Add protected term redefinition tests.

[davidlehn]

A few of these tests look a bit crazy. Sorry. ;-) Suggested improvements would be welcome.

---

Preview | Diff

[dlongley]

@gkellogg,

This is related to #90.

[dlongley]

@davidlehn, can you resolve the conflict?

[davidlehn]

@dlongley Fixed.

[dlongley]

@davidlehn, thanks!

[gkellogg]

@dlongley note the comment in the last commit: "Note that the algorithm sets terms to null in context if they're defined as null or {"@id": null} and doesn't retain the protected status, which it probably should."

If you agree, I'll update that part too.

[dlongley]

@gkellogg,

@dlongley note the comment in the last commit: "Note that the algorithm sets terms to null in context if they're defined as null or {"@id": null} and doesn't retain the protected status, which it probably should."

If you agree, I'll update that part too.

It does seem like it should to retain protected status, i.e. what we're saying is that we'd allow for terms to be defined as null and protect that null definition, right? I know we have some other issue about it perhaps doing that for prefixes for security purposes.

We'd have to implement it to make sure it doesn't do anything strange or break any existing tests. We would also need new tests to check it. Have you happened to have implemented it and run it against the existing tests?

[gkellogg]

My implementation does create an empty term definition for either null or {"@id": null}, although it doesn't presently set @protected on it.

We should update step 5.4 in IRI Expansion to not use a term as a Compact IRI prefix if it's IRI mapping is null.