Skip to main content
Version: v3.0.0

Asynchronous methods

Several methods in the Feature Store Client API have asynchronous variants (methods ending with _async).

For example, starting an ingestion asynchronously:

job = feature_set.ingest_async(source)

This method returns a job immediately instead of blocking. The job exposes:

  • done — a property that is True once the job has finished, False otherwise.
  • get_result() — a method that returns the result of the job. If it is called before the job has finished, an exception is thrown.
  • wait_for_result(poll_interval=2.0, timeout=None) — a method that blocks until the job completes and then returns its result. This is the most convenient way to wait for an asynchronous job.
if job.done:
    result = job.get_result()

# Or simply block until the job finishes and return the result:
result = job.wait_for_result()

Asynchronous schema extraction​

Schema extraction runs as a job, so it has asynchronous variants too. Call them on the workspace you intend to register the feature set in — see Schema extraction in the Workspaces API for why the scope matters.

workspace = client.workspaces.list(name="my_workspace")[0]

# Extract a schema from a data source
job = workspace.extract_schema_from_source_async(source)
schema = job.wait_for_result()

# Extract a derived schema from existing feature sets
job = workspace.extract_derived_schema_async(
    feature_sets=[parent_feature_set],
    transformation=transformation,
)
schema = job.wait_for_result()

Both methods return a Job immediately, and wait_for_result() returns the extracted Schema.

note

The same methods also exist on the client (client.extract_schema_from_source_async(...)), but they scope the job to your default workspace. Prefer the workspace-scoped form.

Feedback