[QUANTITY]: gather requirements for quantity model

[DIDELON Pierre]

> From: Brian Thomas <thomas at mail630.gsfc.nasa.gov>
> To: Patrick Dowler <patrick.dowler at nrc-cnrc.gc.ca>
> Subject: Re: [QUANTITY]: gather requirements for quantity model
> Date: Tue, 28 Oct 2003 18:03:33 -0500
> Cc: dm at ivoa.net
> 
> On Tuesday 28 October 2003 04:46 pm, Patrick Dowler wrote:
> > At some low levels we are just doing CS; if we are not then we are
> > totally screwed because we are mixing domain-specific knowlegde
> > and concepts with basic fundamentals of math and comp sci. That
> > path leads to disaster.
> 
[snip]
> 
> 	And just when was "quantity" ever defined by us (the VO DM community)? 
> 	If it has, then we would have no need for requirements. 

General requirements must come from the whole VO community, and the DM group
have to implement them in a model. Because the outside community did not always
speak the same langage as the DM group, (and even in the DM group langage
uniformity/agreement seems not to be perfect), this process is already a loop
with "translation" each time we cross a group domain. Horrible yet, isn't it?

However we have some indications which can help us.
Discussions during the meetings (without written traces I admit)
and previous mail on the list. For example the one of clive Page
(http://www.ivoa.net/forum/dm/0086.htm), as far I understand the english
and technical part, he seems that he claim for a simplistic, immediate 
( yesterday like always ;-) ) Quantity with one/few values
to exchange very small pieces of informations.

Moreover, if we look things going on in the VO community we (or only me?) 
can see that :
- There is already a container for table (VOTable) and all possible related
tabular value (obslog, timeseries, spectra ...) which tends to be 
enlarge to be used as packaging for more complex data (images, images list and
set, data cube...). This structure "encapsulates" data and metadata, which
is essential for VO, perhaps not in a perfect manner but it has the advantage 
to exists.
- The VO community, or at least a certain part, ask our leader (Jonathan)
to make a DM for spectra. It can be solve by the definition of metadata
to be introduced in a VOTable to make it a Spectra understandable by all
the VO actors. But that's my opinion and I am perhaps (if not certainly)
wrong. The modelling of spectra would tell us this.

So. VO community seems to "have" sufficient tools/concepts to handle
"complex data" and is asking us (?) a model to handle "simple data"
to be able to tranport it across the whole community between services,
computers, humans....

> 
> 	What is the CS "quantity" definition, btw?
> 
It don't include data cube I hope ;-)

In conclusion, I think that we must first agree on what is the QUANTITY
needed by the VO community;
1) simple quantity containing atomic value or arrays of values 
2) complex quantity able to represent any kind of data structure
usable in the VO community.

could we try a vote ;-)

Pierre