An STC "when" and "where" example

[Arnold Rots]

I think Rob is correct in asserting that we are not talking about the
same things.  Let me give my argument another try:

In my view the most basic requirement is:
A VOEvent should include all available information on the event that is
scientifically significant.

The criterion for inclusion ought to be scientific significance, not
what the service feels like implementing.
And it would seem part of our job to agree on a canonical list of items
that are considered scientifically significant.
I don't think there is any guessing here as to what some mysterious
clients might want or need.  It has been solved for the IAU Circulars
and I think we can do it for VOEvent.
As a first stab at what that information would be, at least as far as
STC is concerned, I would think: spatial position, time, spectral
band, position of the observatory - all as accurate as possible and
with errors and sizes/intervals/bandwidths.
An honest effort should be made by services to assemble all that
information, but obviously, if it cannot be produced, it will not be
available and hence cannot be included - that's fair enough.

Proprietary information, authentication, and restricted acc are is a
whole different kettle of fish; that is a separate discussion.

Hope this clarifies,

  - Arnold


Alasdair Allan wrote:
> 
> Arnold Rots wrote:
> > What I heard you say is that a VOEvent service should be allowed to 
> > provide the amount of detail that it feels like providing, and if that 
> > makes it useless to a client, that's too bad for the client and (s)he 
> > should just ignore the event.
> >
> > What I was saying is that that is not acceptable; it does not conform 
> > with accepted astronomical practice, as evidenced, for instance, by 
> > the requirements for IAU Circulars.
> 
> Sorry but that's nuts. What about cases where there isn't enough 
> information existing to provide enough detail for all the clients to 
> act on? Do we not send an VOEvent message out at all? How do you figure 
> out the level of detail required by the clients? You can't, there is an 
> unknowable number of clients, many perhaps hidden behind brokering 
> services which can't be seen by the process sending the alert, which 
> may require different levels of detail to carry out their science 
> programme.
> 
> This is nothing to do with the IAU accepted practices, and everything 
> to do with the way you build the architectures for reliable and robust 
> web services. The process sending the alert literally cannot have the 
> responsibility to assess the level of accuracy or information required 
> client side. It's not possible. It has to collect the information it 
> knows about, or is willing to pass on without authenticated access as 
> some data may be proprietary and must in that case be restricted to 
> authenticated clients only, and push that...
> 
> Architecturally what you're asking for is not possible from services 
> that have to run without human intervention.
> 
> Al.
> 
--------------------------------------------------------------------------
Arnold H. Rots                                Chandra X-ray Science Center
Smithsonian Astrophysical Observatory                tel:  +1 617 496 7701
60 Garden Street, MS 67                              fax:  +1 617 495 7356
Cambridge, MA 02138                             arots at head.cfa.harvard.edu
USA                                     http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------