DATE-OBS='31/12/99'

[Preben Grosbol]

'Peter Bunclark <psb at ast.cam.ac.uk>' wrote:

>What is happening about the millenium?  Is there any proposed way of dealing
>with the 2-digit original spec. of the year of observation.

 We at ESO have also started discussing this issue in the process
of defining the output FITS format for the VLT.  Internally, some
proposals have been made, however, mostly in a style which does
not ensure that old FITS readers would get it right.

 The only safe way seems to be define a new keyword with a more
general format e.g. as in Peter Bunclarks proposal.  I see not
problems with name 'DATAOBS ' and known major organization which
uses it with some other meaning but may be somebody turns up.

 For the value field YYYYMMDD, my main comment would be that
ISO has defined a general date/time format (ISO 8601:1988/1991)
which should be considered.  If I remember right it also starts
YYYYMMDD and is of string type but has an option for time.
As soon as I get the printed copy of ISO 8601 I will check
if it sensible for the value field of 'DATAOBS ' and post it.

 This keyword would be redundant considering 'MJD-OBS' but I
do agree that a more readable format also should be available.

Preben Grosbol