[fitsbits] start of Public Comment Period on the CHECKSUM convention

[THIERRY FORVEILLE]

> Rob Seaman" <seaman at noao.edu> wrote:
> On Jul 5, 2015, at 11:05 AM, THIERRY FORVEILLE
> <thierry.forveille at ujf-grenoble.fr> wrote:
> >> From: "Erik Bray" <embray at stsci.edu>
> >> 
> >> On 07/02/2015 06:14 AM, Mark Calabretta wrote:
> >>> However, because metadata should not be stored in the comment field,
> >> 
> >> Unfortunately the ship has already sailed on that one when it comes to
> >> FITS, due to the use of comments to store unit information.
> >> 
> > I had fought that one, on the argument that there is no good reason to use
> > anything but the base SI units (e.g. meters, not kilometers), but I was
> > outvoted ;-) As you say, that ship has sailed...
> 
> The ship has sailed, but like all ships can return to port whenever we like.
> 
> As with the other issues of long keyword names and values, versioning,
> inheritance, etc, we can address units when transitioning from the current
> ASCII headers to a binary table metadata representation.  Just add a column
> for units.  A column for an (optional) timestamp.  And a separate column for
> human-readable comments with no special-handling for units or timestamps.
> 
My position being that units are a nuisance because writers should 
convert to the non-ambiguous base units of the system (the default when
no unit is specified) rather than bother readers with yet another task,
I am afraid that the "once FITS always FITS" rule prevents that particular
ship's return ;-) At most we could declare this a discouraged style,
on a par with the random groups, but the return doesn't seem worth
that trouble.