[fitsbits] 64-bit integer comments

[William Pence]

Thierry Forveille wrote:
> Unless somebody comes up with a project that will fly within 5 years
> and that needs 64 bits of dynamics in images, what's wrong with just
> waiting until some actual need occurs?

The projects I have dealt with typically make decisions about the format of 
their FITS data products on time scales of weeks or months.  The FITS 
approval process on the other hand has traditionally operated on a time 
scale of years.  Even though we are working to make the FITS approval 
process more efficient, it will never match the speed required by some 
projects.  Thus I think the onus is on us (the keepers of the FITS Standard) 
to anticipate future needs in a timely manner.

This also misses the point that there are literally hundreds (if not 
thousands) of individual students and researchers that write their own 
relatively small analysis programs that read and write FITS data files.  It 
seems very plausible to me that at least some of these researchers would 
find it convenient right now to write out a histogram of Integer*8 values to 
a FITS image, perhaps as an intermediate file that will then be read in and 
further analyzed by a second program.  I would argue we should be working to 
ensure that the FITS "toolbox" contains all the capabilities that reasonably 
might be useful to researchers in doing their work as easily and efficiently 
as possible.  This seems more useful to me than telling those researchers 
"no, you cannot simply write I*8 values into a FITS image;  If you really 
must store I*8 values then you must spend more time and effort to develop 
more complicated software to write them into a column of a binary table".

Bill Pence
-- 
____________________________________________________________________
Dr. William Pence                          William.D.Pence at nasa.gov
NASA/GSFC Code 662         HEASARC         +1-301-286-4599 (voice)
Greenbelt MD 20771                         +1-301-286-1684 (fax)