[jviotti]

Signed-off-by: Juan Cruz Viotti jv@jviotti.com

[augmentcode]

This pull request is abnormally large and would use a significant amount of tokens to review. If you still wish to review it, comment "augment review" and we will review it.

[jviotti]

augment review

[augmentcode]

🤖 Augment PR Summary

Summary: This PR adds support for OpenAPI 3.0 Schema Objects. These schemas cannot declare their dialect with $schema or carry an identifier, so a collection's defaultDialect supplies the dialect instead.

Changes:

• Resolver now removes $schema from the document and adds $schema/$id back only when the dialect's vocabularies define those keywords (dialect_defines). The dialect is now read from the resolver entry instead of the document.
• Adds Resolver::dialect_of. Index generators (materialise, bundle, editor, locations, dependencies, health, blaze templates, stats, explorer metadata/dependents) now pass the entry's dialect and identifier to SchemaFrame, schema_bundle, for_editor, lint, and similar functions.
• Adds RouterAction::artifact_dialect, which reads the dialect from the metapack extension. Enterprise conversion checks now use it through served_schema_dialect (ladder rung plus a "custom" flag) instead of reading $schema from the document.
• Adds the OAS 3.0 base vocabulary to the stats response schema, and updates docs to mention OpenAPI 3.0 and how to configure it.
• Adds new openapi3.0 fixtures, e2e hurl suites, and CLI snapshot tests for both the 2021-09-28 and 2024-10-18 dialect URIs. Updates the large set of existing hurl expectations for the new collection (counts, ordering, pagination cursors).

Technical Notes: Because OAS 3.0 artifacts no longer contain $schema or $id, anything that consumes them must get the dialect from outside the document, either from the resolver entry or the metapack extension.

_{🤖 Was this summary useful? React with 👍 or 👎}

[augmentcode]

Review completed. 1 suggestion posted.

Comment augment review to trigger a new review at any time.

[augmentcode]

Since an OAS 3.0 schema is now returned with no $schema, it can only be read correctly by callers that pass its dialect in from outside. A schema in a different dialect that $refs it (e.g. modern.json (2020-12) referencing identifier, a playground schema, or a conversion closure) will frame, compile and bundle the target under the referrer's dialect, so nullable/boolean exclusiveMinimum get the wrong meaning or fail metaschema checks. Consider how cross-dialect references to these schemas should keep their dialect, for example when bundling or when the subresolver is called, or explicitly reject them.

Severity: medium

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}

[github-actions]

Benchmark (community)

Details

Benchmark suite	Current: a87f797	Previous: 12328ab	Ratio
Add one schema (0 existing)	149 ms	254 ms	0.59
Add one schema (100 existing)	32 ms	37 ms	0.86
Add one schema (1000 existing)	72 ms	75 ms	0.96
Add one schema (10000 existing)	661 ms	535 ms	1.24
Update one schema (1 existing)	22 ms	32 ms	0.69
Update one schema (101 existing)	32 ms	37 ms	0.86
Update one schema (1001 existing)	81 ms	80 ms	1.01
Update one schema (10001 existing)	646 ms	543 ms	1.19
Cached rebuild (1 existing)	9 ms	9 ms	1
Cached rebuild (101 existing)	12 ms	10 ms	1.20
Cached rebuild (1001 existing)	35 ms	35 ms	1
Cached rebuild (10001 existing)	232 ms	184 ms	1.26
Index 100 schemas	345 ms	316 ms	1.09
Index 1000 schemas	819 ms	1067 ms	0.77
Index 10000 schemas	9068 ms	10081 ms	0.90
Index 10000 schemas (custom meta-schema)	10768 ms	13424 ms	0.80
Index 10000 schemas ($ref fan-out)	10240 ms	12314 ms	0.83
test/e2e/html: Schema Fetch (p50)	203 us	295 us	0.69
test/e2e/html: Schema Fetch (p99)	243 us	365 us	0.67

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

Benchmark (enterprise)

Details

Benchmark suite	Current: a87f797	Previous: 12328ab	Ratio
Add one schema (0 existing)	540 ms	604 ms	0.89
Add one schema (100 existing)	125 ms	123 ms	1.02
Add one schema (1000 existing)	178 ms	171 ms	1.04
Add one schema (10000 existing)	793 ms	779 ms	1.02
Update one schema (1 existing)	108 ms	110 ms	0.98
Update one schema (101 existing)	121 ms	122 ms	0.99
Update one schema (1001 existing)	183 ms	179 ms	1.02
Update one schema (10001 existing)	843 ms	821 ms	1.03
Cached rebuild (1 existing)	13 ms	14 ms	0.93
Cached rebuild (101 existing)	17 ms	17 ms	1
Cached rebuild (1001 existing)	52 ms	43 ms	1.21
Cached rebuild (10001 existing)	318 ms	302 ms	1.05
Index 100 schemas	757 ms	718 ms	1.05
Index 1000 schemas	1842 ms	1876 ms	0.98
Index 10000 schemas	16273 ms	16208 ms	1.00
Index 10000 schemas (custom meta-schema)	16309 ms	16106 ms	1.01
Index 10000 schemas ($ref fan-out)	20918 ms	20989 ms	1.00
enterprise/e2e/auth: Schema Anonymous (p50)	391 us	398 us	0.98
enterprise/e2e/auth: Schema Anonymous (p99)	507 us	506 us	1.00
enterprise/e2e/auth: Schema API Key Identity (p50)	395 us	402 us	0.98
enterprise/e2e/auth: Schema API Key Identity (p99)	510 us	514 us	0.99
enterprise/e2e/auth: Schema API Key SHA256 (p50)	400 us	408 us	0.98
enterprise/e2e/auth: Schema API Key SHA256 (p99)	525 us	521 us	1.01
enterprise/e2e/auth: Schema JWT (p50)	524 us	533 us	0.98
enterprise/e2e/auth: Schema JWT (p99)	675 us	684 us	0.99
test/e2e/html: Schema Fetch (p50)	395 us	403 us	0.98
test/e2e/html: Schema Fetch (p99)	490 us	505 us	0.97

This comment was automatically generated by workflow using github-action-benchmark.