metaflow 2.19.3__tar.gz → 2.19.4__tar.gz

{metaflow-2.19.3/metaflow.egg-info → metaflow-2.19.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: metaflow
-Version: 2.19.3
+Version: 2.19.4
 Summary: Metaflow: More AI and ML, Less Engineering
 Author: Metaflow Developers
 Author-email: help@metaflow.org
@@ -26,7 +26,7 @@ License-File: LICENSE
 Requires-Dist: requests
 Requires-Dist: boto3
 Provides-Extra: stubs
-Requires-Dist: metaflow-stubs==2.19.3; extra == "stubs"
+Requires-Dist: metaflow-stubs==2.19.4; extra == "stubs"
 Dynamic: author
 Dynamic: author-email
 Dynamic: classifier

{metaflow-2.19.3 → metaflow-2.19.4}/devtools/Tiltfile

@@ -630,7 +630,7 @@ if "metadata-service" in enabled_components:
             'metadatadb.database=metaflow',
             'metadatadb.host=postgresql',
             'image.repository=public.ecr.aws/outerbounds/metaflow_metadata_service',
-            'image.tag=2.4.13-2-g70af4ed',
+            'image.tag=2.5.0',
             'resources.requests.cpu=25m',
             'resources.requests.memory=64Mi',
             'resources.limits.cpu=50m',

{metaflow-2.19.3 → metaflow-2.19.4}/metaflow/datastore/flow_datastore.py

@@ -167,7 +167,8 @@ class FlowDataStore(object):
         if attempt is not None and attempt <= metaflow_config.MAX_ATTEMPTS - 1:
             attempt_range = range(attempt + 1) if include_prior else [attempt]
         for task_url in task_urls:
-            task_splits = task_url.split("/")
+            # task_url can have a trailing slash, so strip this to avoid empty strings in the split
+            task_splits = task_url.rstrip("/").split("/")
             # Usually it is flow, run, step, task (so 4 components) -- if we have a
             # fifth one, there is a specific attempt number listed as well.
             task_attempt_range = attempt_range

{metaflow-2.19.3 → metaflow-2.19.4}/metaflow/plugins/env_escape/client_modules.py

@@ -1,5 +1,6 @@
 import atexit
 import importlib
+import importlib.util
 import itertools
 import pickle
 import re
@@ -41,6 +42,8 @@ class _WrappedModule(object):
     def __getattr__(self, name):
         if name == "__loader__":
             return self._loader
+        if name == "__spec__":
+            return importlib.util.spec_from_loader(self._prefix, self._loader)
         if name in ("__name__", "__package__"):
             return self._prefix
         if name in ("__file__", "__path__"):
@@ -71,7 +74,8 @@ class _WrappedModule(object):
             # Try to see if this is a submodule that we can load
             m = None
             try:
-                m = self._loader.load_module(".".join([self._prefix, name]))
+                submodule_name = ".".join([self._prefix, name])
+                m = importlib.import_module(submodule_name)
             except ImportError:
                 pass
             if m is None:
@@ -117,7 +121,28 @@ class _WrappedModule(object):
 
 
 class ModuleImporter(object):
-    # This ModuleImporter implements the Importer Protocol defined in PEP 302
+    """
+    A custom import hook that proxies module imports to a different Python environment.
+
+    This class implements the MetaPathFinder and Loader protocols (PEP 451) to enable
+    "environment escape" - allowing the current Python process to import and use modules
+    from a different Python interpreter with potentially different versions or packages.
+
+    When a module is imported through this importer:
+    1. A client spawns a server process in the target Python environment
+    2. The module is loaded in the remote environment
+    3. A _WrappedModule proxy is returned that forwards all operations (function calls,
+       attribute access, etc.) to the remote environment via RPC
+    4. Data is serialized/deserialized using pickle for cross-environment communication
+
+    Args:
+        python_executable: Path to the Python interpreter for the remote environment
+        pythonpath: Python path to use in the remote environment
+        max_pickle_version: Maximum pickle protocol version supported by remote interpreter
+        config_dir: Directory containing configuration for the environment escape
+        module_prefixes: List of module name prefixes to handle
+    """
+
     def __init__(
         self,
         python_executable,
@@ -135,84 +160,89 @@ class ModuleImporter(object):
         self._handled_modules = None
         self._aliases = {}
 
-    def find_module(self, fullname, path=None):
+    def find_spec(self, fullname, path=None, target=None):
         if self._handled_modules is not None:
             if get_canonical_name(fullname, self._aliases) in self._handled_modules:
-                return self
+                return importlib.util.spec_from_loader(fullname, self)
             return None
         if any([fullname.startswith(prefix) for prefix in self._module_prefixes]):
             # We potentially handle this
-            return self
+            return importlib.util.spec_from_loader(fullname, self)
         return None
 
-    def load_module(self, fullname):
-        if fullname in sys.modules:
-            return sys.modules[fullname]
-        if self._client is None:
-            if sys.version_info[0] < 3:
-                raise NotImplementedError(
-                    "Environment escape imports are not supported in Python 2"
-                )
-            # We initialize a client and query the modules we handle
-            # The max_pickle_version is the pickle version that the server (so
-            # the underlying interpreter we call into) supports; we determine
-            # what version the current environment support and take the minimum
-            # of those two
-            max_pickle_version = min(self._max_pickle_version, pickle.HIGHEST_PROTOCOL)
-
-            self._client = Client(
-                self._module_prefixes,
-                self._python_executable,
-                self._pythonpath,
-                max_pickle_version,
-                self._config_dir,
-            )
-            atexit.register(_clean_client, self._client)
-
-            # Get information about overrides and what the server knows about
-            exports = self._client.get_exports()
-
-            prefixes = set()
-            export_classes = exports.get("classes", [])
-            export_functions = exports.get("functions", [])
-            export_values = exports.get("values", [])
-            export_exceptions = exports.get("exceptions", [])
-            self._aliases = exports.get("aliases", {})
-            for name in itertools.chain(
-                export_classes,
-                export_functions,
-                export_values,
-                (e[0] for e in export_exceptions),
-            ):
-                splits = name.rsplit(".", 1)
-                prefixes.add(splits[0])
-            # We will make sure that we create modules even for "empty" prefixes
-            # because packages are always loaded hierarchically so if we have
-            # something in `a.b.c` but nothing directly in `a`, we still need to
-            # create a module named `a`. There is probably a better way of doing this
-            all_prefixes = list(prefixes)
-            for prefix in all_prefixes:
-                parts = prefix.split(".")
-                cur = parts[0]
-                for i in range(1, len(parts)):
-                    prefixes.add(cur)
-                    cur = ".".join([cur, parts[i]])
-
-            # We now know all the modules that we can handle. We update
-            # handled_module and return the module if we have it or raise ImportError
-            self._handled_modules = {}
-            for prefix in prefixes:
-                self._handled_modules[prefix] = _WrappedModule(
-                    self, prefix, exports, self._client
-                )
+    def create_module(self, spec):
+        # Return the pre-created wrapped module for this spec
+        self._initialize_client()
+
+        fullname = spec.name
         canonical_fullname = get_canonical_name(fullname, self._aliases)
-        # Modules are created canonically but we need to return something for any
-        # of the aliases.
-        module = self._handled_modules.get(canonical_fullname)
-        if module is None:
-            raise ImportError
-        sys.modules[fullname] = module
-        return module
+        # Modules are created canonically but we need to handle any of the aliases.
+        wrapped_module = self._handled_modules.get(canonical_fullname)
+        if wrapped_module is None:
+            raise ImportError(f"No module named '{fullname}'")
+        return wrapped_module
+
+    def exec_module(self, module):
+        # No initialization needed since the wrapped module returned by
+        # create_module() is fully initialized
+        pass
+
+    def _initialize_client(self):
+        if self._client is not None:
+            return
+
+        # We initialize a client and query the modules we handle
+        # The max_pickle_version is the pickle version that the server (so
+        # the underlying interpreter we call into) supports; we determine
+        # what version the current environment support and take the minimum
+        # of those two
+        max_pickle_version = min(self._max_pickle_version, pickle.HIGHEST_PROTOCOL)
+
+        self._client = Client(
+            self._module_prefixes,
+            self._python_executable,
+            self._pythonpath,
+            max_pickle_version,
+            self._config_dir,
+        )
+        atexit.register(_clean_client, self._client)
+
+        # Get information about overrides and what the server knows about
+        exports = self._client.get_exports()
+
+        prefixes = set()
+        export_classes = exports.get("classes", [])
+        export_functions = exports.get("functions", [])
+        export_values = exports.get("values", [])
+        export_exceptions = exports.get("exceptions", [])
+        self._aliases = exports.get("aliases", {})
+        for name in itertools.chain(
+            export_classes,
+            export_functions,
+            export_values,
+            (e[0] for e in export_exceptions),
+        ):
+            splits = name.rsplit(".", 1)
+            prefixes.add(splits[0])
+        # We will make sure that we create modules even for "empty" prefixes
+        # because packages are always loaded hierarchically so if we have
+        # something in `a.b.c` but nothing directly in `a`, we still need to
+        # create a module named `a`. There is probably a better way of doing this
+        all_prefixes = list(prefixes)
+        for prefix in all_prefixes:
+            parts = prefix.split(".")
+            cur = parts[0]
+            for i in range(1, len(parts)):
+                prefixes.add(cur)
+                cur = ".".join([cur, parts[i]])
+
+        # We now know all the modules that we can handle. We update
+        # handled_module and return the module if we have it or raise ImportError
+        self._handled_modules = {}
+        for prefix in prefixes:
+            self._handled_modules[prefix] = _WrappedModule(
+                self, prefix, exports, self._client
+            )
 
 
 def create_modules(python_executable, pythonpath, max_pickle_version, path, prefixes):

{metaflow-2.19.3 → metaflow-2.19.4}/metaflow/user_decorators/mutable_flow.py

@@ -239,7 +239,7 @@ class MutableFlow:
                     "Mutable flow removing parameter %s from flow" % var
                 )
                 # Reset so that we don't list it again
-                del self._flow_cls._flow_state[_FlowState.CACHED_PARAMETERS]
+                self._flow_cls._flow_state.pop(_FlowState.CACHED_PARAMETERS, None)
                 return True
         debug.userconf_exec(
             "Mutable flow failed to remove parameter %s from flow" % parameter_name

metaflow-2.19.4/metaflow/version.py

@@ -0,0 +1 @@
+metaflow_version = "2.19.4"

{metaflow-2.19.3 → metaflow-2.19.4/metaflow.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: metaflow
-Version: 2.19.3
+Version: 2.19.4
 Summary: Metaflow: More AI and ML, Less Engineering
 Author: Metaflow Developers
 Author-email: help@metaflow.org
@@ -26,7 +26,7 @@ License-File: LICENSE
 Requires-Dist: requests
 Requires-Dist: boto3
 Provides-Extra: stubs
-Requires-Dist: metaflow-stubs==2.19.3; extra == "stubs"
+Requires-Dist: metaflow-stubs==2.19.4; extra == "stubs"
 Dynamic: author
 Dynamic: author-email
 Dynamic: classifier

{metaflow-2.19.3 → metaflow-2.19.4}/metaflow.egg-info/requires.txt

@@ -2,4 +2,4 @@ requests
 boto3
 
 [stubs]
-metaflow-stubs==2.19.3
+metaflow-stubs==2.19.4

metaflow-2.19.3/metaflow/version.py

@@ -1 +0,0 @@
-metaflow_version = "2.19.3"