Astronomers have been asking for the ability to specify NDF sections
using WCS coords. I'm trying to decide on a way to do this. I think
it's going to require some new syntax because the NDF library needs to
be able to decide wether the supplied bounds refer to pixel axes or
WCS axes. This is because there is not necesarily a one-to-one
relation between WCS axes and pixel axes; the number of WCS axes need
not equal the number of pixel axes, and they can be in a different
order. So, when a user supplies a comma separated list of bounds
expressions after an NDF name, the NDF library has to be able to
determine which axis (pixel or WCS) each expression refers to. In the
old days it was simple; the N'th expression in the list referred to
pixel axis N. Now it's not so simple.
My proposed solution is to say that if a section specifier is enclosed
in parentheses
then it follows the old rules, but if it is enclosed in square
brackets then the NDF library assumes that the N'th expression in the
list refers to WCS axis N, rather than pixel axis N. It will not be
possible to specify pixel indices when using this syntax (i.e. integer
values will be interpreted as WCS axis value).
It's going to need some other syntactic changes as well. For
instance, since ':' is used as a separator in RA and Dec values, it
cannot also be used to separate lower and upper bounds in a NDF bounds
expression. So I'll probably use "|" to separate lower and upper
bounds in the new syntax rather than ":".
If anyone has any comments on any of this, let me know.
David
|