Source code for qcfractal.queue.adapters

Queue backend abstraction manager.

from .executor_adapter import DaskAdapter, ExecutorAdapter
from .fireworks_adapter import FireworksAdapter
from .parsl_adapter import ParslAdapter

[docs]def build_queue_adapter(workflow_client, logger=None, **kwargs) -> "BaseAdapter": """Constructs a queue manager based off the incoming queue socket type. Parameters ---------- workflow_client : object A object wrapper for different distributed workflow types. The following input types are valid - Python Processes: "concurrent.futures.process.ProcessPoolExecutor" - Dask Distributed: "distributed.Client" - Fireworks: "fireworks.LaunchPad" - Parsl: "parsl.config.Config" logger : logging.Logger, Optional. Default: None Logger to report to **kwargs Additional kwargs for the Adapter Returns ------- ret : Adapter Returns a valid Adapter for the selected computational queue """ adapter_type = type(workflow_client).__module__ + "." + type(workflow_client).__name__ if adapter_type == "parsl.config.Config": adapter = ParslAdapter(workflow_client, logger=logger, **kwargs) elif adapter_type == "concurrent.futures.process.ProcessPoolExecutor": adapter = ExecutorAdapter(workflow_client, logger=logger, **kwargs) elif adapter_type == "distributed.client.Client": adapter = DaskAdapter(workflow_client, logger=logger, **kwargs) elif adapter_type == "fireworks.core.launchpad.LaunchPad": adapter = FireworksAdapter(workflow_client, logger=logger, **kwargs) else: raise KeyError("QueueAdapter type '{}' not understood".format(adapter_type)) return adapter