VOEvent session

Rob Seaman seaman at noao.edu
Mon Dec 13 19:01:39 PST 2010


Can a "lightweight voevent packet followed by an arbitrary payload" correspond to "publishing a telegram"?

Rob
--
On Dec 13, 2010, at 5:58 PM, Mike Fitzpatrick wrote:

> So this added complexity appears to be for a single use case, the desire to
> pass light curves in the packet.  What if we have other use cases where I want
> to include a full spectrum so my machine can classify a SN, or a table of
> orbital elements so I can predict a future observation, or the light curve of 
> a moving source?   Where does it end?  And do we do it all with an IVOA
> model of something or a custom simple version?
> 
> If the requirement is that a machine be able to make decisions based on what's
> in the packet, the buy-in to support (from my perspective, parse/create) a packet 
> is pretty huge.  There also seems to be a lot more processing required to sort
> a stream of many thousands of messages to decide if a particular packet is
> of interest (or is supported).
> 
> OTOH, is the transport of events were designed to act like a messaging system
> then we could again have a lightweight voevent packet followed by an arbitrary
> payload of whatever was desired.  Likewise, the "backbone" could be designed
> to make it easy to satisfy requests for reference material (e.g. a SEAP call back 
> to the originating repository).  In the first case, we may be transmitting a lot more
> data, but we're doing that anyway if it's already in the packet.
> 
> -Mike
> 
> 
> 
> On Mon, Dec 13, 2010 at 5:29 PM, Roy Williams <roy at cacr.caltech.edu> wrote:
> On 12/13/2010 3:13 PM, Rob Seaman wrote:
> > A VOEvent
> > message, for instance, could simply be an alert saying "Hey, over
> > here!  New information on object XYZ has been registered at IVORN
> > ivo:ABC".  Would this satisfy the requirements?
> 
> 
> The requirement is that a machine be able to make decisions based on the data in the VOEvent or the resources behind its links. The possibility of machines making complex judgments with multi-authored data -- this is the reason for 200 elements and attributes.
> 
> Yes, the VOEvent can be very simple and just have links. But then the question of structure is simply shifted to the remote resources.
> 
> Cheers
> Roy
> 



More information about the voevent mailing list