Question about publish/update operations.

[Terry Simons]

Does anyone know what will happen if you have two publishers for the
same metadata each with different sessions that attempt to publish
updates to a piece of metadata that already exists?

For example, let's say that a client (session 1) publishes metadata Y to link X.

Now let's say that a different client (session 2) tries to publish
metadata Y to link X.

Will this result in an error of some sort? I would think so, since
you wouldn't want two distinct clients updating data that they don't
"own", but I'm not sure.

Thanks!

- Terry

[Clifford Kahn]

No, the MAP server does not enforce any notion of ownership, other than its vendor-dependent authorization policy.

The newer metadata will replace the old if it is single-valued, and will be added to the old if it is multi-valued.

This allows for certain kinds of cooperation between MAP clients.
Cliff

[Terry Simons]

Ok thanks for that.

One more question about multi-valued metadata.

If two clients publish updates that get appended to a multi-valued
link, and a third client requests the data from the map server, does
that list of data come back in the order it was published, or does it
come back in random order?

That is, can I tell the order in which items were published?

- Terry

[Peter Lee]

Just to make it clear, there is no "multi-valued link". A link can contain
both multi-value metadata and/or single-value metadata.

As for the order of the updates, there is no specific requirement that the
list of data needs to be returned in the same order as published, the only
way that a client can distinguish the order is by looking at the "timestamp"
attribute in each of the metadata, the timestamp attribute is added by the
MAP server when the publish was received, and it will be returned with each
metadata.

Peter