VOunits draft

[Rob Seaman]

On May 22, 2009, at 2:40 AM, Francois Ochsenbein wrote:

> cgs: the IAU recommends the usage of SI since 20years now, and lists  
> especially erg, gauss, etc as "units which use is deprecated"

If the IAU has made no progress on this issue in the last 20 years,  
one might question whether we can expect that the IVOA will in the  
next 20.

> therefore I feel we should encourage this change,

...however, unlike coordinate conversion issues recently discussed,  
converting (basic) cgs units back and forth to SI is straightforward.

> especially if we consider that it is important that our data can  
> easily be understood by users from other disciplines;

More important yet is that the data be understood by the users within  
each discipline.  Whether or not IVOA should adopt a policy of  
encouraging a change in future astronomical usage, we are required to  
support current usage.  This implies that our specifying SI where cgs  
has traditionally held sway is a requirement to support both kinds of  
units.

It may be that we can standardize on SI usage for transport, storage  
and schema, but as with my similar comment a few days ago, this would  
just imply that cgs must pop up again in the external interfaces.   
Surely IVOA recommendations should cover use cases regarding  
coordinate conversions in such interfaces and user applications?

By analogy, consider celestial transients.  A single data set will  
include many classes of event.  Planetary scientists may be looking  
for KBOs, but cosmologists for distant SNe.  VOEvent has to be able to  
describe both, however, because both appear in the data and each class  
of event contaminates the other.  Similarly, cgs is a contaminant to  
SI purists.

We can mandate usage within the IVOA.  We cannot outside.  Our  
standards must cover the broader of the two, their union in a Venn  
diagram.

Perhaps VOunits needs three sections?  1) a list of "thou shalts", 2)  
a list of "never under any circumstances" - or simply a statement that  
"units not mentioned (eg, from firkins to fortnights) will not be  
referenced in IVOA documentation and will not be used in IVOA systems  
or applications", and 3) a (hopefully brief) list of units that are  
deprecated within IVOA, but that may be encountered in external data  
sets and that IVOA applications and external interfaces *may* choose  
to support conversions to/from.

Rob