Skip to content

Latest commit

 

History

History
222 lines (137 loc) · 6.96 KB

CHANGELOG.md

File metadata and controls

222 lines (137 loc) · 6.96 KB
Raw

v5.2.0

  • (Minor) Add support for bootstrapping UnleashClient with an initial configuration.
  • (Minor) Add flag to disable toggle polling from server. Thanks @povilasb!

v5.1.2

  • (Bugfix) Clarify logging if Unleash server doesn't return feature provisioning (i.e. HTTP 304).

v5.1.1

  • (Minor) Move UnleashClient logging to the UnleashClient logger.
  • (Bugfix) Fix logging for UnleashClient.

v5.1.0

  • (Major) Support new constraint operators.
  • (Major) Add cache abstraction. Thanks @walison17!
  • (Minor) Refactor unleash-client-python to modernize tooling (setuptools_scm and centralizing tool config in pyproject.toml).
  • (Minor) Migrate documentation to Sphinx.

v5.0.1

  • (Bugfix) Fix HTTP 304 handling on /client/feature for Gitlab. Thanks @janLo!

v5.0.0

  • (Breaking) Modify client initialization to allow jitter configuration on refresh/background refresh intervals. May break code when parameters to UnleashClient are not used as keyword arguments. Thanks @dryobates and @jstolarski!
  • (Major) If client is already initialized, calling initialize_client() again won't re-run initialization.
  • (Minor) Support HTTP 304 on /client/feature endpoint.
  • (Minor) Rename master to main.
  • (Documentation) Document running UnleashClient in uWSGI. Thanks @sighphyre!
  • (Documentation) Fix links and formatting. Thanks @thomasheartman & @sighphyre!

v4.4.1

  • (Minor) Include py.typed to mark package as type-friendly! Thanks @wbolster!
  • (Minor) Fix API url sanitization. Thanks @romulorosa!

v4.4.0

  • (Minor) Support running Unleash client as a context manager. Thanks @Piojo !

v4.3.0

  • (Minor) initialize_client() will raise exception if UnleashClient is configured with an invalid URL.
  • (Minor) Exclude test package from dist & wheel. Thanks @ameyajoshi99!
  • (Minor) Allow users to specify log-level for when is_enabled() or get_varients() calls fail.

v4.2.0

  • (Minor) Support custom stickiness for FlexibleRollout strategy and variants.

v4.1.0

  • (Minor) Support project-based feature flag loading.

v4.0.0

  • (Major) Deprecate the default_value argument in the is_enabled() method.
  • (Major) Drop Python 3.5 support.
  • (Minor) Remove dependencies versions constraints. Thanks @wbolster and @isra17!
  • (Bugfix) Don't use mutable defaults. Thanks @aviau!

v3.6.2

  • (Minor) Only send metrics to API if feature toggle is in-use (i.e. has been resolved to True/False). Thanks @fwpheckel!
  • (Minor) Remove dangling variations reference in favor of variants verbiage.

v3.6.1

  • (Major) Fix bug where loader didn't properly refresh variants. Thanks @simenaasland!

v3.6.0

  • (Minor) Add Python 3.9 support.
  • (Minor) Only log errors generated when strategy loading fails once.
  • (Minor) Errors submitting metrics will be logged as warnings and not exceptions.
  • (Minor) Update apscheduler version to 3.7.0

v3.5.1

  • (Minor) Better error handling and typo fixes. Thanks @vgerak!
  • (Minor) Update requests version to 2.25.1.

v3.5.0

  • (Major) Stop using the default_value argument in the is_enabled() method (as it can cause counter-intuitive behavior) and add deprecation warning. This argument will be removed in the next major version upgrade!
    • We recommend using the fallback_function argument instead. If you need a blanket True in case of an exception, you can pass in a lambda like: lambda x, y: True.
  • (Minor) Add better logging for API errors.
  • (Minor) Update requests version to v2.25.0.

v3.4.1, v3.4.2

General

  • (Minor) Move CI to Github Actions, add auto-publishing.

v3.4.0

Bugfixes

  • (Major) Fallback function will only be called if exception (feature flag not found, general exception) occurs when calling is_enabled(). It will not be called on successful execution of the method.

v3.3.0

General

  • (Major) Add support for variants on feature toggles.

Bugfixes

  • (Minor) Fixed issue with applying custom constraints to non-standard parameters in context.

v3.2.0

General

  • (Major) Allow users to supply a fallback function to customize the default value of a feature flag.

v3.1.1

Bugfixes

v3.1.0

General

  • (Minor) Add official-ish support for Python 3.8.

v3.0.0

General

  • (Major) Support constraints on all default strategies.
  • (Major) Added flexibleRollout strategy.

v2.6.0

General

  • (Minor) Add ability to add request kwargs when initializing the client. These will be used when registering the client, fetching feature flags, and sending metrics.

v2.5.0

General

  • (Minor) Unleash client will not error if cache is not present and Unleash server not accessible during initialization.

v2.4.0

General

  • (Minor) Added static context values (app name, env) in preparation for Unleash v4 features.

v2.3.0

General

  • (Minor) Add option to disable metrics on client initialization.

Bugfix

  • (Minor) Fixed issue where disable_metrics arugment wasn't honored.

v2.2.1

Bugfixes

  • (Major) Date/time sent to Unleash (in register, metrics, etc) is correctly in UTC w/timestamp format.

v2.2.0

  • Allow configuration of the cache directory.

v2.1.0

General

  • (Major) Support for Python 3.5, 3.6, and 3.7. (Credit to Baaym for 3.5 support!)

v2.0.1

Bugfixes

  • (Major) Fix issue where bucket.start value sent to Unleash was never updated. Credit to Calle for bug report/proposed solution! =)

v2.0.0

Bugfixes

v1.0.2

General

  • unleash-client-python has moved under the general Unleash project!

Bugfixes

  • (Minor) Updated requests version to address security issue in dependency.

v1.0.0

General

  • Implemented custom strategies.

v0.3.0

General

Bugfixes

  • (Major) Fixed interposed arguments in normalized_hash() (aka MurmerHash3 wrapper). Python client will now do the same thing as the other clients!
  • (Major) Fixed issues with logic in random strategies.

v0.2.0

General

  • Changed cache implementation. Instead of caching {feature toggle name: provisioning} we'll now cache the entire API response (and use it if the fetch fails in any way).

v0.1.1

General

  • Fixed Github link on pypi.
  • Removed unused sphinx documentation.
  • Added documentation using mkdocs

v0.1.0

General

  • First implementation of the Unleash Python client! Woo!