Trouble in getting to Solla-Sollew

Rob Seaman seaman at noao.edu
Wed Mar 16 13:23:48 PDT 2011


The engineering trade-off for VOEvent has always been between general purpose elements like params and an explicitly defined schema.  We attempted to formally define a little schema for time series and it was the broader IVOA that objected.  In fact, in the years since the first VOEvent WG meeting it appears that IVOA has decided that explicit XML schema are, well, crap.  Rather, schemata (in the broader sense) are now to be denoted by utypes referencing underlying Platonic data models.

So, from the top:

	1) time series - we are introducing a general table construct that will be able to express IVOA conforming time series when the appropriate utypes are finalized

	1b) IVOA time series - there appears to be some difficulty getting all the time series producers on the same page - in the mean time non-IVOA standards may well gather a critical market share here.  Our customers will use whatever is available when needed.

	2) orbital elements - one can express pretty much anything with a list of params in the <What> element.  Orbital elements are targeting metadata and thus belong in <WhereWhen>.  The STC representation of orbital elements is pretty straightforward, but we are seeking to minimize wrestling with STC issues for v2.0.  (Raise your hand if you've read "I Can Lick 30 Tigers Today!")

	3) external schema - there are objections that we haven't provided explicit instructions suitable for building apps to handle all event streams.  Welcome to time domain astronomy.  The goal is to characterize little flashes of light that may come from next door in the solar system or from remote cosmological distances.  It took ten or fifteen years to figure out that GRBs were the latter.  Astronomers don't care much to study things they already understand.  This is science on the front lines and it is inevitable that the nature of the data structures changes to suit.  To return to the first point, if the IVOA really is ditching XML schemata for expressing anything but the most bland logistical metadata, than "external" is the name of the game, but "schemata" are out the door in any event.  Rather, one presumes that projects will use external references and entrained non-VOEvent data products to adapt.  VOEvent must remain open to these.  Apps will likely fall into two classes of general-purpose utilities and project- specific tools that understand only a subset of VOEventStreams.

	4) formal vocabularies - we have provided for benefitting from the excellent work done by the semantics WG.  Are there ways we need to improve these hooks further?

	5) digital signing - there was no consensus between the Denny mechanism and the Allen mechanism.  In the near term this basically means that Bob wins by default and the necessary hooks for W3C based signatures won't be put in the packets for v2.0.  Long term?

	6) enhanced references - we're discussing 'em now, I'm confident of consensus

	7) registry technologies - I look forward to pursuing this post-v2.0 along with SEAP

We have attempted to identify the minimum set of changes needed.  VOEvent v2.0 is by no means a wish list.  Please comment on changes from v1.11 in that light.  Perhaps there will be a v2.1 in two or three years (under a different chair...)

Rob
--

On Mar 16, 2011, at 12:28 PM, Roy Williams wrote:

> On 03/16/2011 12:15 PM, Matthew Graham wrote:
>> From the 2008 NVO report:
>> 
>> "[VOEvent] V2 will add support for time series, orbital elements,
>> science use case specific external schema, formal vocabularies,
>> digital signing of packets, and enhanced reference and registry
>> technologies"
> 
> I would say we got 5 out of those 7, and the registry stuff will be ready as soon as the VAO allows Matthew to build it. As for the "external schema" thing, is anyone ready to elaborate and argue for its usefulness and feasibility?
> 
> -- There is support for time series: this is represented by a <Table> element with a column that is time. The IVOA strong metadata can be added through Utypes.
> 
> -- Orbital elements, for example:
> <Group type="ivoat:keplerianElements">
>  <Param name="a"
>  <Param name="eccentricity"
> 
> -- Connecting up arbitrary external schema .... well I'm glad that threat is over!
> 
> -- Formal vocabularies are used in the <Reference type="ivoat:lightCurve> and the above example with <Group>.
> 
> -- Digital signing: my understanding is that this can be done with XML comments after the standard is defined
> http://www.ivoa.net/Documents/Notes/VOEvent/VOEventDigiSig-20080514.html
> 
> -- Reference technologies, see the <Reference> discussion, yes we got that.
> 
> -- The registry technology (as you know) is not all built yet, and the speedy REC for VOEvent2 will make that happen faster.
> 
> Roy



More information about the voevent mailing list