[fitsbits] 64-bit integers

[Thierry Forveille]

Markus Dolensky writes:
 > Forgive me if that has been brought up before, but as a relatively new 
 > observer of this list let me share an idea. Emphasizing the fact that 
 > FITS is a 'transport' format and that there will always be various, 
 > coexisting computer architectures (be it 8- or 64-bit computers) 
 > - what about allowing
 > 
 > BITPIX = [-]2**n   / where n = [3..14]
 > 
FITS is a practical transport format, with a strong emphasis on the practical
qualifier. The onus is thus on the proponent of any new functionality to
demonstrate a) that it is needed by existing or near-term datasets, and b) 
that it can be supported by at least of majority of existing systems.