Skip to main content

​
API & Schema

Across the 1.13 → 2.0 jump the REST surface gains 208 endpoints and loses 10 (Collate’s own surface adds 76 and removes 47 on top of that). The JSON Schema specification gains 86 files, modifies 107, and removes 3 — verified directly against the openmetadata-spec module at the 2.0.0 tag. This page covers the schema and contract changes that cut across components; component pages cover the domain-specific ones.

​
The Suggestions API is removed

Breaking. Affects API clients, SDKs, and automation bots. The entire /v1/suggestions namespace (ten endpoints, including accept-all/reject-all) is gone — see Collaboration for the full replacement mapping onto /v1/tasks.

​
search_after becomes a repeated parameter

Breaking. Affects anyone paginating /v1/search/query. Each sort value is now carried as its own search_after parameter, rather than one comma-joined string, so values containing a comma (a glossary term’s fully qualified name, for example) are safe:
Split the cursor on the client and emit one parameter per sort value. The deleted parameter on the same endpoint is deprecated for removal — express deleted-entity filtering through query_filter instead. Separately, GET /v1/dataQuality/testCases now binds testCaseStatus as a repeated parameter too, validated against the enum, with an unknown value returning 400 naming the allowed values.

​
Entity name validation is tightened

Breaking. Affects every write path, ingestion connectors and CSV import. entityName and testCaseEntityName change pattern:
Verified directly against openmetadata-spec/.../type/basic.json — confirmed present at the 2.0.0 tag and absent at 1.13.4-release. In addition to the existing :: restriction, names may no longer contain >, ", or any ASCII control character. Entities already stored with these characters are not rewritten by the migration, but the next PUT or PATCH that revalidates the name will fail — audit source systems whose object names can contain quotes or angle brackets before upgrading.

​
Enum values are removed and added across several schemas

Breaking where values are removed, additive-with-caveats where they’re added. Only Metadata remains, and the corresponding application configuration schemas are deleted entirely. Additions worth checking exhaustive switch statements or pattern matching against: resourceDescriptor.operation gains the five task operations, changeEventType gains taskCreated/taskUpdated/three lineage event types, entityRelationship.relationshipType gains assignedTo, backgroundJob.jobType gains CSV_IMPORT/CSV_EXPORT/AUDIT_EXPORT, and table.tableType gains SemanticView.

​
Default values change for RDF, workflow timing and connector filters

Behavioural. Affects deployments that never set these explicitly.
RDF inference is off by default in 2.0. If you relied on inferred triples in SPARQL queries, set RDF_INFERENCE_ENABLED=true explicitly.

​
Schema reference relocations affect generated models

Behavioural. Affects code generated from the specification and JSON Schema validators. Chart function and kpiDetails definitions moved out of dataInsightCustomChart.json into a new dataInsight/custom/chartFunctions.json — formulaHolder.json, lineChart.json, summaryCard.json and dataInsightCustomChartResultList.json all update their $ref accordingly. Regenerate any client model built from the JSON Schema; if you resolve $ref pointers by hand, update the paths.

​
Notable new endpoint groups replace common client-side patterns

Additive. None of these break anything, but they replace patterns you may currently implement yourself.

​
Two items are deprecated, not yet removed

Deprecated, plan a migration but no immediate action is required:

​
Full detail

For the full narrative walkthrough of API & Schema in Collate 2.0 — including screenshots and step-by-step context — see Release 2.0: API & Schema.