The State of VOEvent

[Rob Seaman]

Hi Bob,

> I'm an old school engineer from the days when design reviews were  
> battlegrounds, and once completed, everyone went to the pub.

I'm buying the first round in Santa Barbara.

> Easy if the XML isn't touched.

Yes, that's the issue.  Of course, there's no reason to have chosen  
XML in that case :-)

> some of VOEvent (the outer schema) [...] is not implemented/used in  
> reality.

We seek to describe a vast range of phenomena.  Virtually the whole  
schema is optional.  The alternative would be to design a least common  
denominator fixed format.  One size does not fit all.

> there are multiple violations of the VOEvent 1.1 schema in the  
> actual messages

Is the point that schema based canonicalization would thus fail?  We  
should fix violating packets.  Signing technology is one way to  
encourage validation, since nonconforming packets won't be signable -  
due to the canonicalization, if nothing else.

> Starting with that, someone [...] with thick skin should be tasked  
> with brokering the additions to the schema going forward.

I've got thick skin.  The additions to v2.0 will be:
   - Simple (but not too) time series
   - STC based orbital elements (may also clarify "why STC?")
   - A hook in <What> for external schemata
   - Expand reference types to include KML

All but KML were planned at the first VOEvent meeting.  VOEvent is  
defined by specification, not by schema.

Not v2.0 (since already supported):  vocabularies, SEAP, registries,  
signatures.

The phrase "elegant simplicity" implies that elegance is distinct from  
simplicity.  The modern world certainly contains many simple, but  
inelegant, things.  VOEvent aims for elegance.  The future will be our  
judge.  We also aim for just enough complexity to represent our  
content.  No more.

> No "it would be nice if" stuff

None exists:
   - Time series expand our reach to include the non-transient time  
domain.
   - One project's noise is another's signal - hence orbital elements.
   - External schemata permit the evolutionary path implicit in your  
message.
   - Controlled vocabularies have been copiously justified previously.
   - SEAP is query.  Query + registry = command and control.
   - Authentication?  No point in any of it without a reliable trust  
model.

> A digital signature, using today's methods, on some bucket of bytes  
> less than a few megabytes should be so reliable that the probability  
> that it can be spoofed is less than the inverse of the number of  
> particles in the universe.

For logistical headaches, see Bruce Schneier.  The canonicalization  
discussion is just a detour on the way to key exchange.

> What's so hard about keeping the original XML?

XML is about content.  FITS, for instance, is about the octets.

> a "relay" shouldn't be interested in the content

XML is content, so the brokers respond to content.  Relays are filters.

Rob