[daip] Re: aips_gripe

[Eric Greisen]

Martin Hardcastle writes:
 > {25-JUL-2000 15:37:20}
 > {scorpius     UN-REGISTERED!!!!!!!}
 > { 2500  31DEC99  SOL}
 > {Martin Hardcastle}
 > {m.hardcastle at bristol.ac.uk}
 > {+44 117 928 9051}
 > {Argh! Can we have the old behaviour of HGEOM wrt reference pixel back,
 >  please! It seems much more sensible to use the reference pixel of the
 >  template image as the default, given that HGEOM's main use is as a
 >  `make this image like that image' task. If people really want to transform
 >  the reference pixel to the new co-ordinate system, then let's have another
 >  option to HGEOM to do it... but the original behaviour should be the default.
 >   }
 > {I liked HGEOM the way it was!}
 > {m.hardcastle at bristol.ac.uk}
 > { }

Which way?  I just changed it this morning somewhat back to the way it
has been traditionally.  The oldest behavior had it select the ref
pixel of the 2nd image if the output image was > (in both axes) than
the 2nd and to jump abruptly to centering the input on the output if
1 or both of the axes of the output was < the 2nd image.  Since the
2nd image is often a dummy its dimensions should not count for
anything.  The NVSS regrids all their small images onto the same
geometry still small, but planning to go to a large field with more
tasks later.  I changed HGEOM and OHGEO so that they allowed the user
some control of the ref pixel - the user had no control before and now
he/she does.  But the NVSS people complained that the oldest bahvior
was needed for their usage, that figuring out what the ref pixel has
to be to center the input on the output is too hard  So I changed last
night the behavior so that the DEFAULT is to center the input on the
output, but to still give the user adverbs to control what he/she
gets.  If you want the ref pixel to be that of the 2nd image, it is
easy to type in the numbers or to use GETHEAD to fill them in.

Frankly, if I hear any more complaints about these two tasks, I will
revert them to their old form - with NO user control over the ref
pixel and with errors in the computation of coordinates or just delete
them from the system entirely.