Links to native format

[PascalLike]

During the Rotterdam Code Sprint in October 2025 we realized there's no way to flag which link is the "native" format of an item. For example, in OGC API Records, clients might want to know the original format of a record among the various "rel"="alternate" representations listed. Common's link schema includes a rel field that could be used to indicate which one is the "native" item representation. For example, a link might have "rel": "native" to mark that it's the original encoding. This small tweak would let clients see which link is the source format, which would really help the Records API (and maybe Features) be clearer about native encodings.

Since the IANA Link Relations registry does not include a "native" relation, we suggest defining a specific OGC relation, for example:

http://www.opengis.net/def/rel/ogc/1.0/native

"Refers to a representation that constitutes the native format in which the resource is stored or maintained by the server." (see comment)

[jerstlouis]

For a general web resources, wouldn't this be canonical ?

Designates the preferred version of a resource (the IRI and its contents).

I believe I have seen OGC API implementations using it that way.

In the 3D Container and Tiles pilot (Engineering Report), we had used original for a similar purpose. original is defined as an IANA entry, however, the meaning is somewhat different referring to something more like the "original version" rather than the source representation.

For the definition of entries in the definition server, this is normally handled by the Naming Authority. However, normally these are only defined once they are used in some standard.

Potentially canonical could be used from an OGC API - Common - Part 2 collection to link to the preferred representation of the data.

@cportele any suggestions how to go about this?

[cportele]

@jerstlouis

See my comment in the original thread that was also referenced by. @PascalLike. As you point out, original is not applicable in this case.

canonical works for cases where the link context is a "non-canonical" / "non-native" representation of the canonical / native representation of the resource.

In practice, canonical seems to be mostly used to declare duplicate pages/resources to search engines, e.g. see here.

Before deciding to use canonical, it might be worth a try how this impacts how search engines index such cases, in particular, if there are HTML representations.

canonical does not work in cases where one resource points to representations of another resource and wants to recommend one of the representations.