Functions Overview¶

Source: https://www.palantir.com/docs/foundry/functions/overview/ Captured during the ontology-parity effort. Concrete feature taxonomy only.

Definition¶

Functions are "code-authored logic that can be executed quickly in operational contexts," with first-class support for authoring logic based on the Ontology.

Languages & runtimes¶

TypeScript: v1 (legacy) and v2 (current).
Python: full support with a dedicated toolchain.
Feature support varies by language/version; language choice is feature-driven.

Function types by execution context¶

Data transformation — return object sets / variable values for dashboards; compute derived columns in Workshop; chart aggregations.
Ontology operations — read object properties, traverse links, execute flexible Ontology edits across multiple objects, back function-backed actions for complex mutations.
Integration — query external systems via webhooks; enrich objects with external data; API calls from the function body.
Domain-specific — Pipeline Builder sidecars (Python); query functions published via an API gateway; custom aggregations on object types; functions-on-objects (relationship-aware); LLM integrations (TS v2 + Python).

Ontology binding mechanisms¶

Object access: direct property access on typed objects; link traversal; object-set queries/filtering; search + aggregation.
Edit capabilities: user-facing mutations through actions; staged writes (TS v2); schema-aware object creation with ID generation; attachment/media handling.

Lifecycle & deployment¶

Development: unit-testing framework with stub objects; local dev (Python); debug modes + logging.
Publishing/versioning: versioned releases with dependency ranges; Marketplace product integration; permission-based execution control.
Monitoring: function metrics/perf tracking; execution telemetry; action logs for edit operations.

Server-side execution model¶

Functions execute server-side in an isolated environment, enabling operational automation without client-side processing overhead.