CHANGELOG.md

Change log

All notable changes to the LaunchDarkly Ruby SDK will be documented in this file. This project adheres to Semantic Versioning.

[6.3.1] - 2021-12-31

Fixed:

• Fixed a bug that could cause a streaming connection to fail intermittently if the feature flag data contained UTF-8 characters outside of the ASCII character set. This would happen if a multi-byte character happened to be split across two successive reads from the stream, so the chances of it happening varied according to how often international characters appeared in the data and how much buffering of reads was done by the OS.
• In JRuby only, stream reconnections would fail if the application explicitly set the initial reconnect delay to zero.

[6.3.0] - 2021-12-09

Added:

• The SDK now supports evaluation of Big Segments. See: https://docs.launchdarkly.com/home/users/big-segments
• LaunchDarkly::Integrations::TestData is a new way to inject feature flag data programmatically into the SDK for testing—either with fixed values for each flag, or with targets and/or rules that can return different values for different users. Unlike FileData, this mechanism does not use any external resources, only the data that your test code has provided.

Changed:

• To use the file data source feature, the preferred entry point is now LaunchDarkly::Integrations::FileData.data_source rather than LaunchDarkly::FileDataSource.factory. This makes the Ruby SDK more consistent with other SDKs, grouping together all of the optional "connecting the SDK to something else" features under Integrations, and using the method name data_source for consistency with the Config property that will receive the object.

Deprecated:

• LaunchDarkly::FileDataSource (see above).

[6.2.5] - 2021-10-12

Fixed:

• Fixed a bug that caused unnecessarily heavy CPU usage when receiving very large sets of flag data from LaunchDarkly.
• Improved the speed of making the initial streaming connection to LaunchDarkly. The delay that happens before reconnecting after a connection failure was mistakenly being applied before the first connection.

[6.2.4] - 2021-08-11

Changed:

• The dependency version constraint for the http gem is now looser: it allows 5.x versions as well as 4.x. The breaking changes in http v5.0.0 do not affect the SDK. (#184)
• The dependency version constraint for the json gem is also looser: it allows any 2.x version that is higher than the SDK's minimum dependency version, not just 2.3. (#184)
• The project's build now uses v2.2.10 of bundler due to known vulnerabilities in other versions.

[6.2.3] - 2021-08-06

Fixed:

• Diagnostic events did not properly set the usingProxy attribute when a proxy was configured with the HTTPS_PROXY environment variable. (#182)

[6.2.2] - 2021-07-23

Fixed:

• Enabling debug logging in polling mode could cause polling to fail with a NameError. (Thanks, mmurphy-notarize!)

[6.2.1] - 2021-07-15

Changed:

• If variation or variation_detail is called with a user object that has no key (an invalid condition that will always result in the default value being returned), the SDK now logs a warn-level message to alert you to this incorrect usage. This makes the Ruby SDK's logging behavior consistent with the other server-side LaunchDarkly SDKs. (#177)

[6.2.0] - 2021-06-17

Added:

• The SDK now supports the ability to control the proportion of traffic allocation to an experiment. This works in conjunction with a new platform feature now available to early access customers.

[6.1.1] - 2021-05-27

Fixed:

• Calling variation with a nil user parameter is invalid, causing the SDK to log an error and return a fallback value, but the SDK was still sending an analytics event for this. An event without a user is meaningless and can't be processed by LaunchDarkly. This is now fixed so the SDK will not send one.

[6.1.0] - 2021-02-04

Added:

• Added the alias method. This can be used to associate two user objects for analytics purposes by generating an alias event.

[6.0.0] - 2021-01-26

Added:

• Added a socket_factory configuration option which can be used for socket creation by the HTTP client if provided. The value of socket_factory must be an object providing an open(uri, timeout) method and returning a connected socket.

Changed:

• Switched to the http gem instead of socketry (with a custom http client) for streaming, and instead of Net::HTTP for polling / events.
• Dropped support for Ruby < version 2.5
• Dropped support for JRuby < version 9.2
• Switched the default polling domain from app.launchdarkly.com to sdk.launchdarkly.com.

[5.8.2] - 2021-01-19

Fixed:

• Fixed a warning within the Redis integration when run with version 4.3 or later of the redis gem. (Thanks, emancu!)

[5.8.1] - 2020-11-09

Fixed:

• Updated json gem to patch CVE-2020-10663.

[5.8.0] - 2020-05-27

Added:

• In LaunchDarkly::Integrations::Redis::new_feature_store, if you pass in an externally created pool, you can now set the new option pool_shutdown_on_close to false to indicate that the SDK should not shut down this pool if the SDK is shut down. The default behavior, as before, is that it will be shut down. (Thanks, jacobthemyth!)

[5.7.4] - 2020-05-04

Fixed:

• Setting a user's custom property explicitly to nil, rather than omitting it entirely or setting it to an empty hash, would cause the SDK to log an error and drop the current batch of analytics events. Now, it will be treated the same as an empty hash. (#147)

[5.7.3] - 2020-04-27

Changed:

• Previously, installing the SDK in an environment that did not have openssl would cause a failure at build time. The SDK still requires openssl at runtime, but this check has been removed because it caused the rake problem mentioned below, and because openssl is normally bundled in modern Ruby versions.

Fixed:

• The LDClient constructor will fail immediately with a descriptive ArgumentError if you provide a nil SDK key in a configuration that requires an SDK key (that is, a configuration that will require communicating with LaunchDarkly services). Previously, it would still fail, but without a clear error message. You are still allowed to omit the SDK key in an offline configuration. (#154)
• Removed a hidden dependency on rake which could cause your build to fail if you had a dependency on this SDK and you did not have rake installed. (#155)
• Previously a clause in a feature flag rule that used a string operator (such as "starts with") or a numeric operator (such as "greater than") could cause evaluation of the flag to completely fail and return a default value if the value on the right-hand side of the expression did not have the right data type-- for instance, "greater than" with a string value. The LaunchDarkly dashboard does not allow creation of such a rule, but it might be possible to do so via the REST API; the correct behavior of the SDK is to simply treat the expression as a non-match.

[5.7.2] - 2020-03-27

Fixed:

• Fixed a bug in the 5.7.0 and 5.7.1 releases that caused analytics events not to be sent unless diagnostic events were explicitly disabled. This also caused an error to be logged: undefined method started?.

[5.7.1] - 2020-03-18

Fixed:

• The backoff delay logic for reconnecting after a stream failure was broken so that if a failure occurred after a stream had been active for at least 60 seconds, retries would use no delay, potentially causing a flood of requests and a spike in CPU usage. This bug was introduced in version 5.5.0 of the SDK.

[5.7.0] - 2020-03-10

Added:

• The SDK now periodically sends diagnostic data to LaunchDarkly, describing the version and configuration of the SDK, the architecture and version of the runtime platform, and performance statistics. No credentials, hostnames, or other identifiable values are included. This behavior can be disabled with Config.diagnostic_opt_out or configured with Config.diagnostic_recording_interval.
• New Config properties wrapper_name and wrapper_version allow a library that uses the Ruby SDK to identify itself for usage data if desired.

Removed:

• Removed an unused dependency on rake.

[5.6.2] - 2020-01-15

Fixed:

• The SDK now specifies a uniquely identifiable request header when sending events to LaunchDarkly to ensure that events are only processed once, even if the SDK sends them two times due to a failed initial attempt.

[5.6.1] - 2020-01-06

Fixed:

• In rare circumstances (depending on the exact data in the flag configuration, the flag's salt value, and the user properties), a percentage rollout could fail and return a default value, logging the error "Data inconsistency in feature flag ... variation/rollout object with no variation or rollout". This would happen if the user's hashed value fell exactly at the end of the last "bucket" (the last variation defined in the rollout). This has been fixed so that the user will get the last variation.

[5.6.0] - 2019-08-28

Added:

• Added support for upcoming LaunchDarkly experimentation features. See LDClient.track().

[5.5.12] - 2019-08-05

Fixed:

• Under conditions where analytics events are being generated at an extremely high rate (for instance, if an application is evaluating a flag repeatedly in a tight loop on many threads), it was possible for the internal event processing logic to fall behind on processing the events, causing them to use more and more memory. The logic has been changed to drop events if necessary so that besides the existing limit on the number of events waiting to be sent to LaunchDarkly (config.capacity), the same limit also applies on the number of events that are waiting to be processed by the worker thread that decides whether or not to send them to LaunchDarkly. If that limit is exceeded, this warning message will be logged once: "Events are being produced faster than they can be processed; some events will be dropped". Under normal conditions this should never happen; this change is meant to avoid a concurrency bottleneck in applications that are already so busy that thread starvation is likely.

[5.5.11] - 2019-07-24

Fixed:

• FileDataSource was using YAML.load, which has a known security vulnerability. This has been changed to use YAML.safe_load, which will refuse to parse any files that contain the ! directives used in this type of attack. This issue does not affect any applications that do not use FileDataSource (which is meant for testing purposes, not production use). (#139)

[5.5.10] - 2019-07-24

This release was an error; it is identical to 5.5.9.

[5.5.9] - 2019-07-23

Fixed:

• Due to the gem name no longer being the same as the require name, Bundler autoloading was no longer working in versions 5.5.7 and 5.5.8 of the SDK. This has been fixed. (Thanks, tonyta!)

[5.5.8] - 2019-07-11

Fixed:

• In streaming mode, depending on the Ruby version, calling close on the client could cause a misleading warning message in the log, such as Unexpected error from event source: #<IOError: stream closed in another thread>. (#135)

[5.5.7] - 2019-05-13

Changed:

• Changed the gem name from ldclient-rb to launchdarkly-server-sdk.

There are no other changes in this release. Substituting ldclient-rb version 5.5.6 with launchdarkly-server-sdk version 5.5.7 will not affect functionality.

[5.5.6] - 2019-05-08

Fixed:

• CI tests now include Ruby 2.6.x.
• Running the SDK unit tests is now simpler, as the database integrations can be skipped. See CONTRIBUTING.md.

Note on future releases

The LaunchDarkly SDK repositories are being renamed for consistency. This repository is now ruby-server-sdk rather than ruby-client.

The gem name will also change. In the 5.5.6 release, it is still ldclient-rb; in all future releases, it will be launchdarkly-server-sdk. No further updates to the ldclient-rb gem will be published after this release.

[5.5.5] - 2019-03-28

Fixed:

• Setting user attributes to non-string values when a string was expected would cause analytics events not to be processed. Also, in the case of the secondary attribute, this could cause evaluations to fail for a flag with a percentage rollout. The SDK will now convert attribute values to strings as needed. (#131)

[5.5.4] - 2019-03-29

Fixed:

• Fixed a missing require that could sometimes cause a NameError to be thrown when starting the client, depending on what other gems were installed. This bug was introduced in version 5.5.3. (#129)
• When an analytics event was generated for a feature flag because it is a prerequisite for another flag that was evaluated, the user data was being omitted from the event. (#128)
• If track or identify is called without a user, the SDK now logs a warning, and does not send an analytics event to LaunchDarkly (since it would not be processed without a user).
• Added a link from the SDK readme to the guide regarding the client initialization.

[5.5.3] - 2019-02-13

Changed:

• The SDK previously used the faraday and net-http-persistent gems for all HTTP requests other than streaming connections. Since faraday lacks a stable version and has a known issue with character encoding, and net-http-persistent is no longer maintained, these have both been removed. This should not affect any SDK functionality.

Fixed:

• The SDK was not usable in Windows because of net-http-persistent. That gem has been removed.
• When running in Windows, the event-processing thread threw a RangeError due to a difference in the Windows implementation of concurrent-ruby. This has been fixed.
• Windows incompatibilities were undetected before because we were not running a Windows CI job. We are now testing on Windows with Ruby 2.5.

[5.5.2] - 2019-01-18

Fixed:

• Like 5.5.1, this release contains only documentation fixes. Implementation classes that are not part of the supported API are now hidden from the generated documentation.

[5.5.1] - 2019-01-17

Fixed:

• Fixed several documentation comments that had the wrong parameter names. There are no other changes in this release; it's only to correct the documentation.

[5.5.0] - 2019-01-17

Added:

• It is now possible to use Consul or DynamoDB as a persistent feature store, similar to the existing Redis integration. See the LaunchDarkly::Integrations::Consul and LaunchDarkly::Integrations::DynamoDB modules, and the reference guide Persistent data stores.
• There is now a LaunchDarkly::Integrations::Redis module, which is the preferred method for creating a Redis feature store.
• All of the database feature stores now support local caching not only for individual feature flag queries, but also for all_flags_state.
• The Config property data_source is the new name for update_processor and update_processor_factory.

Changed:

• The implementation of the SSE protocol for streaming has been moved into a separate gem, ld-eventsource. This has no effect on streaming functionality.

Fixed:

• Added or corrected a large number of documentation comments. All API classes and methods are now documented, and internal implementation details have been hidden from the documentation. You can view the latest documentation on RubyDoc.
• Fixed a problem in the Redis feature store that would only happen under unlikely circumstances: trying to evaluate a flag when the LaunchDarkly client had not yet been fully initialized and the store did not yet have data in it, and then trying again when the client was still not ready but the store did have data (presumably put there by another process). Previously, the second attempt would fail.
• In polling mode, the SDK did not correctly handle non-ASCII Unicode characters in feature flag data. (#90)

Deprecated:

• RedisFeatureStore.new. This implementation class may be changed or moved in the future; use LaunchDarkly::Integrations::Redis::new_feature_store.
• Config.update_processor and Config.update_processor_factory; use Config.data_source.

[5.4.3] - 2019-01-11

Changed:

• The SDK is now compatible with net-http-persistent 3.x. (Thanks, CodingAnarchy!)

[5.4.2] - 2019-01-04

Fixed:

• Fixed overly specific dependency versions of concurrent-ruby and semantic. (#115)
• Removed obsolete dependencies on hashdiff and thread_safe.

[5.4.1] - 2018-11-05

Fixed:

• Fixed a LoadError in file_data_source.rb, which was added in 5.4.0. (Thanks, kbarrette!)

[5.4.0] - 2018-11-02

Added:

• It is now possible to inject feature flags into the client from local JSON or YAML files, replacing the normal LaunchDarkly connection. This would typically be for testing purposes. See file_data_source.rb.

Fixed:

• When shutting down an LDClient, if in polling mode, the client was using Thread.raise to make the polling thread stop sleeping. Thread.raise can cause unpredictable behavior in a worker thread, so it is no longer used.

[5.3.0] - 2018-10-24

Added:

• The all_flags_state method now accepts a new option, details_only_for_tracked_flags, which reduces the size of the JSON representation of the flag state by omitting some metadata. Specifically, it omits any data that is normally used for generating detailed evaluation events if a flag does not have event tracking or debugging turned on.

Fixed:

• JSON data from all_flags_state is now slightly smaller even if you do not use the new option described above, because it omits the flag property for event tracking unless that property is true.

[5.2.0] - 2018-08-29

Added:

• The new LDClient method variation_detail allows you to evaluate a feature flag (using the same parameters as you would for variation) and receive more information about how the value was calculated. This information is returned in an EvaluationDetail object, which contains both the result value and a "reason" object which will tell you, for instance, if the user was individually targeted for the flag or was matched by one of the flag's rules, or if the flag returned the default value due to an error.

Fixed:

• Evaluating a prerequisite feature flag did not produce an analytics event if the prerequisite flag was off.

[5.1.0] - 2018-08-27

Added:

• The new LDClient method all_flags_state() should be used instead of all_flags() if you are passing flag data to the front end for use with the JavaScript SDK. It preserves some flag metadata that the front end requires in order to send analytics events correctly. Versions 2.5.0 and above of the JavaScript SDK are able to use this metadata, but the output of all_flags_state() will still work with older versions.
• The all_flags_state() method also allows you to select only client-side-enabled flags to pass to the front end, by using the option client_side_only: true.

Changed:

• Unexpected exceptions are now logged at ERROR level, and exception stacktraces at DEBUG level. Previously, both were being logged at WARN level.

Deprecated:

• LDClient.all_flags()

[5.0.1] - 2018-07-02

Fixed:

Fixed a regression in version 5.0.0 that could prevent the client from reconnecting if the stream connection was dropped by the server.

[5.0.0] - 2018-06-26

Changed:

• The client no longer uses Celluloid for streaming I/O. Instead, it uses socketry.
• The client now treats most HTTP 4xx errors as unrecoverable: that is, after receiving such an error, it will not make any more HTTP requests for the lifetime of the client instance, in effect taking the client offline. This is because such errors indicate either a configuration problem (invalid SDK key) or a bug, which is not likely to resolve without a restart or an upgrade. This does not apply if the error is 400, 408, 429, or any 5xx error.
• During initialization, if the client receives any of the unrecoverable errors described above, the client constructor will return immediately; previously it would continue waiting until a timeout. The initialized? method will return false in this case.

Removed:

• The SDK no longer supports Ruby versions below 2.2.6, or JRuby below 9.1.16.

[4.0.0] - 2018-05-10

Changed:

• To reduce the network bandwidth used for analytics events, feature request events are now sent as counters rather than individual events, and user details are now sent only at intervals rather than in each event. These behaviors can be modified through the LaunchDarkly UI and with the new configuration option inline_users_in_events.

Removed:

• JRuby 1.7 is no longer supported.
• Greatly reduced the number of indirect gem dependencies by removing moneta, which was previously a requirement for the Redis feature store.

[3.0.3] - 2018-03-23

Fixed

• In the Redis feature store, fixed a synchronization problem that could cause a feature flag update to be missed if several of them happened in rapid succession.

[3.0.2] - 2018-03-06

Fixed

• Improved efficiency of logging by not constructing messages that won't be visible at the current log level. (Thanks, julik!)

[3.0.1] - 2018-02-26

Fixed

• Fixed a bug that could prevent very large feature flags from being updated in streaming mode.

[3.0.0] - 2018-02-22

Added

• Support for a new LaunchDarkly feature: reusable user segments.

Changed

• The feature store interface has been changed to support user segment data as well as feature flags. Existing code that uses InMemoryFeatureStore or RedisFeatureStore should work as before, but custom feature store implementations will need to be updated.

[2.5.0] - 2018-02-12

Added

• Adds support for a future LaunchDarkly feature, coming soon: semantic version user attributes.

Changed

• It is now possible to compute rollouts based on an integer attribute of a user, not just a string attribute.

[2.4.1] - 2018-01-23

Changed

• Reduce logging level for missing flags
• Relax json and faraday dependencies

Fixed

• Wrap redis bulk updates in a transaction
• Fixed documentation links

[2.4.0] - 2018-01-12

Changed

• Will use feature store if already initialized even if connection to service could not be established. This is useful when flags have been initialized in redis.
• Increase default and minimum polling interval to 30s
• Strip out unknown top-level attributes

[2.3.2] - 2017-12-02

Fixed

• Make sure redis store initializations are atomic

[2.3.1] - 2017-11-16

Changed

• Include source code for changes described in 2.3.0

[2.3.0] - 2017-11-16

Added

• Add close method to Ruby client to stop processing events
• Add support for Redis feature store
• Add support for LDD mode
• Allow user to disable outgoing event stream.

Changed

• Stop retrying on 401 responses (due to bad sdk keys)

[2.2.7] - 2017-07-26

Changed

• Update Readme to fix instructions on installing gem using command line
• Cleaned up formatting on various files (Rubocop)

[2.2.5] - 2017-05-08

Changed

• Added proxy support to streaming and http connections. Respects HTTP_PROXY and http_proxy environment variables as well as the :proxy => protocol://user:pass@host configuration parameter.

[2.1.5] - 2017-03-28

Changed

• Updated changelog

[2.1.1] - 2017-03-28

Changed

• Bumped nio4r to 2.0

[2.0.6] - 2017-02-10

Changed

• Improved handling of http status codes that may not be integers.

[2.0.5] - 2017-01-31

Changed

• Improved error handling when connected to flag update stream.

[2.0.3] - 2016-10-21

Fixed

• Indirect stream events are now correctly processed

[2.0.2] - 2016-08-08

Changed

• The default logger now logs at info level

[2.0.0] - 2016-08-08

Added

• Support for multivariate feature flags. In addition to booleans, feature flags can now return numbers, strings, dictionaries, or arrays via the variation method.
• New all_flags method returns all flag values for a specified user.
• If streaming is disabled, the client polls for feature flag changes. If streaming is disabled, the client will default to polling LaunchDarkly every second for updates. The poll interval is configurable via poll_interval.
• New secure_mode_hash function computes a hash suitable for the new LaunchDarkly JavaScript client's secure mode feature.
• Support for extremely large feature flags. When a large feature flag changes, the stream will include a directive to fetch the updated flag.

Changed

• You can now initialize the LaunchDarkly client with an optional timeout (specified in seconds). This will block initialization until the client has finished bootstrapping and is able to serve feature flags.
• The streaming implementation (StreamProcessor) uses Celluloid under the hood instead of EventMachine. The dependency on EventMachine has been removed.
• The store option has been renamed to cache_store.
• Offline mode can no longer be set dynamically. Instead, at configuration time, the offline parameter can be set to put the client in offline mode. It is no longer possible to dynamically change whether the client is online and offline (via set_online and set_offline). Call offline? to determine whether or not the client is offline.
• The debug_stream configuration option has been removed.
• The log_timings configuration option has been removed.

Deprecated

• The toggle call has been deprecated in favor of variation.

Removed

• update_user_flag_setting has been removed. To change user settings, use the LaunchDarkly REST API.