Annotation Collections: different IRIs for embedded IRIs and embedded descriptions.

[gsergiu]

Within the section Representations with Annotation IRIs the same annotation collection ( in the sense that it contains the same annotations objects), gets 2 different IRIs.

"id": "http://example.org/annotations/?iris=1"
"id": "http://example.org/annotations/?iris=0"

Consequently, a client will consider that these 2 responses contain 2 different collections. However, the fact is that the client retrieves 2 different representations of the same collection.

Is there a posibility to move the iris param from the url? Can we consider using different profiles and/or headers so that the ids of collections are the same, independently from the representation of annotations?

This seems to be conceptually related to #371

[iherman]

On 14 Dec 2016, at 16:30, gsergiu [email protected] wrote:

Within the section Representations with Annotation IRIs https://www.w3.org/TR/annotation-protocol/#representations-with-annotation-iris the same annotation collection ( in the sense that it contains the same annotations objects), gets 2 different IRIs.

"id": "http://example.org/annotations/?iris=1" "id": "http://example.org/annotations/?iris=0" Consequently, a client will consider that these 2 responses contain 2 different collections. However, the fact is that the client retrieves 2 different representations of the same collection.

Is there a posibility to move the iris param from the url? Can we consider using different profiles and/or headers so that the ids of collections are the same, independently from the representation of annotations?

I do not think the standard specifies any of that, those are only examples for the URI-s that a specific implementation may use.

This seems to be conceptually related to #371 https://github.com/w3c/web-annotation/issues/371

And just as that one, I would like to label this issue as 'postponed'. At this point, when we plan to vote for a Proposed Rec this week, we cannot do any technical change on the documents any more.

[gsergiu]

@iherman it is ok for me to set it to postponed. I just wanted to report it as an issue. When using the get method, you typically used query params in the URLs, which has the consequence that we get different URLs that actually point to the same conceptual thing. The basic question here is: can we consider this URLs as being IRIs? under which conditions/circumstances?

[gsergiu]

Just a note that this issue is partially addressed by the introduction of container prefferences, in the sense that the IRIs or embedded annotations can be asled through the Prefer header. https://www.w3.org/TR/annotation-protocol/#container-representation-preferences

In this case the iris query param is redundant, but still has the effect of having two different IRIs for the same colection.