make format validate by default

[gregsdennis]

What kind of change does this PR introduce?

Feature update

Issue & Discussion References

Closes #1520
Depends on json-schema-org/TSC#19

Summary

Makes format an assertion to align with user expectations.

Quick notes:

• Implementations SHOULD support format with the defined values
• Implementations MAY support others, but only by explicit config
• Implementations MUST refuse to process a schema that contains an unsupported format
• Permissive allowances removed since every format is clearly defined by referencing a specification

Does this PR introduce a breaking change?

Technically a breaking change, but aligns to user expectations. So... ????

[jviotti]

If I'm reading this correctly, it means we let format apply to instances other than strings. If I recall correctly, the JSON Schema test suite very explicitly tests that format only applies to instances, which might be a change that could confuse various people and implementations.

[gregsdennis]

If I'm reading this correctly, it means we let format apply to instances other than strings.

Yes, format can apply to instances of any type. This isn't a new feature, however, the spec only defines formats that apply to strings.

[jdesrosiers]

This being a breaking change and a potentially controversial one, I think there should be a TSC vote to formally document that we have consensus before this moves forward. It should also have an ADR to document the decision.

[gregsdennis]

It should also have an ADR to document the decision.

I'll add one.

I think there should be a TSC vote to formally document that we have consensus before this moves forward

I'll create a TSC voting issue summarizing this PR.

[jdesrosiers]

I have a few editorial suggestions, but otherwise, this looks great.

[Relequestual]

I was initially concerned with some of the wording and the implication for required or optional tests, but it was changed, and I'm now happy. Sharing after the merge as @gregsdennis asked on the TSC vote issue for me to do so, although that was a week ago (sorry). (Partly for my own records

Specifically https://github.com/json-schema-org/json-schema-spec/pull/1553/files#diff-a2feaccf762dd55c4d3dd625aa1b57c48df6f782b8088cd4210fc9b93f8aea17R314-R316

Implementations SHOULD provide assertion behavior for the format values defined
by this document[^2] and MUST refuse to process any schema which contains a
format value it doesn't support.

I was concerned the "SHOULD" was too weak here, but the proceeding "MUST" gives it enough balance. Primarily, this avoids any false positives where a user expects assertion behavior and an implementation goes "shrugs I'll just allow it".