# UID Types in Curiosity

Curiosity uses several types of Unique Identifiers (UIDs) to manage data efficiently. Understanding these types is crucial for data modeling and low-level graph operations.

# Node UIDs (UID128)

Nodes are identified by a 128-bit UID (UID128).

• Stable: These UIDs are stable across all user-defined schemas. If you recreate a node with the same Type and Key, it will have the same UID.
• Deterministic: The UID is deterministically derived from the node's Type (string) and Key (string) using a hashing algorithm.
• Usage: This is equivalent to a Primary Key in a traditional database.
• Size: 16 bytes.

# Node Type and Edge Type UIDs

Internally, Curiosity maps string type names (e.g., "Person", "KNOWS") to numeric UIDs for performance.

• Sequential: These are always sequential integers (uint).
• Schema Association: They are strictly associated with the current schema.
• Not Stable: These UIDs are not stable across deployments. If you export data and import it into a fresh workspace, the numeric Type UID for "Person" might change (e.g., from 10 to 15). Always refer to types by their string name in code/queries, and let the system handle the mapping.
• Size: 4 bytes.

# Index UIDs (UID64)

Indexes and some internal structures use a 64-bit UID (UID64).

• Usage: Identifies specific indexes (e.g., a specific Lucene Text Index instance).
• Size: 8 bytes.

# Edge Definition

An Edge in Curiosity is a lightweight structure connecting two nodes. Its definition includes:

1. Edge UID: A derived UID identifying the edge itself (often a combination of the source node, edge type, and an internal counter).
2. Target Node UID: The UID128 of the node the edge points to.
3. Target Node Type UID: The numeric Type UID of the target node. This allows the engine to know the type of the target without loading the target node itself.