE might imply that cellsize is determined by the UV coverage,
the image size is determined by the primary beam and all channels are
imaged.  Widefield would imply that w-projection/wfclark are used and would
set the primary beam options, fft machine, ... Some parameters (convolution
type is a good example) would only be settable if EXPERT is chosen.  In the
hypothetical "gravy gui" all parameters could be viewable with ignored
parameters greyed-out or collapsed in a tree.

I think this might be one way towards a "one size fits all" task interface
for complicated tasks such as imager.

-W

Date: Tue, 28 Mar 2006 12:30:34 -0700 (MST)
From: Michael Rupen <mrupen@aoc.nrao.edu>
Subject: Re: Complex TASKs : an idea

Hi Walter,

I personally prefer the hierarchical scheme that David suggested,
where one can access more detailed parameters whilst working on the
actual inputs.  I would be really annoyed to get halfway through the
inouts and then realize that (e.g.) cellsize wasn't settable in the
view I chose when starting the process.  I haven't much cared for
"expert" vs. "casual" views when I've run across them in the past
(e.g., AIPS table flagging) -- they seem to place a barrier to
learning as one goes, and to accessing the one "expert" ability that
one really needs in the midst of mostly "casual" use.  But probably
you've finished this discussion already this morning :)

  Cheers,

            Michael