make memoizer into an interface class and impls

Author: benclifford

parsl/dataflow/dflow.py

@@ -29,7 +29,7 @@
 from parsl.dataflow.dependency_resolvers import SHALLOW_DEPENDENCY_RESOLVER
 from parsl.dataflow.errors import DependencyError, JoinError
 from parsl.dataflow.futures import AppFuture
-from parsl.dataflow.memoization import Memoizer
+from parsl.dataflow.memoization import BasicMemoizer, Memoizer
 from parsl.dataflow.rundirs import make_rundir
 from parsl.dataflow.states import FINAL_FAILURE_STATES, FINAL_STATES, States
 from parsl.dataflow.taskrecord import TaskRecord
@@ -164,7 +164,7 @@ def __init__(self, config: Config) -> None:
         else:
             checkpoint_files = []
 
-        self.memoizer = Memoizer(self, memoize=config.app_cache, checkpoint_files=checkpoint_files)
+        self.memoizer: Memoizer = BasicMemoizer(self, memoize=config.app_cache, checkpoint_files=checkpoint_files)
         self.memoizer.run_dir = self.run_dir
 
         self._checkpoint_timer = None

parsl/dataflow/memoization.py

@@ -121,6 +121,17 @@ def id_for_memo_function(f: types.FunctionType, output_ref: bool = False) -> byt
 
 
 class Memoizer:
+    def update_memo(self, task: TaskRecord, r: Future[Any]) -> None:
+        raise NotImplementedError
+
+    def checkpoint(self, tasks: Sequence[TaskRecord]) -> None:
+        raise NotImplementedError
+
+    def check_memo(self, task: TaskRecord) -> Optional[Future[Any]]:
+        raise NotImplementedError
+
+
+class BasicMemoizer(Memoizer):
     """Memoizer is responsible for ensuring that identical work is not repeated.
 
     When a task is repeated, i.e., the same function is called with the same exact arguments, the