optimizely-sdk 5.4.0__tar.gz → 5.5.0__tar.gz

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/CHANGELOG.md

@@ -1,5 +1,33 @@
 # Optimizely Python SDK Changelog
 
+## 5.5.0
+April 30th, 2026
+
+### New Features
+
+#### **Feature Rollout Support**
+
+- Added support for Feature Rollouts, a new rule type that combines Targeted Delivery simplicity with A/B test measurement capabilities. During project config parsing, the "everyone else" variation from the flag's rollout is automatically injected into feature rollout experiments, enabling correct evaluation without changes to decision logic. ([#499](https://github.com/optimizely/python-sdk/pull/499))
+- Removed experiment type validation from config parsing to support flexible rule types. ([#500](https://github.com/optimizely/python-sdk/pull/500))
+
+---
+
+### Enhancements
+
+- Added event retry support for reliable event dispatching. ([#475](https://github.com/optimizely/python-sdk/pull/475))
+- Excluded CMAB experiments from UserProfileService to prevent stale cached decisions. ([#474](https://github.com/optimizely/python-sdk/pull/474))
+- Cleaned up flag-level holdout fields for simplified configuration. ([#507](https://github.com/optimizely/python-sdk/pull/507))
+
+---
+
+### Bug Fixes
+
+- Fixed `return` in `finally` block silently swallowing exceptions. ([#505](https://github.com/optimizely/python-sdk/pull/505))
+- Fixed `PollingConfigManager` rejecting valid URL-only configuration. ([#497](https://github.com/optimizely/python-sdk/pull/497))
+
+---
+
+
 ## 5.4.0
 December 19th, 2025
 

{optimizely-sdk-5.4.0/optimizely_sdk.egg-info → optimizely_sdk-5.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: optimizely-sdk
-Version: 5.4.0
+Version: 5.5.0
 Summary: Python SDK for Optimizely Feature Experimentation, Optimizely Full Stack (legacy), and Optimizely Rollouts.
 Home-page: https://github.com/optimizely/python-sdk
 Author: Optimizely
@@ -17,8 +17,34 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Description-Content-Type: text/markdown
-Provides-Extra: test
 License-File: LICENSE
+Requires-Dist: jsonschema>=3.2.0
+Requires-Dist: pyrsistent>=0.16.0
+Requires-Dist: requests>=2.21
+Requires-Dist: idna>=2.10
+Requires-Dist: rpds-py<0.20.0; python_version < "3.11"
+Provides-Extra: test
+Requires-Dist: requests>=2.21; extra == "test"
+Requires-Dist: idna>=2.10; extra == "test"
+Requires-Dist: ruff>=0.15.0; extra == "test"
+Requires-Dist: python-coveralls; extra == "test"
+Requires-Dist: coverage; extra == "test"
+Requires-Dist: pytest-cov; extra == "test"
+Requires-Dist: rpds-py<0.20.0; python_version < "3.11" and extra == "test"
+Requires-Dist: jsonschema>=3.2.0; extra == "test"
+Requires-Dist: funcsigs>=0.4; extra == "test"
+Requires-Dist: pytest>=6.2.0; extra == "test"
+Requires-Dist: pyrsistent>=0.16.0; extra == "test"
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: home-page
+Dynamic: license-file
+Dynamic: provides-extra
+Dynamic: requires-dist
+Dynamic: summary
 
 Optimizely Feature Experimentation is A/B testing and feature management for product development teams. Experiment in any application. Make every feature on your roadmap an opportunity to learn. Learn more at https://www.optimizely.com/products/experiment/feature-experimentation/ or see our documentation at https://docs.developers.optimizely.com/experimentation/v4.0.0-full-stack/docs/welcome. # Optimizely Python SDK
 
@@ -278,6 +304,34 @@ idna (BSD 3-Clause License https://github.com/kjd/idna/blob/master/LICENSE.md)
 - Swift - https://github.com/optimizely/swift-sdk
 # Optimizely Python SDK Changelog
 
+## 5.5.0
+April 30th, 2026
+
+### New Features
+
+#### **Feature Rollout Support**
+
+- Added support for Feature Rollouts, a new rule type that combines Targeted Delivery simplicity with A/B test measurement capabilities. During project config parsing, the "everyone else" variation from the flag's rollout is automatically injected into feature rollout experiments, enabling correct evaluation without changes to decision logic. ([#499](https://github.com/optimizely/python-sdk/pull/499))
+- Removed experiment type validation from config parsing to support flexible rule types. ([#500](https://github.com/optimizely/python-sdk/pull/500))
+
+---
+
+### Enhancements
+
+- Added event retry support for reliable event dispatching. ([#475](https://github.com/optimizely/python-sdk/pull/475))
+- Excluded CMAB experiments from UserProfileService to prevent stale cached decisions. ([#474](https://github.com/optimizely/python-sdk/pull/474))
+- Cleaned up flag-level holdout fields for simplified configuration. ([#507](https://github.com/optimizely/python-sdk/pull/507))
+
+---
+
+### Bug Fixes
+
+- Fixed `return` in `finally` block silently swallowing exceptions. ([#505](https://github.com/optimizely/python-sdk/pull/505))
+- Fixed `PollingConfigManager` rejecting valid URL-only configuration. ([#497](https://github.com/optimizely/python-sdk/pull/497))
+
+---
+
+
 ## 5.4.0
 December 19th, 2025
 

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/config_manager.py

@@ -145,14 +145,14 @@ class StaticConfigManager(BaseConfigManager):
         except optimizely_exceptions.UnsupportedDatafileVersionException as error:
             error_msg = error.args[0]
             error_to_handle = error
-        except:
+        except Exception:
             error_msg = enums.Errors.INVALID_INPUT.format('datafile')
             error_to_handle = optimizely_exceptions.InvalidInputException(error_msg)
-        finally:
-            if error_msg or config is None:
-                self.logger.error(error_msg)
-                self.error_handler.handle_error(error_to_handle or Exception('Unknown Error'))
-                return
+
+        if error_msg or config is None:
+            self.logger.error(error_msg)
+            self.error_handler.handle_error(error_to_handle or Exception('Unknown Error'))
+            return
 
         previous_revision = self._config.get_revision() if self._config else None
 
@@ -236,9 +236,6 @@ class PollingConfigManager(StaticConfigManager):
         )
         self._sdk_key = sdk_key or self._sdk_key
 
-        if self._sdk_key is None:
-            raise optimizely_exceptions.InvalidInputException(enums.Errors.MISSING_SDK_KEY)
-
         self.datafile_url = self.get_datafile_url(
             self._sdk_key, url, url_template or self.DATAFILE_URL_TEMPLATE
         )

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/decision_service.py

@@ -515,6 +515,11 @@ class DecisionService:
                     'reasons': decide_reasons,
                     'variation': None
                 }
+            ignore_user_profile = True
+            self.logger.debug(
+                f'Skipping user profile service for CMAB experiment "{experiment.key}". '
+                f'CMAB decisions are dynamic and not stored for sticky bucketing.'
+            )
             variation_id = cmab_decision['variation_id'] if cmab_decision else None
             cmab_uuid = cmab_decision['cmab_uuid'] if cmab_decision else None
             variation = project_config.get_variation_from_id(experiment_key=experiment.key,

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/entities.py

@@ -87,6 +87,7 @@ class Experiment(BaseEntity):
         groupId: Optional[str] = None,
         groupPolicy: Optional[str] = None,
         cmab: Optional[CmabDict] = None,
+        type: Optional[str] = None,
         **kwargs: Any
     ):
         self.id = id
@@ -101,6 +102,7 @@ class Experiment(BaseEntity):
         self.groupId = groupId
         self.groupPolicy = groupPolicy
         self.cmab = cmab
+        self.type = type
 
     def get_audience_conditions_or_ids(self) -> Sequence[str | list[str]]:
         """ Returns audienceConditions if present, otherwise audienceIds. """
@@ -220,8 +222,6 @@ class Holdout(BaseEntity):
         variations: list[VariationDict],
         trafficAllocation: list[TrafficAllocation],
         audienceIds: list[str],
-        includedFlags: Optional[list[str]] = None,
-        excludedFlags: Optional[list[str]] = None,
         audienceConditions: Optional[Sequence[str | list[str]]] = None,
         **kwargs: Any
     ):
@@ -232,8 +232,6 @@ class Holdout(BaseEntity):
         self.trafficAllocation = trafficAllocation
         self.audienceIds = audienceIds
         self.audienceConditions = audienceConditions
-        self.includedFlags = includedFlags or []
-        self.excludedFlags = excludedFlags or []
 
     def get_audience_conditions_or_ids(self) -> Sequence[str | list[str]]:
         """Returns audienceConditions if present, otherwise audienceIds.

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/event_dispatcher.py

@@ -49,7 +49,7 @@ class EventDispatcher:
             session = requests.Session()
 
             retries = Retry(total=EventDispatchConfig.RETRIES,
-                            backoff_factor=0.1,
+                            backoff_factor=0.2,
                             status_forcelist=[500, 502, 503, 504])
             adapter = HTTPAdapter(max_retries=retries)
 

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/helpers/enums.py

@@ -229,6 +229,14 @@ class OdpManagerConfig:
     EVENT_TYPE: Final = 'fullstack'
 
 
+class ExperimentTypes:
+    ab: Final = 'ab'
+    mab: Final = 'mab'
+    cmab: Final = 'cmab'
+    td: Final = 'td'
+    fr: Final = 'fr'
+
+
 class OdpSegmentsCacheConfig:
     """ODP Segment Cache configs."""
     DEFAULT_CAPACITY: Final = 10_000

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/helpers/types.py

@@ -117,6 +117,10 @@ class CmabDict(BaseEntity):
     trafficAllocation: int
 
 
+ExperimentType = Literal[
+    'ab', 'mab', 'cmab', 'td', 'fr'
+]
+
 HoldoutStatus = Literal['Draft', 'Running', 'Concluded', 'Archived']
 
 
@@ -126,5 +130,3 @@ class HoldoutDict(ExperimentDict):
     Extends ExperimentDict with holdout-specific properties.
     """
     holdoutStatus: HoldoutStatus
-    includedFlags: list[str]
-    excludedFlags: list[str]

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/helpers/validator.py

@@ -193,7 +193,7 @@ def is_user_profile_valid(user_profile: dict[str, Any]) -> bool:
     if not user_profile:
         return False
 
-    if not type(user_profile) is dict:
+    if type(user_profile) is not dict:
         return False
 
     if UserProfile.USER_ID_KEY not in user_profile:
@@ -203,7 +203,7 @@ def is_user_profile_valid(user_profile: dict[str, Any]) -> bool:
         return False
 
     experiment_bucket_map = user_profile.get(UserProfile.EXPERIMENT_BUCKET_MAP_KEY)
-    if not type(experiment_bucket_map) is dict:
+    if type(experiment_bucket_map) is not dict:
         return False
 
     for decision in experiment_bucket_map.values():

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/odp/odp_event_manager.py

@@ -163,6 +163,8 @@ class OdpEventManager:
 
         self.logger.debug(f'ODP event queue: flushing batch size {batch_len}.')
         should_retry = False
+        initial_retry_interval = 0.2  # 200ms
+        max_retry_interval = 1.0  # 1 second
 
         for i in range(1 + self.retry_count):
             try:
@@ -176,7 +178,12 @@ class OdpEventManager:
             if not should_retry:
                 break
             if i < self.retry_count:
-                self.logger.debug('Error dispatching ODP events, scheduled to retry.')
+                # Exponential backoff: 200ms, 400ms, 800ms, ... capped at 1s
+                delay = initial_retry_interval * (2 ** i)
+                if delay > max_retry_interval:
+                    delay = max_retry_interval
+                self.logger.debug(f'Error dispatching ODP events, retrying after {delay}s.')
+                time.sleep(delay)
 
         if should_retry:
             self.logger.error(Errors.ODP_EVENT_FAILED.format(f'Failed after {i} retries: {self._current_batch}'))

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/project_config.py

@@ -93,9 +93,6 @@ class ProjectConfig:
         holdouts_data: list[types.HoldoutDict] = config.get('holdouts', [])
         self.holdouts: list[entities.Holdout] = []
         self.holdout_id_map: dict[str, entities.Holdout] = {}
-        self.global_holdouts: list[entities.Holdout] = []
-        self.included_holdouts: dict[str, list[entities.Holdout]] = {}
-        self.excluded_holdouts: dict[str, list[entities.Holdout]] = {}
         self.flag_holdouts_map: dict[str, list[entities.Holdout]] = {}
 
         # Convert holdout dicts to Holdout entities
@@ -104,33 +101,13 @@ class ProjectConfig:
             holdout = entities.Holdout(**holdout_data)
             self.holdouts.append(holdout)
 
-            # Only process Running holdouts but doing it here for efficiency like the original Python implementation)
+            # Only process Running holdouts
             if not holdout.is_activated:
                 continue
 
             # Map by ID for quick lookup
             self.holdout_id_map[holdout.id] = holdout
 
-            # Categorize as global vs flag-specific
-            # Global holdouts: apply to all flags unless explicitly excluded
-            # Flag-specific holdouts: only apply to explicitly included flags
-            if not holdout.includedFlags:
-                # This is a global holdout
-                self.global_holdouts.append(holdout)
-
-                # Track which flags this global holdout excludes
-                if holdout.excludedFlags:
-                    for flag_id in holdout.excludedFlags:
-                        if flag_id not in self.excluded_holdouts:
-                            self.excluded_holdouts[flag_id] = []
-                        self.excluded_holdouts[flag_id].append(holdout)
-            else:
-                # This holdout applies to specific flags only
-                for flag_id in holdout.includedFlags:
-                    if flag_id not in self.included_holdouts:
-                        self.included_holdouts[flag_id] = []
-                    self.included_holdouts[flag_id].append(holdout)
-
         # Utility maps for quick lookup
         self.group_id_map: dict[str, entities.Group] = self._generate_key_map(self.groups, 'id', entities.Group)
         self.experiment_id_map: dict[str, entities.Experiment] = self._generate_key_map(
@@ -232,19 +209,39 @@ class ProjectConfig:
                 self.experiment_feature_map[exp_id] = [feature.id]
                 rules.append(self.experiment_id_map[exp_id])
 
-            flag_id = feature.id
-            applicable_holdouts: list[entities.Holdout] = []
-
-            # Add global holdouts first, excluding any that are explicitly excluded for this flag
-            excluded_holdouts = self.excluded_holdouts.get(flag_id, [])
-            for holdout in self.global_holdouts:
-                if holdout not in excluded_holdouts:
-                    applicable_holdouts.append(holdout)
-
-            # Add flag-specific local holdouts AFTER global holdouts
-            if flag_id in self.included_holdouts:
-                applicable_holdouts.extend(self.included_holdouts[flag_id])
-
+            # Feature Rollout support: inject the "everyone else" variation
+            # into any experiment with type == "feature_rollout"
+            everyone_else_variation = self._get_everyone_else_variation(feature)
+            if everyone_else_variation is not None:
+                for experiment in rules:
+                    if experiment.type == enums.ExperimentTypes.fr:
+                        experiment.variations.append({
+                            'id': everyone_else_variation.id,
+                            'key': everyone_else_variation.key,
+                            'featureEnabled': everyone_else_variation.featureEnabled,
+                            'variables': cast(
+                                list[types.VariableDict],
+                                everyone_else_variation.variables,
+                            ),
+                        })
+                        experiment.trafficAllocation.append({
+                            'entityId': everyone_else_variation.id,
+                            'endOfRange': 10000,
+                        })
+                        self.variation_key_map[experiment.key][everyone_else_variation.key] = everyone_else_variation
+                        self.variation_id_map[experiment.key][everyone_else_variation.id] = everyone_else_variation
+                        self.variation_id_map_by_experiment_id[experiment.id][everyone_else_variation.id] = (
+                            everyone_else_variation
+                        )
+                        self.variation_key_map_by_experiment_id[experiment.id][everyone_else_variation.key] = (
+                            everyone_else_variation
+                        )
+                        self.variation_variable_usage_map[everyone_else_variation.id] = self._generate_key_map(
+                            everyone_else_variation.variables, 'id', entities.Variation.VariableUsage
+                        )
+
+            # Map all running holdouts to this flag
+            applicable_holdouts = list(self.holdout_id_map.values())
             if applicable_holdouts:
                 self.flag_holdouts_map[feature.key] = applicable_holdouts
 
@@ -667,6 +664,41 @@ class ProjectConfig:
         self.logger.error(f'Rollout with ID "{rollout_id}" is not in datafile.')
         return None
 
+    def _get_everyone_else_variation(self, flag: entities.FeatureFlag) -> Optional[entities.Variation]:
+        """ Get the "everyone else" variation for a feature flag.
+
+        The "everyone else" rule is the last experiment in the flag's rollout,
+        and its first variation is the "everyone else" variation.
+
+        Args:
+            flag: The feature flag to get the everyone else variation for.
+
+        Returns:
+            The "everyone else" Variation entity, or None if not available.
+        """
+        if not flag.rolloutId:
+            return None
+
+        rollout = self.get_rollout_from_id(flag.rolloutId)
+        if not rollout or not rollout.experiments:
+            return None
+
+        everyone_else_rule = rollout.experiments[-1]
+        variations = everyone_else_rule.get('variations', [])
+        if not variations:
+            return None
+
+        variation_dict = variations[0]
+        return entities.Variation(
+            id=variation_dict['id'],
+            key=variation_dict['key'],
+            featureEnabled=bool(variation_dict.get('featureEnabled', False)),
+            variables=cast(
+                Optional[list[entities.Variable]],
+                variation_dict.get('variables'),
+            ),
+        )
+
     def get_variable_value_for_variation(
         self, variable: Optional[entities.Variable], variation: Optional[Union[entities.Variation, VariationDict]]
     ) -> Optional[str]:

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely/version.py

@@ -11,5 +11,5 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-version_info = (5, 4, 0)
+version_info = (5, 5, 0)
 __version__ = '.'.join(str(v) for v in version_info)

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0/optimizely_sdk.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: optimizely-sdk
-Version: 5.4.0
+Version: 5.5.0
 Summary: Python SDK for Optimizely Feature Experimentation, Optimizely Full Stack (legacy), and Optimizely Rollouts.
 Home-page: https://github.com/optimizely/python-sdk
 Author: Optimizely
@@ -17,8 +17,34 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Description-Content-Type: text/markdown
-Provides-Extra: test
 License-File: LICENSE
+Requires-Dist: jsonschema>=3.2.0
+Requires-Dist: pyrsistent>=0.16.0
+Requires-Dist: requests>=2.21
+Requires-Dist: idna>=2.10
+Requires-Dist: rpds-py<0.20.0; python_version < "3.11"
+Provides-Extra: test
+Requires-Dist: requests>=2.21; extra == "test"
+Requires-Dist: idna>=2.10; extra == "test"
+Requires-Dist: ruff>=0.15.0; extra == "test"
+Requires-Dist: python-coveralls; extra == "test"
+Requires-Dist: coverage; extra == "test"
+Requires-Dist: pytest-cov; extra == "test"
+Requires-Dist: rpds-py<0.20.0; python_version < "3.11" and extra == "test"
+Requires-Dist: jsonschema>=3.2.0; extra == "test"
+Requires-Dist: funcsigs>=0.4; extra == "test"
+Requires-Dist: pytest>=6.2.0; extra == "test"
+Requires-Dist: pyrsistent>=0.16.0; extra == "test"
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: home-page
+Dynamic: license-file
+Dynamic: provides-extra
+Dynamic: requires-dist
+Dynamic: summary
 
 Optimizely Feature Experimentation is A/B testing and feature management for product development teams. Experiment in any application. Make every feature on your roadmap an opportunity to learn. Learn more at https://www.optimizely.com/products/experiment/feature-experimentation/ or see our documentation at https://docs.developers.optimizely.com/experimentation/v4.0.0-full-stack/docs/welcome. # Optimizely Python SDK
 
@@ -278,6 +304,34 @@ idna (BSD 3-Clause License https://github.com/kjd/idna/blob/master/LICENSE.md)
 - Swift - https://github.com/optimizely/swift-sdk
 # Optimizely Python SDK Changelog
 
+## 5.5.0
+April 30th, 2026
+
+### New Features
+
+#### **Feature Rollout Support**
+
+- Added support for Feature Rollouts, a new rule type that combines Targeted Delivery simplicity with A/B test measurement capabilities. During project config parsing, the "everyone else" variation from the flag's rollout is automatically injected into feature rollout experiments, enabling correct evaluation without changes to decision logic. ([#499](https://github.com/optimizely/python-sdk/pull/499))
+- Removed experiment type validation from config parsing to support flexible rule types. ([#500](https://github.com/optimizely/python-sdk/pull/500))
+
+---
+
+### Enhancements
+
+- Added event retry support for reliable event dispatching. ([#475](https://github.com/optimizely/python-sdk/pull/475))
+- Excluded CMAB experiments from UserProfileService to prevent stale cached decisions. ([#474](https://github.com/optimizely/python-sdk/pull/474))
+- Cleaned up flag-level holdout fields for simplified configuration. ([#507](https://github.com/optimizely/python-sdk/pull/507))
+
+---
+
+### Bug Fixes
+
+- Fixed `return` in `finally` block silently swallowing exceptions. ([#505](https://github.com/optimizely/python-sdk/pull/505))
+- Fixed `PollingConfigManager` rejecting valid URL-only configuration. ([#497](https://github.com/optimizely/python-sdk/pull/497))
+
+---
+
+
 ## 5.4.0
 December 19th, 2025
 

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/optimizely_sdk.egg-info/requires.txt

@@ -7,16 +7,16 @@ idna>=2.10
 rpds-py<0.20.0
 
 [test]
-coverage
-pyrsistent>=0.16.0
-idna>=2.10
-pytest>=6.2.0
-pytest-cov
-funcsigs>=0.4
 requests>=2.21
+idna>=2.10
+ruff>=0.15.0
 python-coveralls
+coverage
+pytest-cov
 jsonschema>=3.2.0
-flake8>=4.0.1
+funcsigs>=0.4
+pytest>=6.2.0
+pyrsistent>=0.16.0
 
 [test:python_version < "3.11"]
 rpds-py<0.20.0

{optimizely-sdk-5.4.0 → optimizely_sdk-5.5.0}/requirements/test.txt

@@ -1,5 +1,5 @@
 coverage
-flake8 >= 4.0.1
+ruff >= 0.15.0
 funcsigs >= 0.4
 pytest >= 6.2.0
 pytest-cov