[fitswcs] WCS announcements

[Mark Calabretta]

On Fri 2002/12/20 11:36:08 CDT, Doug Mink wrote
in a message to: Mark Calabretta <Mark.Calabretta at atnf.csiro.au>
and copied to: fitswcs at cv.nrao.edu

>1.  Changingthe format of subroutine names from *fwd and *rev to
>    *x2y and *y2x eliminates confusion.  I have always had to read
>    the code to figure out which was which.

Also the cartographic meaning of forward/reverse was lost with the
spectral WCS.

>3.  Will wcsini() recognized CDi_j and CROTAi as legitimate WCS keywords?
>    If it doesn't, hiding piximg and imgpix makes it harder to process
>    the CDi_j keywords which are present in a huge amount of data.  We
>    have found that it is not straightforward to go from CD to CDELT and
>    rotation, but it is easy to go to piximg and imgpix.  There might need
>    to be an alternate version of linset() or additional arguments--I
>    currently pass my equivalent program multiple parallel arguments
>    which are null pointers if unused.

The main complication with CDi_j is that its default values differ from
PCi_j so it's not simply a matter of setting CDELTi to unity.  A detailed
solution was given in an earlier draft of Paper III (since excised); I
need to dig it up and refresh my memory.  A solution for CROTAn will be
provided along the lines of Sect. 6.1 of Paper II.  I'll make the WCSLIB
headers available for comment again once it's done.

>4.  Maintaining a single code base by wrapping the C code with Fortran
>    should be easier.  That's how I do it now, and it works well.

It also has (good) implications for PGSBOX.  It's interesting how all
the baggage in the wcsprm struct and other structs (pointers to allocated
memory, pointers to functions, etc.) can be passed to and from fortran
apparently with impunity.  Portability between platforms is an issue that
I've not closely investigated though.

Thanks for the comments.

Cheers, Mark