flwr 1.15.1__tar.gz → 1.16.0__tar.gz

{flwr-1.15.1 → flwr-1.16.0}/PKG-INFO

@@ -1,13 +1,13 @@
 Metadata-Version: 2.1
 Name: flwr
-Version: 1.15.1
+Version: 1.16.0
 Summary: Flower: A Friendly Federated AI Framework
 Home-page: https://flower.ai
 License: Apache-2.0
 Keywords: Artificial Intelligence,Federated AI,Federated Analytics,Federated Evaluation,Federated Learning,Flower,Machine Learning
 Author: The Flower Authors
 Author-email: hello@flower.ai
-Requires-Python: >=3.9,<4.0
+Requires-Python: >=3.9.2,<4.0.0
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research
@@ -16,12 +16,12 @@ Classifier: Operating System :: MacOS :: MacOS X
 Classifier: Operating System :: POSIX :: Linux
 Classifier: Programming Language :: Python
 Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3 :: Only
 Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: Implementation :: CPython
 Classifier: Topic :: Scientific/Engineering
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
@@ -32,7 +32,7 @@ Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Classifier: Typing :: Typed
 Provides-Extra: rest
 Provides-Extra: simulation
-Requires-Dist: cryptography (>=43.0.1,<44.0.0)
+Requires-Dist: cryptography (>=44.0.1,<45.0.0)
 Requires-Dist: grpcio (>=1.62.3,<2.0.0,!=1.65.0)
 Requires-Dist: iterators (>=0.0.2,<0.0.3)
 Requires-Dist: numpy (>=1.26.0,<3.0.0)
@@ -40,7 +40,7 @@ Requires-Dist: pathspec (>=0.12.1,<0.13.0)
 Requires-Dist: protobuf (>=4.21.6,<5.0.0)
 Requires-Dist: pycryptodome (>=3.18.0,<4.0.0)
 Requires-Dist: pyyaml (>=6.0.2,<7.0.0)
-Requires-Dist: ray (==2.10.0) ; (python_version >= "3.9" and python_version < "3.12") and (extra == "simulation")
+Requires-Dist: ray (==2.31.0) ; (python_version >= "3.9" and python_version < "3.13") and (extra == "simulation")
 Requires-Dist: requests (>=2.31.0,<3.0.0)
 Requires-Dist: rich (>=13.5.0,<14.0.0)
 Requires-Dist: starlette (>=0.45.2,<0.46.0) ; extra == "rest"

{flwr-1.15.1 → flwr-1.16.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "flwr"
-version = "1.15.1"
+version = "1.16.0"
 description = "Flower: A Friendly Federated AI Framework"
 license = "Apache-2.0"
 authors = ["The Flower Authors <hello@flower.ai>"]
@@ -51,8 +51,6 @@ exclude = ["src/py/**/*_test.py"]
 [tool.poetry.scripts]
 # `flwr` CLI
 flwr = "flwr.cli.app:app"
-# SuperExec (can run with either Deployment Engine or Simulation Engine)
-flower-superexec = "flwr.superexec.app:run_superexec" # Deprecated
 # Simulation Engine
 flwr-simulation = "flwr.simulation.app:flwr_simulation"
 flower-simulation = "flwr.simulation.run_simulation:run_simulation_from_cli"
@@ -60,17 +58,15 @@ flower-simulation = "flwr.simulation.run_simulation:run_simulation_from_cli"
 flower-superlink = "flwr.server.app:run_superlink"
 flower-supernode = "flwr.client.supernode.app:run_supernode"
 flwr-serverapp = "flwr.server.serverapp:flwr_serverapp"
-flower-server-app = "flwr.server.run_serverapp:run_server_app" # Deprecated
 flwr-clientapp = "flwr.client.clientapp:flwr_clientapp"
-flower-client-app = "flwr.client.supernode:run_client_app" # Deprecated
 
 [tool.poetry.dependencies]
-python = "^3.9"
+python = "^3.9.2"
 # Mandatory dependencies
 numpy = ">=1.26.0,<3.0.0"
 grpcio = "^1.62.3,!=1.65.0"
 protobuf = "^4.21.6"
-cryptography = "^43.0.1"
+cryptography = "^44.0.1"
 pycryptodome = "^3.18.0"
 iterators = "^0.0.2"
 typer = "^0.12.5"
@@ -81,7 +77,7 @@ rich = "^13.5.0"
 pyyaml = "^6.0.2"
 requests = "^2.31.0"
 # Optional dependencies (Simulation Engine)
-ray = { version = "==2.10.0", optional = true, python = ">=3.9,<3.12" }
+ray = { version = "==2.31.0", optional = true, python = ">=3.9,<3.13" }
 # Optional dependencies (REST transport layer)
 starlette = { version = "^0.45.2", optional = true }
 uvicorn = { version = "^0.34.0", extras = ["standard"], optional = true }

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/build.py

@@ -138,6 +138,8 @@ def build(
                 and f.name != "pyproject.toml"  # Exclude the original pyproject.toml
             ]
 
+            all_files.sort()
+
             for file_path in all_files:
                 # Read the file content manually
                 with open(file_path, "rb") as f:

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/log.py

@@ -38,6 +38,10 @@ from flwr.proto.exec_pb2_grpc import ExecStub
 from .utils import init_channel, try_obtain_cli_auth_plugin, unauthenticated_exc_handler
 
 
+class AllLogsRetrieved(BaseException):
+    """Raised when all logs are retrieved."""
+
+
 def start_stream(
     run_id: int, channel: grpc.Channel, refresh_period: int = CONN_REFRESH_PERIOD
 ) -> None:
@@ -56,10 +60,10 @@ def start_stream(
         # pylint: disable=E1101
         if e.code() == grpc.StatusCode.NOT_FOUND:
             logger(ERROR, "Invalid run_id `%s`, exiting", run_id)
-        if e.code() == grpc.StatusCode.CANCELLED:
-            pass
         else:
             raise e
+    except AllLogsRetrieved:
+        pass
     finally:
         channel.close()
 
@@ -94,6 +98,7 @@ def stream_logs(
         with unauthenticated_exc_handler():
             for res in stub.StreamLogs(req, timeout=duration):
                 print(res.log_output, end="")
+        raise AllLogsRetrieved()
     except grpc.RpcError as e:
         # pylint: disable=E1101
         if e.code() != grpc.StatusCode.DEADLINE_EXCEEDED:
@@ -108,27 +113,21 @@ def stream_logs(
 def print_logs(run_id: int, channel: grpc.Channel, timeout: int) -> None:
     """Print logs from the beginning of a run."""
     stub = ExecStub(channel)
-    req = StreamLogsRequest(run_id=run_id)
+    req = StreamLogsRequest(run_id=run_id, after_timestamp=0.0)
 
     try:
-        while True:
-            try:
-                with unauthenticated_exc_handler():
-                    # Enforce timeout for graceful exit
-                    for res in stub.StreamLogs(req, timeout=timeout):
-                        print(res.log_output)
-            except grpc.RpcError as e:
-                # pylint: disable=E1101
-                if e.code() == grpc.StatusCode.DEADLINE_EXCEEDED:
-                    break
-                if e.code() == grpc.StatusCode.NOT_FOUND:
-                    logger(ERROR, "Invalid run_id `%s`, exiting", run_id)
-                    break
-                if e.code() == grpc.StatusCode.CANCELLED:
-                    break
-                raise e
-    except KeyboardInterrupt:
-        logger(DEBUG, "Stream interrupted by user")
+        with unauthenticated_exc_handler():
+            # Enforce timeout for graceful exit
+            for res in stub.StreamLogs(req, timeout=timeout):
+                print(res.log_output)
+                break
+    except grpc.RpcError as e:
+        if e.code() == grpc.StatusCode.NOT_FOUND:  # pylint: disable=E1101
+            logger(ERROR, "Invalid run_id `%s`, exiting", run_id)
+        elif e.code() == grpc.StatusCode.DEADLINE_EXCEEDED:  # pylint: disable=E1101
+            pass
+        else:
+            raise e
     finally:
         channel.close()
         logger(DEBUG, "Channel closed")

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/new.py

@@ -213,7 +213,7 @@ def new(
         else:
             challenge_name = "Code"
             num_clients = "10"
-            dataset_name = "lucasmccabe-lmi/CodeAlpaca-20k"
+            dataset_name = "flwrlabs/code-alpaca-20k"
 
         context["llm_challenge_str"] = llm_challenge_str
         context["fraction_fit"] = fraction_fit

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/README.baseline.md.tpl

@@ -13,7 +13,7 @@ dataset: [dataset1, dataset2] # TODO: list of datasets you include in your basel
 
 > [!IMPORTANT]
 > To help having all baselines similarly formatted and structured, we have included two scripts in `baselines/dev` that when run will format your code and run some tests checking if it's formatted.
-> These checks use standard packages such as `isort`, `black`, `pylint` and others. You as a baseline creator will need to install additional pacakges. These are already specified in the `pyproject.toml` of 
+> These checks use standard packages such as `isort`, `black`, `pylint` and others. You as a baseline creator will need to install additional packages. These are already specified in the `pyproject.toml` of
 > your baseline. Follow these steps:
 
 ```bash
@@ -66,9 +66,9 @@ cd .. # so you are in the `flower/baselines` directory
 
 ## About this baseline
 
-**What’s implemented:** :warning: *_Concisely describe what experiment(s) (e.g. Figure 1, Table 2, etc) in the publication can be replicated by running the code. Please only use a few sentences. ”_*
+**What’s implemented:** :warning: *_Concisely describe what experiment(s) (e.g. Figure 1, Table 2, etc.) in the publication can be replicated by running the code. Please only use a few sentences. ”_*
 
-**Datasets:** :warning: *_List the datasets you used (if you used a medium to large dataset, >10GB please also include the sizes of the dataset). We highly recommend using [FlowerDatasets](https://flower.ai/docs/datasets/index.html) to download and partition your dataset. If you have other ways to download the data, you can also use `FlowerDatasets` to partiion it._*
+**Datasets:** :warning: *_List the datasets you used (if you used a medium to large dataset, >10GB please also include the sizes of the dataset). We highly recommend using [FlowerDatasets](https://flower.ai/docs/datasets/index.html) to download and partition your dataset. If you have other ways to download the data, you can also use `FlowerDatasets` to partition it._*
 
 **Hardware Setup:** :warning: *_Give some details about the hardware (e.g. a server with 8x V100 32GB and 256GB of RAM) you used to run the experiments for this baseline. Indicate how long it took to run the experiments. Someone out there might not have access to the same resources you have so, could you list the absolute minimum hardware needed to run the experiment in a reasonable amount of time ? (e.g. minimum is 1x 16GB GPU otherwise a client model can’t be trained with a sufficiently large batch size). Could you test this works too?_*
 
@@ -122,6 +122,6 @@ flwr run . --run-config learning-rate=0.1,coefficient=0.123
 flwr run . --run-config <my-big-experiment-config>.toml
 ```
 
-:warning: _It is preferable to show a single commmand (or multilple commands if they belong to the same experiment) and then a table/plot with the expected results, instead of showing all the commands first and then all the results/plots._
+:warning: _It is preferable to show a single command (or multiple commands if they belong to the same experiment) and then a table/plot with the expected results, instead of showing all the commands first and then all the results/plots._
 :warning: _If you present plots or other figures, please include either a Jupyter notebook showing how to create them or include a utility function that can be called after the experiments finish running._
 :warning: If you include plots or figures, save them in `.png` format and place them in a new directory named `_static` at the same level as your `README.md`.

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/README.md.tpl

@@ -18,7 +18,7 @@ Refer to the [How to Run Simulations](https://flower.ai/docs/framework/how-to-ru
 
 ## Run with the Deployment Engine
 
-Follow this [how-to guide](https://flower.ai/docs/framework/how-to-run-flower-with-deployment-engine.html) to run the same app in this example but with Flower's Deployment Engine. After that, you might be intersted in setting up [secure TLS-enabled communications](https://flower.ai/docs/framework/how-to-enable-tls-connections.html) and [SuperNode authentication](https://flower.ai/docs/framework/how-to-authenticate-supernodes.html) in your federation. 
+Follow this [how-to guide](https://flower.ai/docs/framework/how-to-run-flower-with-deployment-engine.html) to run the same app in this example but with Flower's Deployment Engine. After that, you might be interested in setting up [secure TLS-enabled communications](https://flower.ai/docs/framework/how-to-enable-tls-connections.html) and [SuperNode authentication](https://flower.ai/docs/framework/how-to-authenticate-supernodes.html) in your federation.
 
 You can run Flower on Docker too! Check out the [Flower with Docker](https://flower.ai/docs/framework/docker/index.html) documentation.
 

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.baseline.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "flwr-datasets[vision]>=0.5.0",
     "torch==2.5.1",
     "torchvision==0.20.1",

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.flowertune.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "flwr-datasets>=0.5.0",
     "torch==2.3.1",
     "trl==0.8.1",

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.huggingface.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "flwr-datasets>=0.5.0",
     "torch==2.5.1",
     "transformers>=4.30.0,<5.0",

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.jax.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "jax==0.4.30",
     "jaxlib==0.4.30",
     "scikit-learn==1.6.1",

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.mlx.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "flwr-datasets[vision]>=0.5.0",
     "mlx==0.21.1",
 ]

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.numpy.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "numpy>=2.0.2",
 ]
 

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.pytorch.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "flwr-datasets[vision]>=0.5.0",
     "torch==2.5.1",
     "torchvision==0.20.1",

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.sklearn.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "flwr-datasets[vision]>=0.5.0",
     "scikit-learn>=1.6.1",
 ]

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/cli/new/templates/app/pyproject.tensorflow.toml.tpl

@@ -8,7 +8,7 @@ version = "1.0.0"
 description = ""
 license = "Apache-2.0"
 dependencies = [
-    "flwr[simulation]>=1.15.1",
+    "flwr[simulation]>=1.16.0",
     "flwr-datasets[vision]>=0.5.0",
     "tensorflow>=2.11.1,<2.18.0",
 ]

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/client/client_app.py

@@ -16,6 +16,8 @@
 
 
 import inspect
+from collections.abc import Iterator
+from contextlib import contextmanager
 from typing import Callable, Optional
 
 from flwr.client.client import Client
@@ -71,6 +73,11 @@ def _inspect_maybe_adapt_client_fn_signature(client_fn: ClientFnExt) -> ClientFn
     return client_fn
 
 
+@contextmanager
+def _empty_lifespan(_: Context) -> Iterator[None]:
+    yield
+
+
 class ClientAppException(Exception):
     """Exception raised when an exception is raised while executing a ClientApp."""
 
@@ -95,15 +102,6 @@ class ClientApp:
     >>>    return FlowerClient().to_client()
     >>>
     >>> app = ClientApp(client_fn)
-
-    If the above code is in a Python module called `client`, it can be started as
-    follows:
-
-    >>> flower-client-app client:app --insecure
-
-    In this `client:app` example, `client` refers to the Python module `client.py` in
-    which the previous code lives in and `app` refers to the global attribute `app` that
-    points to an object of type `ClientApp`.
     """
 
     def __init__(
@@ -135,35 +133,41 @@ class ClientApp:
         self._train: Optional[ClientAppCallable] = None
         self._evaluate: Optional[ClientAppCallable] = None
         self._query: Optional[ClientAppCallable] = None
+        self._lifespan = _empty_lifespan
 
     def __call__(self, message: Message, context: Context) -> Message:
         """Execute `ClientApp`."""
-        # Execute message using `client_fn`
-        if self._call:
-            return self._call(message, context)
-
-        # Execute message using a new
-        if message.metadata.message_type == MessageType.TRAIN:
-            if self._train:
-                return self._train(message, context)
-            raise ValueError("No `train` function registered")
-        if message.metadata.message_type == MessageType.EVALUATE:
-            if self._evaluate:
-                return self._evaluate(message, context)
-            raise ValueError("No `evaluate` function registered")
-        if message.metadata.message_type == MessageType.QUERY:
-            if self._query:
-                return self._query(message, context)
-            raise ValueError("No `query` function registered")
-
-        # Message type did not match one of the known message types abvoe
-        raise ValueError(f"Unknown message_type: {message.metadata.message_type}")
-
-    def train(self) -> Callable[[ClientAppCallable], ClientAppCallable]:
+        with self._lifespan(context):
+            # Execute message using `client_fn`
+            if self._call:
+                return self._call(message, context)
+
+            # Execute message using a new
+            if message.metadata.message_type == MessageType.TRAIN:
+                if self._train:
+                    return self._train(message, context)
+                raise ValueError("No `train` function registered")
+            if message.metadata.message_type == MessageType.EVALUATE:
+                if self._evaluate:
+                    return self._evaluate(message, context)
+                raise ValueError("No `evaluate` function registered")
+            if message.metadata.message_type == MessageType.QUERY:
+                if self._query:
+                    return self._query(message, context)
+                raise ValueError("No `query` function registered")
+
+            # Message type did not match one of the known message types abvoe
+            raise ValueError(f"Unknown message_type: {message.metadata.message_type}")
+
+    def train(
+        self, mods: Optional[list[Mod]] = None
+    ) -> Callable[[ClientAppCallable], ClientAppCallable]:
         """Return a decorator that registers the train fn with the client app.
 
         Examples
         --------
+        Registering a train function:
+
         >>> app = ClientApp()
         >>>
         >>> @app.train()
@@ -171,6 +175,17 @@ class ClientApp:
         >>>    print("ClientApp training running")
         >>>    # Create and return an echo reply message
         >>>    return message.create_reply(content=message.content())
+
+        Registering a train function with a function-specific modifier:
+
+        >>> from flwr.client.mod import message_size_mod
+        >>>
+        >>> app = ClientApp()
+        >>>
+        >>> @app.train(mods=[message_size_mod])
+        >>> def train(message: Message, context: Context) -> Message:
+        >>>    print("ClientApp training running with message size mod")
+        >>>    return message.create_reply(content=message.content())
         """
 
         def train_decorator(train_fn: ClientAppCallable) -> ClientAppCallable:
@@ -182,18 +197,22 @@ class ClientApp:
 
             # Register provided function with the ClientApp object
             # Wrap mods around the wrapped step function
-            self._train = make_ffn(train_fn, self._mods)
+            self._train = make_ffn(train_fn, self._mods + (mods or []))
 
             # Return provided function unmodified
             return train_fn
 
         return train_decorator
 
-    def evaluate(self) -> Callable[[ClientAppCallable], ClientAppCallable]:
+    def evaluate(
+        self, mods: Optional[list[Mod]] = None
+    ) -> Callable[[ClientAppCallable], ClientAppCallable]:
         """Return a decorator that registers the evaluate fn with the client app.
 
         Examples
         --------
+        Registering an evaluate function:
+
         >>> app = ClientApp()
         >>>
         >>> @app.evaluate()
@@ -201,6 +220,18 @@ class ClientApp:
         >>>    print("ClientApp evaluation running")
         >>>    # Create and return an echo reply message
         >>>    return message.create_reply(content=message.content())
+
+        Registering an evaluate function with a function-specific modifier:
+
+        >>> from flwr.client.mod import message_size_mod
+        >>>
+        >>> app = ClientApp()
+        >>>
+        >>> @app.evaluate(mods=[message_size_mod])
+        >>> def evaluate(message: Message, context: Context) -> Message:
+        >>>    print("ClientApp evaluation running with message size mod")
+        >>>    # Create and return an echo reply message
+        >>>    return message.create_reply(content=message.content())
         """
 
         def evaluate_decorator(evaluate_fn: ClientAppCallable) -> ClientAppCallable:
@@ -212,18 +243,22 @@ class ClientApp:
 
             # Register provided function with the ClientApp object
             # Wrap mods around the wrapped step function
-            self._evaluate = make_ffn(evaluate_fn, self._mods)
+            self._evaluate = make_ffn(evaluate_fn, self._mods + (mods or []))
 
             # Return provided function unmodified
             return evaluate_fn
 
         return evaluate_decorator
 
-    def query(self) -> Callable[[ClientAppCallable], ClientAppCallable]:
+    def query(
+        self, mods: Optional[list[Mod]] = None
+    ) -> Callable[[ClientAppCallable], ClientAppCallable]:
         """Return a decorator that registers the query fn with the client app.
 
         Examples
         --------
+        Registering a query function:
+
         >>> app = ClientApp()
         >>>
         >>> @app.query()
@@ -231,6 +266,18 @@ class ClientApp:
         >>>    print("ClientApp query running")
         >>>    # Create and return an echo reply message
         >>>    return message.create_reply(content=message.content())
+
+        Registering a query function with a function-specific modifier:
+
+        >>> from flwr.client.mod import message_size_mod
+        >>>
+        >>> app = ClientApp()
+        >>>
+        >>> @app.query(mods=[message_size_mod])
+        >>> def query(message: Message, context: Context) -> Message:
+        >>>    print("ClientApp query running with message size mod")
+        >>>    # Create and return an echo reply message
+        >>>    return message.create_reply(content=message.content())
         """
 
         def query_decorator(query_fn: ClientAppCallable) -> ClientAppCallable:
@@ -242,13 +289,77 @@ class ClientApp:
 
             # Register provided function with the ClientApp object
             # Wrap mods around the wrapped step function
-            self._query = make_ffn(query_fn, self._mods)
+            self._query = make_ffn(query_fn, self._mods + (mods or []))
 
             # Return provided function unmodified
             return query_fn
 
         return query_decorator
 
+    def lifespan(
+        self,
+    ) -> Callable[
+        [Callable[[Context], Iterator[None]]], Callable[[Context], Iterator[None]]
+    ]:
+        """Return a decorator that registers the lifespan fn with the client app.
+
+        The decorated function should accept a `Context` object and use `yield`
+        to define enter and exit behavior.
+
+        Examples
+        --------
+        >>> app = ClientApp()
+        >>>
+        >>> @app.lifespan()
+        >>> def lifespan(context: Context) -> None:
+        >>>     # Perform initialization tasks before the app starts
+        >>>     print("Initializing ClientApp")
+        >>>
+        >>>     yield  # ClientApp is running
+        >>>
+        >>>     # Perform cleanup tasks after the app stops
+        >>>     print("Cleaning up ClientApp")
+        """
+
+        def lifespan_decorator(
+            lifespan_fn: Callable[[Context], Iterator[None]]
+        ) -> Callable[[Context], Iterator[None]]:
+            """Register the lifespan fn with the ServerApp object."""
+            warn_preview_feature("ClientApp-register-lifespan-function")
+
+            @contextmanager
+            def decorated_lifespan(context: Context) -> Iterator[None]:
+                # Execute the code before `yield` in lifespan_fn
+                try:
+                    if not isinstance(it := lifespan_fn(context), Iterator):
+                        raise StopIteration
+                    next(it)
+                except StopIteration:
+                    raise RuntimeError(
+                        "lifespan function should yield at least once."
+                    ) from None
+
+                try:
+                    # Enter the context
+                    yield
+                finally:
+                    try:
+                        # Execute the code after `yield` in lifespan_fn
+                        next(it)
+                    except StopIteration:
+                        pass
+                    else:
+                        raise RuntimeError("lifespan function should only yield once.")
+
+            # Register provided function with the ClientApp object
+            # Ignore mypy error because of different argument names (`_` vs `context`)
+            self._lifespan = decorated_lifespan  # type: ignore
+
+            # Return provided function unmodified
+            return lifespan_fn
+
+        return lifespan_decorator
+
 
 class LoadClientAppError(Exception):
     """Error when trying to load `ClientApp`."""

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/client/clientapp/app.py

@@ -16,6 +16,7 @@
 
 
 import argparse
+import gc
 import time
 from logging import DEBUG, ERROR, INFO
 from typing import Optional
@@ -160,6 +161,9 @@ def run_clientapp(  # pylint: disable=R0914
                 stub=stub, token=token, message=reply_message, context=context
             )
 
+            del client_app, message, context, run, fab, reply_message
+            gc.collect()
+
             # Reset token to `None` to prevent flwr-clientapp from trying to pull the
             # same inputs again
             token = None

{flwr-1.15.1 → flwr-1.16.0}/src/py/flwr/client/message_handler/message_handler.py

@@ -82,7 +82,7 @@ def handle_control_message(message: Message) -> tuple[Optional[Message], int]:
         recordset = RecordSet()
         recordset.configs_records["config"] = ConfigsRecord({"reason": reason})
         out_message = message.create_reply(recordset)
-        # Return TaskRes and sleep duration
+        # Return Message and sleep duration
         return out_message, sleep_duration
 
     # Any other message