Schema

ApertureDB derives its schema from stored objects and their properties. Adding an object does not require a schema declaration; the schema evolves as data is added. GetSchema reports the resulting classes, properties, and types. Writes must still satisfy the database’s property-type rules. A conflicting write can return status: 5; see Responses.

Every node has an explicit type and class. The nine node types can share class names without sharing categories. Customer connections are categorized by their class and both endpoint type/class pairs. See class.

GetSchema returns flat nodes and edges arrays with full category selectors, counts and property types. Internal ownership links are excluded. Its scope is node, edge or all. A ref inspects an earlier selection; refresh: false permits the transaction’s cached schema.

Properties are objects such as {"count":123,"type":"String"}. Index definitions are available separately through GetIndexes. Connection aggregates maintain ordinary properties on source objects. Inspect their definitions with GetAggregates. Descriptor classes use vector indexes on their binary column; see Descriptor classes and vector indexes.

RemoveClass requires an empty category with no indexes, aggregate definitions, or dependent customer connection categories. Its committed removal makes the numeric category ID available for reuse, including by another node type. New object birth transactions prevent old _uniqueid values from selecting replacement objects. Existing readers keep their original category snapshot. DeleteAll preserves active category registrations even when they become empty; remove those categories explicitly when they are no longer needed.