Define when an IRI is considered to be absolute and if IRI normalization is supported

[lanthaler]

We still have to define how we detect whether an IRI is an absolute IRI or not (this should be easy as we just have to look for a colon and an undefined prefix). We also have to define whether we allow stuff like "../" and "./" in relative IRIs and suffixes. Doing so would require IRI normalization and would go beyond pure textual concatenation.

[gkellogg]

This is redundant now, as the spec only allows absolute IRIs.

[msporny]

The spec supports relative IRIs, absolute IRIs, and prefixes:

http://json-ld.org/spec/ED/json-ld-syntax/20120122/#iris

[gkellogg]

Definitions of absolute and relative IRIs added to commit ae24034.

[lanthaler]

This defines what an absolute IRI is but not necessarily how we recognize one. Should we have something like:

if no term and prefix matches but the following regex (TBD) matches, the value is considered to be an absolute IRI?

[gkellogg]

The problem is, it is very difficult to distinguish an IRI from a compact IRI (CURIE). As mentioned on the call, one way is to limit the value space of the suffix portion to exclude values that begin with '//', but this is not fool-proof, as there are IRI schemes that do not need to begin with '//', such as 'mailto'. The proposed updated CURIE grammar for RDFa is the following:

 curie       ::= [ [ prefix ] ':' ] reference
 reference   ::= ( ipath-absolute / ipath-rootless / ipath-empty )
                     [ "?" iquery ] [ "#" ifragment ]

We could consider the reference portion for the definition of suffix in JSON-LD, but I think we agreed on today's telecon that this was premature.

Practically speaking, in JSON-LD, a CURIE is recognized when the prefix is defined as a term in the active context, and otherwise, the value is treated as an absolute IRI. I propose that we add this definition and resolve the issue.

[lanthaler]

In other words you would propose that if a term contains a colon but the prefix isn't defined it is treated as an absolute IRI, right?

[gkellogg]

Yes, I believe that's what I just said.

[lanthaler]

OK.. that would basically that all terms with colons would result in an IRI but probably there's no other technical solution without relying on a list of allowed schemas.

[msporny]

PROPOSAL: If a key in a JSON-LD document contains a colon, it is a CompactIRI if the prefix is defined as a term in the active context, otherwise it is an AbsoluteIRI.

[lanthaler]

+1

If we are going to define it that way we need to add a note in the spec saying that keys with colons are ALWAYS mapped to an IRI - there's no way around this. The only thing we could do is to specify an escape character (which could also work with the @ keywords)... but I don't know if we really need to go that far.

[msporny]

RESOLVED: If a key in a JSON-LD document contains a colon and the first colon is not followed by "//", it is a CompactIRI if the prefix is defined as a term in the active context, otherwise it is an AbsoluteIRI.