SIA Evolution Proposal

[martin hill]

If I understand this right, you're saying you want to associate the same data 
with several elements.  For example, a set of observations of several thousand 
objects might have been carried out using two or three filters.

There is an 'XPointer' concept for XML documents which allows you to do this.  
Basically you can describe the filter in some metadata bit, giving it an 
attribute 'ID' and assigning it a unique reference.   Then you refer to that ID 
in an attribute or element in the main data using the XPointer format. 

This is pretty standard now (W3C have adopted it I believe) and seems to solve 
what you want; it also means you can happily describe the filter in as much 
detail as you like without having it all replicated throughout the list.

Cheers,

Martin


Quoting Clive Page <cgp at star.le.ac.uk>:

> On Fri, 16 Apr 2004, Alberto Micol wrote:
> 
> > >you could model your data as:
> > >
> > >results -- Filter -- Observation
> > >
> > >instead of
> > >
> > >results -- Observation -- Filter
> 
> > It looks to me that your model is indeed imposing the data provider view
> > to the users.
> 
> I'm inclined to agree with Alberto here: of course the famous rules of
> Codd and Date tell us never to duplicate data.  But here we have two
> observations (which are pretty natural units in the astronomical world)
> which just happen to have the same filter setting.  I don't see it as
> terribly wasteful to have the filter values specified twice, once within
> each observation chunk, even though that duplicates the information,
> wastes file space, etc.  When you (or your software) access the data, I
> would have thought it better to have the filter information in an expected
> place than to have to root around in the structure for it.
> 
> 
> -- 
> Clive Page
> Dept of Physics & Astronomy,
> University of Leicester,
> Leicester, LE1 7RH,  U.K.
> 
> 
> 


-- 
Martin Hill
07901 55 24 66
www.mchill.net