Standardising units and formats (and ref frames?) in transmission

[Alberto Micol]

On 19 May 2009, at 18:53, Rob Seaman wrote:

> Ok, so an astronomer has a list of sexagesimal coordinates. Say the VO 
> only accepts decimal degrees. The astronomer needs a tool to manually 
> perform the translation. Somebody has to write that tool. One could 
> (and users will) interpret the person who writes that tool as being a 
> "VO programmer". We simply can't escape these requirements. 

Very good example to illustrate the point of an astronomer that
wants to build her VO tool...

I guess any astronomer is smart enough to be able to (and is also 
used to) perform such simple transformation (sexagesimal to decimal
degrees). 

At the contrary let's examine how the astronomer would have to work out
that same example, were such simple standardisation not in place...

If the VO accepts anything, well that "anything"
must be qualified by the astronomer in VO terminology. 
The astronomer will have to know how to represent that "anything" such that the VO can make sense of it, and this might mean knowing how to compose UCD, units, etc. for *the specific representation* this astronomer has in mind.


My bet is that it would be much simpler for the astronomer to perform
the conversion to the "standard/unified" representation, and use the ucd/units/format the VO recommends (or dictates ;-)) for such quantity,
than having to (1) understand all those VO metadata concepts, and (2)
having to *choose* among the *huge* set of options the current VO 
offers (what is the UCD/datatype/etc I should use for my "anything"?). 
It is faster, and less error prone, to grasp a fixed (dictated) 
standard instead of having to take various decisions on various 
VO attributes. And it is more interoperable too.


The user wants to use a standard, WITHOUT the need of understanding it
(if not really necessary).

For example, it could be written somewhere that angular coordinates are 
always in decimal degrees, the units are hence fixed to "deg", and the 
UCD is pos.eq.ra for a right ascension, pos.galactic.lat for galactic latitude, etc. 
With such list (which should cover the most tipical cases) the 
astronomer will gladly copy paste this info in her tool. 

The less the users need to understand the VO, the higher the user uptake.

>
>     The VO can only help by passing around information in a standard way. 
>
> Indeed! But this is a separate (and orthogonal) project requirement. 
> Standardizing on as simple a set of interfaces as possible will 
> clarify the VO system architecture, but it won't magically remove the 
> complexity inherent in astronomy.

Nobody wants --nor can-- remove the inherent complexity. 
But isolating it at the right level will greatly help the 
development of the VO.

Alberto