Clarification Needed on Handling Duplicate `anchor` Definitions Across JSON Schema Tests

[kaptinlin]

Description: There seems to be a discrepancy in how duplicate $anchor definitions are handled across different tests in the JSON Schema Test Suite. Specifically, the tests present conflicting interpretations of whether a subsequent $anchor with the same name should override an earlier definition or whether the first occurrence should be retained.

Relevant Test Files and Observations:

1. Ref Test (ref.json#L632) - In this test, when $anchor is defined multiple times, the reference points to the first occurrence of the $anchor.

   • ref.json#L632

2. Anchor Test (anchor.json#L84) - This test appears to indicate that the latest $anchor definition overrides earlier ones.

   • anchor.json#L84

Concern: The handling of duplicate $anchor names varies between tests, which may lead to inconsistent behavior in schema processing across different implementations. This inconsistency could impact the predictability and reliability of JSON Schema validation.

Questions and Suggested Actions:

• Clarification Request: Could the JSON Schema specification provide clearer guidance on how duplicate $anchor names should be handled? Specifically, should a later $anchor override an earlier one, or should the first defined $anchor be the one that is referenced?
• Specification Update Proposal: I propose considering an update to the JSON Schema specification to explicitly address this scenario, ensuring that all implementations handle duplicate $anchor names consistently.
• Test Suite Adjustment: Depending on the clarification, adjustments may be necessary in the JSON Schema Test Suite to align with the clarified behavior, either by modifying existing tests or by adding new tests to explicitly verify the expected handling of duplicate $anchors.

Thank you for considering this issue. Addressing this ambiguity will help in enhancing the robustness and interoperability of JSON Schema implementations and tools.

[gregsdennis]

In both of the tests you reference, one of the $anchor keywords has a different base URI, provided by an adjacent $id keyword.

These aren't duplicate anchors, just same-named anchors in different resources.

[gregsdennis]

Truly duplicate $anchors should ("MAY") be an error.

At the end of Core 8.2.2:

The effect of specifying the same fragment name multiple times within the same resource, using any combination of "$anchor" and/or "$dynamicAnchor", is undefined. Implementations MAY raise an error if such usage is detected.

[kaptinlin]

thanks