[fitsbits] 64-bit integers

[William Thompson]

Folks:

I've recently been asked whether my binary table software for IDL could handle 
64-bit integers, and had to respond that FITS does not currently support them. 
I remember that there was discussion a while back about adding support, but the 
last message I could find on the subject was from October 2001, reproduced below.

It's probably about time to revisit the subject.  Personally, I can only speak 
from the perspective of IDL.  64-bit integers were first introduced into IDL in 
vresion 5.2, which was released in late 1998.  In 2001, that would still have 
been a fairly recent change, but nowadays we tend to consider anything before 
IDL 5.2 (and sometimes 5.4) too old to be supported.

In the spirit of supporting innovations on a trial basis, as Bill Pence mentions 
below, I could certainly add support for 64-bit integers into my software.  The 
obvious implementation would be to use BITPIX=64 in primary HDUs and IMAGE 
extensions, and TFORM='nK' in binary tables.  Does anybody see any problems with 
that?

Bill Thompson


The earlier message from Bill Pence on 10-Sep-2001 is below.

> This reports on the result of the straw poll that was taken during the week
> of 25 September 2001 on the issue of adding 64-bit integers as a native FITS
> data type.  The possible choices in the poll were:
> 
>   YES - I think 64-bit integers should be added to FITS
>   NO  - I do not think 64-bit integers should be added to FITS, at
>           least not at this time.
>   UNDECIDED - I have not yet made up my mind
>   NO OPINION - I don't care one way or the other
> 
> In total, 29 votes were cast in this poll.  To put this into context, there
> are about 50 - 75 subscribers to this FITSBITS mail list, plus an unknown
> number of readers of the sci.astro.fits newsgroup.  The distribution of
> votes that were received is as follows:
> 
>    YES         20    69%
>    NO           4    14%
>    UNDECIDED    3    10%
>    NO OPINION   2     7%
>       Total    29   100%
> 
> This suggests fairly strong, but certainly not unanimous support for this
> proposal. 
> 
> Where do we go from here?  My personal opinion is that we can afford to wait
> ~~1 year and then raise the issue again.  By then, the percentage of computer
> platforms with 64-bit operating systems will have increased, and the need
> for supporting 64-bit integers may become more apparent.  In the meantime,
> the CFITSIO library as well as several other FITS libraries will support
> FITS files containing 64-bit integers on an trial basis so that programmers
> and users can begin experimenting with this data type if they wish.
> 
> -Bill Pence



-- 
William Thompson
NASA Goddard Space Flight Center
Code 612.1
Greenbelt, MD  20771
USA

301-286-2040
William.T.Thompson.1 at gsfc.nasa.gov