[fitsbits] 64-bit integers

[Preben Grosbol]

With respect to the 64-bit integer suggestion by Bill Pence, I have
the following remarks:

 1) One should not add a new number format just because one can but
    because it is scientifically required.  I still do not see a
    strong science case for 64-bit integers especially in astronomy.

 2) Although 64-bit computers are more common now one should first
    introduce a new format when it is available on the vast majority
    of systems in the community.  I would still think that this will
    take some years more.

 3) Should there be a science driver for an integer format with
    more than 32-bit, one should very seriously consider to use
    the 64-bit floating point format with its 52-bit fraction part.
    There would be many advantages in adding BITPIX = -64 instead
    of 64 e.g.

      a) It is already available in the tables.
      b) It is well defined with NaN and Inf values.
      c) It is a standard format on current computers.

For the reasons listed above, I would not think a 64-bit integer
format is required yet.

Preben Grosbol