RA, DEC and UNSIGNED

[Peter Bunclark]

1. Please let's continue to allow RA and DEC in sexadecimal. One of the
most common
        quotes is ``FITS is good because when you have problems you can
just more the header'';
        the easy readability of the header is seen as a great strength. 
The RA and DEC are almost
        the most important keywords for end users, and should be in the
        format the consumer wants.

2. The lack of unsigned data type was, in my opinion, nearly the biggest
cockup in original
        FITS, right behind the EPOCH and DATE problems.  The overheads
of
        the BZERO hack are:
        a)  Compute overhead; either an extra bit of arithmetic, or,
more likely, the
            reader gives up and casts it all to float, causing lots of
extra arithmetic
            and I/O.
        b)  It *is* a hack;  I bet that some of the systems that FITS
was originally 
             designed for, eg 60-bit machines, need different code from
that on base-2
             wordlength machines (? or not?).
        c)  It brings FITS into disrepute every time you have the
embarrassment of
            introducing the system to a new astronomer or software
engineer.

Peter.