enveloping, batching, signing

[Rob Seaman]

On Feb 4, 2008, at 12:29 PM, Steve Allen wrote:

> On Mon 2008-02-04T12:12:50 -0700, Rob Seaman hath writ:
>> We have to support both.  Can the envelope also be made lightweight
>> for the single packet case?
>
> Quite literally the hardest part of making the envelope is
> agreeing upon what to name the envelope document type.

type = "hancock"

> Other than that it need merely allow one or more VOEvent
> elements and zero or more Signature elements.

Make that zero or more and zero or more.  We might want an envelope  
(call it a "first day cover") that contains only signatures.

> This doesn't answer the question of whether it makes sense to
> allow Signature inside the VOEvent.

The peanuts (from the gallery) should speak up at this juncture.

> *May* travel separately and *should* are very different things, for
> separate travel places extra requirements on the handling systems in
> order to maintain and allow access to their association, and to
> distinguish followups which are events from followups which are
> signatures.

Yup.

> These are some of the serious questions of engineering for which
> we will wish to agree on answers:
>
> Is a Signature important enough warrant inclusion in every event?

Nope.

> Is a Signature cheap enough (creation, transmission, validation, or
> even just ignoring) to allow inclusion in every event?

Yes.  But the load for the subscriber can be minimized by ignoring the  
signature.  A publisher that can't afford the overhead might want to  
reconsider participating in the transient community.

> Are there any use cases which want enveloped batches of events?

Probably.  Are there other use cases than signing that would benefit  
from envelopes?

- Rob