nv_dfm_core.session.SessionDelegate#

class nv_dfm_core.session.SessionDelegate(session, federation_name, homesite, logger)[source]#

Abstract base class for session delegates that handle different execution targets.

A delegate implements the specific logic for executing pipelines on different backends (e.g., FLARE, local execution).

Parameters:

  • session (Any)

  • federation_name (str)

  • homesite (str)

  • logger (Logger)

abstractmethod execute(pipeline, next_frame, input_params, callback_runner, debug, options=None, force_modgen=False)[source]#

Execute a prepared pipeline.

Parameters:

  • pipeline (PreparedPipeline) – The prepared pipeline to execute.

  • next_frame (Frame) – The next frame for the pipeline.

  • input_params (list[tuple[Frame, dict[str, Any]]]) – List of (frame, params) tuples for input.

  • callback_runner (CallbackRunner) – The callback runner for receiving results.

  • debug (bool) – Whether to run in debug mode.

  • options (Any) – Target-specific options.

  • force_modgen (bool) – If True, force code regeneration.

Returns:

Job object for the execution.

Return type:

Job

load_fed_info()[source]#

Load federation information from the federation runtime module.

Return type:

FedInfo

abstractmethod reattach(job_id, callback_runner)[source]#

Reattach to an existing job.

Parameters:

  • job_id (str) – The ID of the job to reattach to.

  • callback_runner (CallbackRunner | None) – The callback runner for receiving results, or None for informational-only attachment (status/abort only).

Returns:

Job object connected to the existing job.

Return type:

Job