STC-S Regions (was: Boxes and Polygons)

[Arnold Rots]

I have been trying all day to find time to wade through the email
deluge of the weekend. I think I have collected the relevant issues
and unanswered question that concern STC-S's definition of Box and
Polygon.

In discussions with Markus Demleitner and David Berry, I have been
working over the past year on an update of the STC-S note.
you will find the current version at:

    http://hea-www.harvard.edu/~arots/nvometa/STC/STC-S-20091030.pdf

Several restrictions have been removed, most notably compound regions.
And there are two remaining issues to be worked out:
- STC-X does not allow more than one PositionInterval, for subtle
reasons that have to do with XSD syntax; but there may be a way out.
- STC-S spherical Polygons are limited to great circle sides; adding
small circles would not be too difficult, if it were not for the fact
that if one were to allow a pole to be specified things get very
messy; maybe only small circles with default pole should be allowed?
Note, though, that small circles can be specified if one uses
Convexes; at the same time, though, it means that this is a case
where STC-S does not allow full and transparent translation between
Convex and Polygon (as opposed to STC-X).

Inside and outside:
Tom (I think) is right that users may not pay attention to the
ordering of their vertices, but somehow they must specify what it is
that they mean, and going around the area counter-clockwise is as good
as any. This really should be taken care of in the user interface.
Note that this is an issue for Polygon, not for Convex - the operation
there is more involved than simply reversing the order.

Small circles
Francois is correct: reversing the order of the vertices without
moving the SmallCircle elements changes not only the meaning of inside
and outside, but also the specification of the small circles. The
reason is simply that the Vertex elements are a combined specification
of vertex and side.
Note that this is not an issue when using Convexes.

Boxes:
Read the definition carefully. It says that the arms of the cross must
be parallel to the coordinate axes AT THE CENTER POSITION. And it says
that the sides are great circles perpendicular to the arms of the
cross AT THEIR END POINTS.
It also says that the Box is a special case of a Polygon and that the
vertices of a Polygon shall be less than 180 deg apart.
The intent is indeed, as John Good and other remarked, for the Box to
be used to specify image bounds - i.e., rather small Boxes.

For "coordinate rectangles" there is little need for a special region
definition, since it is so easy to specify in SQL:
  ... where ra between 60 and 70 and dec between 30 and 40
In STC-S one would use a PositionInterval.


I think this covers the issues that have been raised in the last few
days. Let me know if I missed anything - I guess I should say: let me
know what I missed. I have not checked the coordinate transformations
that were suggested.

Cheers,

  - Arnold

--------------------------------------------------------------------------
Arnold H. Rots                                Chandra X-ray Science Center
Smithsonian Astrophysical Observatory                tel:  +1 617 496 7701
60 Garden Street, MS 67                              fax:  +1 617 495 7356
Cambridge, MA 02138                             arots at head.cfa.harvard.edu
USA                                     http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------