[tune] Remove checkpoint_dir and reporter deprecation notices (ra…

…y-project#42698)

These APIs were hard-deprecated in Ray 2.7, and the utility functions to detect them are incorrectly flagging some cases, causing users to run into DeprecationWarning errors when they shouldn't. This PR removes the deprecated APIs and these unnecessary checks. 

---------

Signed-off-by: Justin Yu <[email protected]>

python/ray/train/base_trainer.py

@@ -733,7 +733,7 @@ def train_func(config):
         # stdout messages and the results directory.
         train_func.__name__ = trainer_cls.__name__
 
-        trainable_cls = wrap_function(train_func, warn=False)
+        trainable_cls = wrap_function(train_func)
         has_base_dataset = bool(self.datasets)
         if has_base_dataset:
             from ray.data.context import DataContext

python/ray/tune/BUILD

@@ -132,14 +132,6 @@ py_test(
     tags = ["team:ml", "exclusive"],
 )
 
-py_test(
-    name = "test_function_api_legacy",
-    size = "small",
-    srcs = ["tests/test_function_api_legacy.py"],
-    deps = [":tune_lib"],
-    tags = ["team:ml", "exclusive"],
-)
-
 py_test(
     name = "test_integration_pytorch_lightning",
     size = "small",

python/ray/tune/registry.py

@@ -102,10 +102,10 @@ def register_trainable(name: str, trainable: Union[Callable, Type], warn: bool =
         logger.debug("Detected class for trainable.")
     elif isinstance(trainable, FunctionType) or isinstance(trainable, partial):
         logger.debug("Detected function for trainable.")
-        trainable = wrap_function(trainable, warn=warn)
+        trainable = wrap_function(trainable)
     elif callable(trainable):
         logger.info("Detected unknown callable for trainable. Converting to class.")
-        trainable = wrap_function(trainable, warn=warn)
+        trainable = wrap_function(trainable)
 
     if not issubclass(trainable, Trainable):
         raise TypeError("Second argument must be convertable to Trainable", trainable)
@@ -246,7 +246,7 @@ def unregister(self, category, key):
 
     def unregister_all(self, category: Optional[str] = None):
         remaining = set()
-        for (cat, key) in self._registered:
+        for cat, key in self._registered:
             if category and category == cat:
                 self.unregister(cat, key)
             else:

python/ray/tune/trainable/function_trainable.py

@@ -27,11 +27,7 @@
     SHOULD_CHECKPOINT,
 )
 from ray.tune.trainable import Trainable
-from ray.tune.utils import (
-    _detect_checkpoint_function,
-    _detect_config_single,
-    _detect_reporter,
-)
+from ray.tune.utils import _detect_config_single
 from ray.util.annotations import DeveloperAPI
 
 
@@ -44,64 +40,6 @@
 TEMP_MARKER = ".temp_marker"
 
 
-_CHECKPOINT_DIR_ARG_DEPRECATION_MSG = """Accepting a `checkpoint_dir` argument in your training function is deprecated.
-Please use `ray.train.get_checkpoint()` to access your checkpoint as a
-`ray.train.Checkpoint` object instead. See below for an example:
-
-Before
-------
-
-from ray import tune
-
-def train_fn(config, checkpoint_dir=None):
-    if checkpoint_dir:
-        torch.load(os.path.join(checkpoint_dir, "checkpoint.pt"))
-    ...
-
-tuner = tune.Tuner(train_fn)
-tuner.fit()
-
-After
------
-
-from ray import train, tune
-
-def train_fn(config):
-    checkpoint: train.Checkpoint = train.get_checkpoint()
-    if checkpoint:
-        with checkpoint.as_directory() as checkpoint_dir:
-            torch.load(os.path.join(checkpoint_dir, "checkpoint.pt"))
-    ...
-
-tuner = tune.Tuner(train_fn)
-tuner.fit()"""  # noqa: E501
-
-_REPORTER_ARG_DEPRECATION_MSG = """Accepting a `reporter` in your training function is deprecated.
-Please use `ray.train.report()` to report results instead. See below for an example:
-
-Before
-------
-
-from ray import tune
-
-def train_fn(config, reporter):
-    reporter(metric=1)
-
-tuner = tune.Tuner(train_fn)
-tuner.fit()
-
-After
------
-
-from ray import train, tune
-
-def train_fn(config):
-    train.report({"metric": 1})
-
-tuner = tune.Tuner(train_fn)
-tuner.fit()"""  # noqa: E501
-
-
 @DeveloperAPI
 class FunctionTrainable(Trainable):
     """Trainable that runs a user function reporting results.
@@ -271,29 +209,20 @@ def _report_thread_runner_error(self, block=False):
 
 @DeveloperAPI
 def wrap_function(
-    train_func: Callable[[Any], Any], warn: bool = True, name: Optional[str] = None
+    train_func: Callable[[Any], Any], name: Optional[str] = None
 ) -> Type["FunctionTrainable"]:
     inherit_from = (FunctionTrainable,)
 
     if hasattr(train_func, "__mixins__"):
         inherit_from = train_func.__mixins__ + inherit_from
 
     func_args = inspect.getfullargspec(train_func).args
-    use_checkpoint = _detect_checkpoint_function(train_func)
     use_config_single = _detect_config_single(train_func)
-    use_reporter = _detect_reporter(train_func)
-
-    if use_checkpoint:
-        raise DeprecationWarning(_CHECKPOINT_DIR_ARG_DEPRECATION_MSG)
-
-    if use_reporter:
-        raise DeprecationWarning(_REPORTER_ARG_DEPRECATION_MSG)
 
     if not use_config_single:
-        # use_reporter is hidden
         raise ValueError(
             "Unknown argument found in the Trainable function. "
-            "The function args must include a 'config' positional parameter."
+            "The function args must include a single 'config' positional parameter.\n"
             "Found: {}".format(func_args)
         )
 

python/ray/tune/trainable/util.py

@@ -10,7 +10,6 @@
 )
 from ray.air.config import ScalingConfig
 from ray.tune.registry import _ParameterRegistry
-from ray.tune.utils import _detect_checkpoint_function
 from ray.util.annotations import PublicAPI
 
 if TYPE_CHECKING:
@@ -124,12 +123,6 @@ def setup(self, config):
         trainable_with_params = _Inner
     else:
         # Function trainable
-        if _detect_checkpoint_function(trainable, partial=True):
-            from ray.tune.trainable.function_trainable import (
-                _CHECKPOINT_DIR_ARG_DEPRECATION_MSG,
-            )
-
-            raise DeprecationWarning(_CHECKPOINT_DIR_ARG_DEPRECATION_MSG)
 
         def inner(config):
             fn_kwargs = {}
@@ -223,13 +216,6 @@ def train_fn(config):
     if not inspect.isclass(trainable):
         if isinstance(trainable, types.MethodType):
             # Methods cannot set arbitrary attributes, so we have to wrap them
-            if _detect_checkpoint_function(trainable, partial=True):
-                from ray.tune.trainable.function_trainable import (
-                    _CHECKPOINT_DIR_ARG_DEPRECATION_MSG,
-                )
-
-                raise DeprecationWarning(_CHECKPOINT_DIR_ARG_DEPRECATION_MSG)
-
             def _trainable(config):
                 return trainable(config)
 

python/ray/tune/utils/__init__.py

@@ -8,8 +8,6 @@
     validate_save_restore,
     warn_if_slow,
     diagnose_serialization,
-    _detect_checkpoint_function,
-    _detect_reporter,
     _detect_config_single,
     wait_for_gpu,
 )
@@ -24,8 +22,6 @@
     "validate_save_restore",
     "warn_if_slow",
     "diagnose_serialization",
-    "_detect_checkpoint_function",
-    "_detect_reporter",
     "_detect_config_single",
     "wait_for_gpu",
 ]

python/ray/tune/utils/util.py

@@ -584,40 +584,6 @@ def validate_save_restore(
     return True
 
 
-def _detect_checkpoint_function(train_func, abort=False, partial=False):
-    """Use checkpointing if any arg has "checkpoint_dir" and args = 2"""
-    func_sig = inspect.signature(train_func)
-    validated = True
-    try:
-        # check if signature is func(config, checkpoint_dir=None)
-        if partial:
-            func_sig.bind_partial({}, checkpoint_dir="tmp/path")
-        else:
-            func_sig.bind({}, checkpoint_dir="tmp/path")
-    except Exception as e:
-        logger.debug(str(e))
-        validated = False
-    if abort and not validated:
-        func_args = inspect.getfullargspec(train_func).args
-        raise ValueError(
-            "Provided training function must have 1 `config` argument "
-            "`func(config)`. Got {}".format(func_args)
-        )
-    return validated
-
-
-def _detect_reporter(func):
-    """Use reporter if any arg has "reporter" and args = 2"""
-    func_sig = inspect.signature(func)
-    use_reporter = True
-    try:
-        func_sig.bind({}, reporter=None)
-    except Exception as e:
-        logger.debug(str(e))
-        use_reporter = False
-    return use_reporter
-
-
 def _detect_config_single(func):
     """Check if func({}) works."""
     func_sig = inspect.signature(func)