Per-Entity Metadata Working Group

[Tom Scavo <>]

On Wed, Jul 27, 2016 at 11:59 AM, Cantor, Scott 
<>
 wrote:
> On 7/27/16, 11:57 AM, "Walter Forbes Hoehn (wassa)" 
> <>
>  wrote:
>
>> I just assumed it was a way to roll “current practice” into the protocol. 
>> At a practical
>> level, I can’t imagine anyone computing the entire aggregate on a 
>> per-query basis.
>
> No, but that's kind of the point. Making it a requirement ties it to 
> current practice, and there's also a slightly subtle issue: eventually the 
> aggregate will grow to a size that makes signing it really painful for the 
> same reason that we're engaged in this whole exercise.

+1

I don't expect Ian to run off and change the spec, however ;-) As
Scott and others continue to observe, the discovery issue is still
outstanding (although I don't necessarily assume a full aggregate is
THE solution to discovery).

I think our initial deployment of a MDQ server should not expose an
"all entities" aggregate. I don't want deployers to migrate to the MDQ
server for a full aggregate, I want deployers to migrate for
per-entity metadata. Moreover, if we created an "all entities"
aggregate on the MDQ server AND we introduce a new metadata signing
key (like the UK federation is planning to do), then we would be
forced to introduce a new aggregate up front (because the spec says
so). Recall that Rhys would like to deliberately introduce a new
aggregate (if at all). I totally agree with that sentiment. The spec
shouldn't force a migration.

Tom