The State of VOEvent

[Rob Seaman]

Hi Josh,

> we've now been using for a few months an internal working scheme  
> (not yet formalized into a proper schema) for time-series (which we  
> call VOTimeseries). On the timescale of a ~week, I think we could  
> pass around instances and a first hack at an XSD and let the group  
> punch holes in it.

Yes, please.

> We could also expose to this group a warehouse of public data that  
> has already been marshalled into this format.

Ditto.

> Much of the conditions and restrictions set forth at the Tuscon  
> meeting as well as some of the structure of the Spectral schema are  
> included (as well as Drake's first work on such a format). Our  
> versions have been evolving with time as we ingest more and more  
> public timeseries datasets and start learning how others like to  
> describe their timeseries.

Ok.  Sounds like a good place to start.

> Some limitations in the current draft:
>    [1] No one in my group is an XML expert so we've likely screwed  
> up some syntax.
>        - we're not validating against a schema because there is none  
> yet
>        - we're not dealing with namespaces

Ok.  I'm sure the group can help tidy this up.

>    [2] We're focusing on optical/IR photometric data so we might be  
> missing some obvious issues with other EM & non-EM reporting.

We can try to recruit folks from outside O/IR to comment, but I would  
prefer to move forward with a useful, but limited purview, capability  
rather than to wait on all possible stakeholders to sign off on it.   
If we later discover significant missing functionality or failures of  
the conceptual model, we can enhance the format in a future version of  
the standard.

>    [3] I'm sure we're not doing STC justice.

VOEvent's use of STC is for targeting coordinates - that is, the  
independent variables of the observation (or future observations) as  
expressed in <WhereWhen>.  On the other hand, a time series is part of  
the description of the dependent variables of past observations, and  
thus should reside in <What>.  We hope to use some derivative of the  
SDM for time series, not native STC.

On the other hand, the expression of orbital elements belongs in  
<WhereWhen> and STC, because it represents targeting for future  
observations - the recovery of the solar system object in the future.

> Attached is a teaser instance. It's embedded in another internal  
> format we're calling VOSource (that's another can of worms).

One can of worms at a time.  Your VOSource also has something to do  
with the characterization data model.

We won't go far wrong if VOEvent always errs on the side of pragmatic  
utility, rather than seeking formal correctness for its own sake.   
Complexity should be introduced for a purpose.

Rob