ldclient-rb nearly burnt my Mac due to CPU throttling #148
Comments
|
Do I understand correctly that it was repeating those connection attempts with no delays in between? Like, do you have a larger sample of the log with timestamps? The SDK is not supposed to retry failed connections immediately, and we haven't seen it behave in this way. I also can't think of any reason that the specific type of network failure (DNS query failing, vs. network not being available at all even though wifi is active) would make any difference to its behavior; those are both just I/O exceptions as far as it's concerned. (By the way, 5.5.6 is not the latest version of the SDK; it is just the last version that was released under the gem name |
|
I would also be curious to know whether you've been able to reproduce this behavior after the initial incident. |
|
I guess I don't have that log. It was looking a following way - rails console was opened and log was scrolling like crazy with that same error 🤷♂ From processes it was ruby only throttling, other were ok. Overall I guess surely it is not very common, but just keep an eye on this, maybe add some non-linear safeguards somewhere, e.g. like sidekiq do 1, 2, 5, 15, 30 etc. |
Well, that's why I'm having some trouble understanding how this could happen: we do explicitly use exponential backoff logic. So if there is some way to make it skip that logic, I don't know what it is (which is why I was asking if you've been able to reproduce the problem). |
|
We did find a bug that could cause the delay to be skipped, and it was fixed in the 5.7.1 release. At that time we forgot to check against open issues so I did not flag it as a fix for this issue, but I'm fairly sure that it's the same problem you ran into. The reason I didn't manage to reproduce it earlier, I think, was that it would not happen if you interrupted the connection too soon after starting the SDK— only if the app had bee running for a little while. So I'm going to close this issue, but please feel free to reopen it if you think that the fix did not address your situation. |
LaunchDarklyCI
pushed a commit
that referenced
this issue
May 27, 2021
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Mar 18, 2022
* fix doc comments * add YARD config so our docs show up correctly everywhere * don't need markup-provider option * rm obsolete proxy param * remove net-http-persistent * fix concurrent-ruby usage that breaks on Windows * add pipeline and clean up with with rm_rf instead of rm * fix highlight blocks * Hr/azure3 (#103) * Add Consul and Redis services to Windows. * Enable Consul and Redis testing * add dynamo (#104) * add experimentation event overrides for rules and fallthrough * warn & don't send event if identify or track has no valid user * include user in prereq flag events * rm unnecessary logic * more factory methods * update readme to refer to docs * add Ruby 2.6.2 to CI * fix missing require for net/http * stringify built-in user attributes in events, and secondary key for evals * make const names consistent * support metric value with track() * update method description * applying markdown templates and updating repository url references * Cleaning up markdown files * allow skipping database tests * Updating the package name (#115) * update package name * missed one * revert module entry point name change * bump ld-eventsource version for stream logging fix * use YAML.safe_load * add unit test and temporarily revert fix to demonstrate failure * restore fix * add comment about not using FileDataSource in production * drop events if inbox is full * update doc comment for track with metric_value * don't let user fall outside of last bucket in rollout * refactor evaluation logic and move it out of the main namespace * comments * fix type coercion behavior * make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: Harpo roeder <hroeder@launchdarkly.com> Co-authored-by: hroederld <46500128+hroederld@users.noreply.github.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com>
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Jun 15, 2022
* rm obsolete proxy param * remove net-http-persistent * fix concurrent-ruby usage that breaks on Windows * add pipeline and clean up with with rm_rf instead of rm * fix highlight blocks * Hr/azure3 (#103) * Add Consul and Redis services to Windows. * Enable Consul and Redis testing * add dynamo (#104) * add experimentation event overrides for rules and fallthrough * warn & don't send event if identify or track has no valid user * include user in prereq flag events * rm unnecessary logic * more factory methods * update readme to refer to docs * add Ruby 2.6.2 to CI * fix missing require for net/http * stringify built-in user attributes in events, and secondary key for evals * make const names consistent * support metric value with track() * update method description * applying markdown templates and updating repository url references * Cleaning up markdown files * allow skipping database tests * Updating the package name (#115) * update package name * missed one * revert module entry point name change * bump ld-eventsource version for stream logging fix * use YAML.safe_load * add unit test and temporarily revert fix to demonstrate failure * restore fix * add comment about not using FileDataSource in production * drop events if inbox is full * update doc comment for track with metric_value * don't let user fall outside of last bucket in rollout * refactor evaluation logic and move it out of the main namespace * comments * fix type coercion behavior * make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: Harpo roeder <hroeder@launchdarkly.com> Co-authored-by: hroederld <46500128+hroederld@users.noreply.github.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com>
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Jun 30, 2022
* remove net-http-persistent * fix concurrent-ruby usage that breaks on Windows * add pipeline and clean up with with rm_rf instead of rm * fix highlight blocks * Hr/azure3 (#103) * Add Consul and Redis services to Windows. * Enable Consul and Redis testing * add dynamo (#104) * add experimentation event overrides for rules and fallthrough * warn & don't send event if identify or track has no valid user * include user in prereq flag events * rm unnecessary logic * more factory methods * update readme to refer to docs * add Ruby 2.6.2 to CI * fix missing require for net/http * stringify built-in user attributes in events, and secondary key for evals * make const names consistent * support metric value with track() * update method description * applying markdown templates and updating repository url references * Cleaning up markdown files * allow skipping database tests * Updating the package name (#115) * update package name * missed one * revert module entry point name change * bump ld-eventsource version for stream logging fix * use YAML.safe_load * add unit test and temporarily revert fix to demonstrate failure * restore fix * add comment about not using FileDataSource in production * drop events if inbox is full * update doc comment for track with metric_value * don't let user fall outside of last bucket in rollout * refactor evaluation logic and move it out of the main namespace * comments * fix type coercion behavior * make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: Harpo roeder <hroeder@launchdarkly.com> Co-authored-by: hroederld <46500128+hroederld@users.noreply.github.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com>
eli-darkly
added a commit
that referenced
this issue
Sep 7, 2022
* Hr/azure3 (#103) * Add Consul and Redis services to Windows. * Enable Consul and Redis testing * add dynamo (#104) * add experimentation event overrides for rules and fallthrough * warn & don't send event if identify or track has no valid user * include user in prereq flag events * rm unnecessary logic * more factory methods * update readme to refer to docs * add Ruby 2.6.2 to CI * fix missing require for net/http * stringify built-in user attributes in events, and secondary key for evals * make const names consistent * support metric value with track() * update method description * applying markdown templates and updating repository url references * Cleaning up markdown files * allow skipping database tests * Updating the package name (#115) * update package name * missed one * revert module entry point name change * bump ld-eventsource version for stream logging fix * use YAML.safe_load * add unit test and temporarily revert fix to demonstrate failure * restore fix * add comment about not using FileDataSource in production * drop events if inbox is full * update doc comment for track with metric_value * don't let user fall outside of last bucket in rollout * refactor evaluation logic and move it out of the main namespace * comments * fix type coercion behavior * make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) * Add polling support for contract test service (#198) * Add windows tests in circleci (#199) At some point in the past, we were experimenting with using Azure to verify Window builds. Now that CircleCI supports Windows, we should keep everything on a single CI provider. * Add application info support (#194) * reuse EvaluationDetail instances by precomputing results Co-authored-by: hroederld <46500128+hroederld@users.noreply.github.com> Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: hroederld <hroeder@launchdarkly.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> Co-authored-by: Louis Chan <lchan@launchdarkly.com>
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Dec 30, 2022
* don't let user fall outside of last bucket in rollout * refactor evaluation logic and move it out of the main namespace * comments * fix type coercion behavior * make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) * Drop support for EOL ruby versions (#196) Ruby 2.5 was EOL 2021-04-05 As of June 27th, 2022, the latest jRuby is Ruby 2.6 compatible. * Remove alias support (#195) * Add polling support for contract test service (#198) * Update rubocop and enable in CI (#197) Several of the Rubocop cop definitions have been renamed or moved to entirely other gems. This brings the configuration up to date with the latest naming conventions. * Add windows tests in circleci (#199) At some point in the past, we were experimenting with using Azure to verify Window builds. Now that CircleCI supports Windows, we should keep everything on a single CI provider. * Add application info support (#194) * reuse EvaluationDetail instances by precomputing results * rubocop reformatting * add super constructor calls * disable rubocop Rails rules and fix some remaining syntax offenses * fix super calls * Add big segment support to contract tests (#201) * Initial creation of LDContext (#206) This introduces the initial structure and usage of the LDContext class. Instances of this class are expected to be created through static factory methods: ```ruby LaunchDarkly::LDContext.create({...}) LaunchDarkly::LDContext.create_multi([...]) ``` This class is not completed yet. Rather, this initial commit is focused on the creation patterns and the most basic operations. Subsequent commits will continue fleshing out this class and its operation. The `get_value` method will see significant changes as we introduce attribute reference support. Its current more simplistic implementation exists only to serve some interim unit tests. * Add reference based value retrieval (#207) This commit introduces the References type used for targeting complex attributes in the new LDContexts. References are expected to be created through static factory methods: ```ruby LaunchDarkly::Reference.create("/a/b") LaunchDarkly::Reference.create_literal("/a/b") ``` These references can be used to retrieve values from an existing LDContext ```ruby ref = LaunchDarkly::Reference.create("/a/b") result = context.get_value_for_reference(ref) ``` * Basic changes to use contexts in evaluations instead of users (#208) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. This replaces `LDUser` with `LDContext` in the parameters for evaluations, and makes the minimum necessary adjustments to allow evaluations to keep working as before as long as the context kind is "user". None of the new behavior defined in the U2C spec is implemented yet. Generation of evaluation events is temporarily disabled because the event logic hasn't been updated yet. U2C contract tests for evaluations are partially enabled; a lot of functionality is still missing, but all the tests that only cover previously-existing evaluation behavior are passing. [pr]: launchdarkly/php-server-sdk-private#103 * Support ContextKind in Clauses (#209) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. The main features of this commit are: - introduction of `individual_context` and `individual_context_count` methods - context kind matching in clauses [pr]: launchdarkly/php-server-sdk-private#108 * Support included / excluded contexts in segments (#210) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. Segments are now able to provide `includedContext` and `excludedContext` properties which can target values within a specific context kind. ```json { "includedContexts": [ { "contextKind": "org", "values": ["orgkey1", "orgkey2"] } ] } ``` [pr]: launchdarkly/php-server-sdk-private#111 * Add contextKind support for rollouts & experiements (#211) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. [pr]: launchdarkly/php-server-sdk-private#110 * Style and test matcher improvements (#212) This commit enables several rubocop rules that were previously disabled. Once enabled, `rubocop -A` was run to automatically apply these fixes. There are a couple of additional changes that were made by hand: - I added the rubocop and rubocop-performance gems as dev packages. This should help address the original installation issue we ran into when I introduced these tools. - By default, new rubocop rules are disabled. This was the default before, but if you don't explicitly set this value, each run generates a ton of warning noise. This quiets that down. - Updates some LDContext tests to be more strict in their expectations of truth. * Remove support for secondary attribute (#213) As decided in the [spec], we are removing the special behavior of the secondary attribute. Going forward, secondary will be treated like any other attribute, and will no longer be included when determining the bucket for a context. [spec]: https://launchdarkly.atlassian.net/wiki/spaces/ENG/pages/2165212563/Consistent+and+Transparent+Rollout+Behavior+Unifying+Percent+Rollout+and+Traffic+Allocation * Remove deprecated APIs (#214) Since the users to context change requires a version break, this is the perfect time to remove previously deprecated bits of functionality. This includes: - Removing `update_processor*` config entries - `FileDataSource` entry point - `RedisFeatureStore` entry point - `Redis::sadd?` warning in unit tests * store data model with classes that aren't Hash * lint * remove [] override methods in places where we don't need them * comments * migrate some more of the model to be non-hash classes * lint * Anonymous cannot be nil in new context format (#216) The legacy user format allowed anonymous to be missing or explicitly provided but set to nil. The new context format requires anonymous to either not be set, or if it is explicitly set, it must be a boolean value. * Tweak error message language and style (#217) Our previous error messages suffered from a couple drawbacks: - The messages were complete sentences, limiting our ability to compose error messages - The messages were overly broad in many cases - The messages unnecessarily required string interpolation that rarely provided much value These new messages are more succinct and are written as small clauses which can be used in conjunction with other error messages more easily. * copyedit Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> * Implement prerequisite cycle detection (#219) * Support attribute reference lookups (#215) This adds support for slash-delimited paths in clause attributes, bucketBy, etc. It does not do anything related to private attribute redaction because none of the U2C event logic is implemented yet. * Implement segment recursion and cycle detection (#220) Clauses in segment rules are now allowed to reference segments. To prevent an infinite recursion edge case, we implement a similar cycle detection mechanism as used on prerequisites. * Update event logic to support users to context change (#221) * Add legacy user-type support to the contract tests (#222) * Remove inline user configuration option (#223) * Add context_ configuration options (#224) These new context_ configuration options are meant to replace the historic user_ options. If both are provided, the context_ variant will take precedence. * Add support for flag context targets (#225) * Bump diplomat * Bump redis * Remove oga * Bump connection_pool * Favor set for faster target lookups (#228) A few of our internal models maintain arrays of values. These arrays can frequently be checked to see if they contain specific values. Since set lookups are much faster than array lookups, this commit changes the internal structure to a set for the values stored in Target and SegmentTarget. * Add secure mode hash to contract tests (#229) * Update big segment support for users to context (#226) To support the users to context change for big segments, this commit makes the following changes: - Introduces a new `Segment.unboundedContextKind` attribute. This will default to `LDContext::KIND_DEFAULT` and is only referenced when `Segment.unbounded` is true. - With the creation of multi-kind contexts, a single evaluation may result in multiple queries to the big segment store. This is reflected in the changes to the `EvalResult` processing. * Drop support for ruby 2.6 (#227) Ruby 2.6 went EOL in March 2022. We originally didn't drop support for it as doing so would require dropping support for jRuby as well. However, jRuby recently released 9.4 which is Ruby 2.7+ compatible. * Update remaining references from user to contexts (#231) There are multiple places throughout the code where we are still referencing users. I have tried to update all the places where a rename seems reasonable or appropriate. There is still some work to do in the test flag builders, but that will be done in a subsequent commit. * Remove new relic integration (#233) The new relic integration was removed many versions ago but a small trace remained behind. * Rename config option private_attribute_names (#234) Co-authored-by: Eli Bishop <eli@launchdarkly.com> * Update test data integration to support contexts (#232) Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: hroederld <hroeder@launchdarkly.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> Co-authored-by: Louis Chan <lchan@launchdarkly.com>
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Jan 19, 2023
* don't let user fall outside of last bucket in rollout * refactor evaluation logic and move it out of the main namespace * comments * fix type coercion behavior * make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) * Drop support for EOL ruby versions (#196) Ruby 2.5 was EOL 2021-04-05 As of June 27th, 2022, the latest jRuby is Ruby 2.6 compatible. * Remove alias support (#195) * Add polling support for contract test service (#198) * Update rubocop and enable in CI (#197) Several of the Rubocop cop definitions have been renamed or moved to entirely other gems. This brings the configuration up to date with the latest naming conventions. * Add windows tests in circleci (#199) At some point in the past, we were experimenting with using Azure to verify Window builds. Now that CircleCI supports Windows, we should keep everything on a single CI provider. * Add application info support (#194) * reuse EvaluationDetail instances by precomputing results * rubocop reformatting * add super constructor calls * disable rubocop Rails rules and fix some remaining syntax offenses * fix super calls * Add big segment support to contract tests (#201) * Initial creation of LDContext (#206) This introduces the initial structure and usage of the LDContext class. Instances of this class are expected to be created through static factory methods: ```ruby LaunchDarkly::LDContext.create({...}) LaunchDarkly::LDContext.create_multi([...]) ``` This class is not completed yet. Rather, this initial commit is focused on the creation patterns and the most basic operations. Subsequent commits will continue fleshing out this class and its operation. The `get_value` method will see significant changes as we introduce attribute reference support. Its current more simplistic implementation exists only to serve some interim unit tests. * Add reference based value retrieval (#207) This commit introduces the References type used for targeting complex attributes in the new LDContexts. References are expected to be created through static factory methods: ```ruby LaunchDarkly::Reference.create("/a/b") LaunchDarkly::Reference.create_literal("/a/b") ``` These references can be used to retrieve values from an existing LDContext ```ruby ref = LaunchDarkly::Reference.create("/a/b") result = context.get_value_for_reference(ref) ``` * Basic changes to use contexts in evaluations instead of users (#208) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. This replaces `LDUser` with `LDContext` in the parameters for evaluations, and makes the minimum necessary adjustments to allow evaluations to keep working as before as long as the context kind is "user". None of the new behavior defined in the U2C spec is implemented yet. Generation of evaluation events is temporarily disabled because the event logic hasn't been updated yet. U2C contract tests for evaluations are partially enabled; a lot of functionality is still missing, but all the tests that only cover previously-existing evaluation behavior are passing. [pr]: launchdarkly/php-server-sdk-private#103 * Support ContextKind in Clauses (#209) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. The main features of this commit are: - introduction of `individual_context` and `individual_context_count` methods - context kind matching in clauses [pr]: launchdarkly/php-server-sdk-private#108 * Support included / excluded contexts in segments (#210) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. Segments are now able to provide `includedContext` and `excludedContext` properties which can target values within a specific context kind. ```json { "includedContexts": [ { "contextKind": "org", "values": ["orgkey1", "orgkey2"] } ] } ``` [pr]: launchdarkly/php-server-sdk-private#111 * Add contextKind support for rollouts & experiements (#211) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. [pr]: launchdarkly/php-server-sdk-private#110 * Style and test matcher improvements (#212) This commit enables several rubocop rules that were previously disabled. Once enabled, `rubocop -A` was run to automatically apply these fixes. There are a couple of additional changes that were made by hand: - I added the rubocop and rubocop-performance gems as dev packages. This should help address the original installation issue we ran into when I introduced these tools. - By default, new rubocop rules are disabled. This was the default before, but if you don't explicitly set this value, each run generates a ton of warning noise. This quiets that down. - Updates some LDContext tests to be more strict in their expectations of truth. * Remove support for secondary attribute (#213) As decided in the [spec], we are removing the special behavior of the secondary attribute. Going forward, secondary will be treated like any other attribute, and will no longer be included when determining the bucket for a context. [spec]: https://launchdarkly.atlassian.net/wiki/spaces/ENG/pages/2165212563/Consistent+and+Transparent+Rollout+Behavior+Unifying+Percent+Rollout+and+Traffic+Allocation * Remove deprecated APIs (#214) Since the users to context change requires a version break, this is the perfect time to remove previously deprecated bits of functionality. This includes: - Removing `update_processor*` config entries - `FileDataSource` entry point - `RedisFeatureStore` entry point - `Redis::sadd?` warning in unit tests * store data model with classes that aren't Hash * lint * remove [] override methods in places where we don't need them * comments * migrate some more of the model to be non-hash classes * lint * Anonymous cannot be nil in new context format (#216) The legacy user format allowed anonymous to be missing or explicitly provided but set to nil. The new context format requires anonymous to either not be set, or if it is explicitly set, it must be a boolean value. * Tweak error message language and style (#217) Our previous error messages suffered from a couple drawbacks: - The messages were complete sentences, limiting our ability to compose error messages - The messages were overly broad in many cases - The messages unnecessarily required string interpolation that rarely provided much value These new messages are more succinct and are written as small clauses which can be used in conjunction with other error messages more easily. * copyedit Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> * Implement prerequisite cycle detection (#219) * Support attribute reference lookups (#215) This adds support for slash-delimited paths in clause attributes, bucketBy, etc. It does not do anything related to private attribute redaction because none of the U2C event logic is implemented yet. * Implement segment recursion and cycle detection (#220) Clauses in segment rules are now allowed to reference segments. To prevent an infinite recursion edge case, we implement a similar cycle detection mechanism as used on prerequisites. * Update event logic to support users to context change (#221) * Add legacy user-type support to the contract tests (#222) * Remove inline user configuration option (#223) * Add context_ configuration options (#224) These new context_ configuration options are meant to replace the historic user_ options. If both are provided, the context_ variant will take precedence. * Add support for flag context targets (#225) * Bump diplomat * Bump redis * Remove oga * Bump connection_pool * Favor set for faster target lookups (#228) A few of our internal models maintain arrays of values. These arrays can frequently be checked to see if they contain specific values. Since set lookups are much faster than array lookups, this commit changes the internal structure to a set for the values stored in Target and SegmentTarget. * Add secure mode hash to contract tests (#229) * Update big segment support for users to context (#226) To support the users to context change for big segments, this commit makes the following changes: - Introduces a new `Segment.unboundedContextKind` attribute. This will default to `LDContext::KIND_DEFAULT` and is only referenced when `Segment.unbounded` is true. - With the creation of multi-kind contexts, a single evaluation may result in multiple queries to the big segment store. This is reflected in the changes to the `EvalResult` processing. * Drop support for ruby 2.6 (#227) Ruby 2.6 went EOL in March 2022. We originally didn't drop support for it as doing so would require dropping support for jRuby as well. However, jRuby recently released 9.4 which is Ruby 2.7+ compatible. * Update remaining references from user to contexts (#231) There are multiple places throughout the code where we are still referencing users. I have tried to update all the places where a rename seems reasonable or appropriate. There is still some work to do in the test flag builders, but that will be done in a subsequent commit. * Remove new relic integration (#233) The new relic integration was removed many versions ago but a small trace remained behind. * Rename config option private_attribute_names (#234) Co-authored-by: Eli Bishop <eli@launchdarkly.com> * Update test data integration to support contexts (#232) * improve data model validation logging; allow missing/empty attribute for segmentMatch (#236) Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: hroederld <hroeder@launchdarkly.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> Co-authored-by: Louis Chan <lchan@launchdarkly.com>
|
We've been having similar issues in a client environment since rolling out Launch Darkly. It caused our apps to run 5 CPU x 60 instances of our app (whereas it used to run about 20 instances at 0.3 CPU each). Usually it is proceeded by some kind of network error. We are using version 6.4.0. Rebooting all our apps fixes it temporarily. I see this was supposed to have been fixed, but I'm curious what the ultimate root cause was and which commit in particular fixed this, and whether a similar issue might exist in version 6.4.0. |
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Jan 27, 2023
* don't let user fall outside of last bucket in rollout * refactor evaluation logic and move it out of the main namespace * comments * fix type coercion behavior * make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) * Drop support for EOL ruby versions (#196) Ruby 2.5 was EOL 2021-04-05 As of June 27th, 2022, the latest jRuby is Ruby 2.6 compatible. * Remove alias support (#195) * Add polling support for contract test service (#198) * Update rubocop and enable in CI (#197) Several of the Rubocop cop definitions have been renamed or moved to entirely other gems. This brings the configuration up to date with the latest naming conventions. * Add windows tests in circleci (#199) At some point in the past, we were experimenting with using Azure to verify Window builds. Now that CircleCI supports Windows, we should keep everything on a single CI provider. * Add application info support (#194) * reuse EvaluationDetail instances by precomputing results * rubocop reformatting * add super constructor calls * disable rubocop Rails rules and fix some remaining syntax offenses * fix super calls * Add big segment support to contract tests (#201) * Initial creation of LDContext (#206) This introduces the initial structure and usage of the LDContext class. Instances of this class are expected to be created through static factory methods: ```ruby LaunchDarkly::LDContext.create({...}) LaunchDarkly::LDContext.create_multi([...]) ``` This class is not completed yet. Rather, this initial commit is focused on the creation patterns and the most basic operations. Subsequent commits will continue fleshing out this class and its operation. The `get_value` method will see significant changes as we introduce attribute reference support. Its current more simplistic implementation exists only to serve some interim unit tests. * Add reference based value retrieval (#207) This commit introduces the References type used for targeting complex attributes in the new LDContexts. References are expected to be created through static factory methods: ```ruby LaunchDarkly::Reference.create("/a/b") LaunchDarkly::Reference.create_literal("/a/b") ``` These references can be used to retrieve values from an existing LDContext ```ruby ref = LaunchDarkly::Reference.create("/a/b") result = context.get_value_for_reference(ref) ``` * Basic changes to use contexts in evaluations instead of users (#208) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. This replaces `LDUser` with `LDContext` in the parameters for evaluations, and makes the minimum necessary adjustments to allow evaluations to keep working as before as long as the context kind is "user". None of the new behavior defined in the U2C spec is implemented yet. Generation of evaluation events is temporarily disabled because the event logic hasn't been updated yet. U2C contract tests for evaluations are partially enabled; a lot of functionality is still missing, but all the tests that only cover previously-existing evaluation behavior are passing. [pr]: launchdarkly/php-server-sdk-private#103 * Support ContextKind in Clauses (#209) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. The main features of this commit are: - introduction of `individual_context` and `individual_context_count` methods - context kind matching in clauses [pr]: launchdarkly/php-server-sdk-private#108 * Support included / excluded contexts in segments (#210) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. Segments are now able to provide `includedContext` and `excludedContext` properties which can target values within a specific context kind. ```json { "includedContexts": [ { "contextKind": "org", "values": ["orgkey1", "orgkey2"] } ] } ``` [pr]: launchdarkly/php-server-sdk-private#111 * Add contextKind support for rollouts & experiements (#211) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. [pr]: launchdarkly/php-server-sdk-private#110 * Style and test matcher improvements (#212) This commit enables several rubocop rules that were previously disabled. Once enabled, `rubocop -A` was run to automatically apply these fixes. There are a couple of additional changes that were made by hand: - I added the rubocop and rubocop-performance gems as dev packages. This should help address the original installation issue we ran into when I introduced these tools. - By default, new rubocop rules are disabled. This was the default before, but if you don't explicitly set this value, each run generates a ton of warning noise. This quiets that down. - Updates some LDContext tests to be more strict in their expectations of truth. * Remove support for secondary attribute (#213) As decided in the [spec], we are removing the special behavior of the secondary attribute. Going forward, secondary will be treated like any other attribute, and will no longer be included when determining the bucket for a context. [spec]: https://launchdarkly.atlassian.net/wiki/spaces/ENG/pages/2165212563/Consistent+and+Transparent+Rollout+Behavior+Unifying+Percent+Rollout+and+Traffic+Allocation * Remove deprecated APIs (#214) Since the users to context change requires a version break, this is the perfect time to remove previously deprecated bits of functionality. This includes: - Removing `update_processor*` config entries - `FileDataSource` entry point - `RedisFeatureStore` entry point - `Redis::sadd?` warning in unit tests * store data model with classes that aren't Hash * lint * remove [] override methods in places where we don't need them * comments * migrate some more of the model to be non-hash classes * lint * Anonymous cannot be nil in new context format (#216) The legacy user format allowed anonymous to be missing or explicitly provided but set to nil. The new context format requires anonymous to either not be set, or if it is explicitly set, it must be a boolean value. * Tweak error message language and style (#217) Our previous error messages suffered from a couple drawbacks: - The messages were complete sentences, limiting our ability to compose error messages - The messages were overly broad in many cases - The messages unnecessarily required string interpolation that rarely provided much value These new messages are more succinct and are written as small clauses which can be used in conjunction with other error messages more easily. * copyedit Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> * Implement prerequisite cycle detection (#219) * Support attribute reference lookups (#215) This adds support for slash-delimited paths in clause attributes, bucketBy, etc. It does not do anything related to private attribute redaction because none of the U2C event logic is implemented yet. * Implement segment recursion and cycle detection (#220) Clauses in segment rules are now allowed to reference segments. To prevent an infinite recursion edge case, we implement a similar cycle detection mechanism as used on prerequisites. * Update event logic to support users to context change (#221) * Add legacy user-type support to the contract tests (#222) * Remove inline user configuration option (#223) * Add context_ configuration options (#224) These new context_ configuration options are meant to replace the historic user_ options. If both are provided, the context_ variant will take precedence. * Add support for flag context targets (#225) * Bump diplomat * Bump redis * Remove oga * Bump connection_pool * Favor set for faster target lookups (#228) A few of our internal models maintain arrays of values. These arrays can frequently be checked to see if they contain specific values. Since set lookups are much faster than array lookups, this commit changes the internal structure to a set for the values stored in Target and SegmentTarget. * Add secure mode hash to contract tests (#229) * Update big segment support for users to context (#226) To support the users to context change for big segments, this commit makes the following changes: - Introduces a new `Segment.unboundedContextKind` attribute. This will default to `LDContext::KIND_DEFAULT` and is only referenced when `Segment.unbounded` is true. - With the creation of multi-kind contexts, a single evaluation may result in multiple queries to the big segment store. This is reflected in the changes to the `EvalResult` processing. * Drop support for ruby 2.6 (#227) Ruby 2.6 went EOL in March 2022. We originally didn't drop support for it as doing so would require dropping support for jRuby as well. However, jRuby recently released 9.4 which is Ruby 2.7+ compatible. * Update remaining references from user to contexts (#231) There are multiple places throughout the code where we are still referencing users. I have tried to update all the places where a rename seems reasonable or appropriate. There is still some work to do in the test flag builders, but that will be done in a subsequent commit. * Remove new relic integration (#233) The new relic integration was removed many versions ago but a small trace remained behind. * Rename config option private_attribute_names (#234) Co-authored-by: Eli Bishop <eli@launchdarkly.com> * Update test data integration to support contexts (#232) * improve data model validation logging; allow missing/empty attribute for segmentMatch (#236) * Fix JSON serialization failure (#237) When we introduced models for the flag and segment data, we added to_json methods which proxy to the underlying to_json method provided by the `json` gem. We defined the method with a variadic parameter, but we failed to unpack it when passing it on to the underlying implementation. This resulted in a serialization failure which prevented the redis data store from initializing. --------- Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: hroederld <hroeder@launchdarkly.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> Co-authored-by: Louis Chan <lchan@launchdarkly.com>
|
An update - we upgraded to 7.0.2 and that did not fix it. |
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Mar 17, 2023
* don't let user fall outside of last bucket in rollout * refactor evaluation logic and move it out of the main namespace * comments * fix type coercion behavior * make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) * Drop support for EOL ruby versions (#196) Ruby 2.5 was EOL 2021-04-05 As of June 27th, 2022, the latest jRuby is Ruby 2.6 compatible. * Remove alias support (#195) * Add polling support for contract test service (#198) * Update rubocop and enable in CI (#197) Several of the Rubocop cop definitions have been renamed or moved to entirely other gems. This brings the configuration up to date with the latest naming conventions. * Add windows tests in circleci (#199) At some point in the past, we were experimenting with using Azure to verify Window builds. Now that CircleCI supports Windows, we should keep everything on a single CI provider. * Add application info support (#194) * reuse EvaluationDetail instances by precomputing results * rubocop reformatting * add super constructor calls * disable rubocop Rails rules and fix some remaining syntax offenses * fix super calls * Add big segment support to contract tests (#201) * Initial creation of LDContext (#206) This introduces the initial structure and usage of the LDContext class. Instances of this class are expected to be created through static factory methods: ```ruby LaunchDarkly::LDContext.create({...}) LaunchDarkly::LDContext.create_multi([...]) ``` This class is not completed yet. Rather, this initial commit is focused on the creation patterns and the most basic operations. Subsequent commits will continue fleshing out this class and its operation. The `get_value` method will see significant changes as we introduce attribute reference support. Its current more simplistic implementation exists only to serve some interim unit tests. * Add reference based value retrieval (#207) This commit introduces the References type used for targeting complex attributes in the new LDContexts. References are expected to be created through static factory methods: ```ruby LaunchDarkly::Reference.create("/a/b") LaunchDarkly::Reference.create_literal("/a/b") ``` These references can be used to retrieve values from an existing LDContext ```ruby ref = LaunchDarkly::Reference.create("/a/b") result = context.get_value_for_reference(ref) ``` * Basic changes to use contexts in evaluations instead of users (#208) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. This replaces `LDUser` with `LDContext` in the parameters for evaluations, and makes the minimum necessary adjustments to allow evaluations to keep working as before as long as the context kind is "user". None of the new behavior defined in the U2C spec is implemented yet. Generation of evaluation events is temporarily disabled because the event logic hasn't been updated yet. U2C contract tests for evaluations are partially enabled; a lot of functionality is still missing, but all the tests that only cover previously-existing evaluation behavior are passing. [pr]: launchdarkly/php-server-sdk-private#103 * Support ContextKind in Clauses (#209) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. The main features of this commit are: - introduction of `individual_context` and `individual_context_count` methods - context kind matching in clauses [pr]: launchdarkly/php-server-sdk-private#108 * Support included / excluded contexts in segments (#210) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. Segments are now able to provide `includedContext` and `excludedContext` properties which can target values within a specific context kind. ```json { "includedContexts": [ { "contextKind": "org", "values": ["orgkey1", "orgkey2"] } ] } ``` [pr]: launchdarkly/php-server-sdk-private#111 * Add contextKind support for rollouts & experiements (#211) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. [pr]: launchdarkly/php-server-sdk-private#110 * Style and test matcher improvements (#212) This commit enables several rubocop rules that were previously disabled. Once enabled, `rubocop -A` was run to automatically apply these fixes. There are a couple of additional changes that were made by hand: - I added the rubocop and rubocop-performance gems as dev packages. This should help address the original installation issue we ran into when I introduced these tools. - By default, new rubocop rules are disabled. This was the default before, but if you don't explicitly set this value, each run generates a ton of warning noise. This quiets that down. - Updates some LDContext tests to be more strict in their expectations of truth. * Remove support for secondary attribute (#213) As decided in the [spec], we are removing the special behavior of the secondary attribute. Going forward, secondary will be treated like any other attribute, and will no longer be included when determining the bucket for a context. [spec]: https://launchdarkly.atlassian.net/wiki/spaces/ENG/pages/2165212563/Consistent+and+Transparent+Rollout+Behavior+Unifying+Percent+Rollout+and+Traffic+Allocation * Remove deprecated APIs (#214) Since the users to context change requires a version break, this is the perfect time to remove previously deprecated bits of functionality. This includes: - Removing `update_processor*` config entries - `FileDataSource` entry point - `RedisFeatureStore` entry point - `Redis::sadd?` warning in unit tests * store data model with classes that aren't Hash * lint * remove [] override methods in places where we don't need them * comments * migrate some more of the model to be non-hash classes * lint * Anonymous cannot be nil in new context format (#216) The legacy user format allowed anonymous to be missing or explicitly provided but set to nil. The new context format requires anonymous to either not be set, or if it is explicitly set, it must be a boolean value. * Tweak error message language and style (#217) Our previous error messages suffered from a couple drawbacks: - The messages were complete sentences, limiting our ability to compose error messages - The messages were overly broad in many cases - The messages unnecessarily required string interpolation that rarely provided much value These new messages are more succinct and are written as small clauses which can be used in conjunction with other error messages more easily. * copyedit Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> * Implement prerequisite cycle detection (#219) * Support attribute reference lookups (#215) This adds support for slash-delimited paths in clause attributes, bucketBy, etc. It does not do anything related to private attribute redaction because none of the U2C event logic is implemented yet. * Implement segment recursion and cycle detection (#220) Clauses in segment rules are now allowed to reference segments. To prevent an infinite recursion edge case, we implement a similar cycle detection mechanism as used on prerequisites. * Update event logic to support users to context change (#221) * Add legacy user-type support to the contract tests (#222) * Remove inline user configuration option (#223) * Add context_ configuration options (#224) These new context_ configuration options are meant to replace the historic user_ options. If both are provided, the context_ variant will take precedence. * Add support for flag context targets (#225) * Bump diplomat * Bump redis * Remove oga * Bump connection_pool * Favor set for faster target lookups (#228) A few of our internal models maintain arrays of values. These arrays can frequently be checked to see if they contain specific values. Since set lookups are much faster than array lookups, this commit changes the internal structure to a set for the values stored in Target and SegmentTarget. * Add secure mode hash to contract tests (#229) * Update big segment support for users to context (#226) To support the users to context change for big segments, this commit makes the following changes: - Introduces a new `Segment.unboundedContextKind` attribute. This will default to `LDContext::KIND_DEFAULT` and is only referenced when `Segment.unbounded` is true. - With the creation of multi-kind contexts, a single evaluation may result in multiple queries to the big segment store. This is reflected in the changes to the `EvalResult` processing. * Drop support for ruby 2.6 (#227) Ruby 2.6 went EOL in March 2022. We originally didn't drop support for it as doing so would require dropping support for jRuby as well. However, jRuby recently released 9.4 which is Ruby 2.7+ compatible. * Update remaining references from user to contexts (#231) There are multiple places throughout the code where we are still referencing users. I have tried to update all the places where a rename seems reasonable or appropriate. There is still some work to do in the test flag builders, but that will be done in a subsequent commit. * Remove new relic integration (#233) The new relic integration was removed many versions ago but a small trace remained behind. * Rename config option private_attribute_names (#234) Co-authored-by: Eli Bishop <eli@launchdarkly.com> * Update test data integration to support contexts (#232) * improve data model validation logging; allow missing/empty attribute for segmentMatch (#236) * Fix JSON serialization failure (#237) When we introduced models for the flag and segment data, we added to_json methods which proxy to the underlying to_json method provided by the `json` gem. We defined the method with a variadic parameter, but we failed to unpack it when passing it on to the underlying implementation. This resulted in a serialization failure which prevented the redis data store from initializing. * fix: Bump eventsource to resolve header parsing (#239) The underlying event source library had an issue where, in certain environments, content-type header detection was failing. This was resolved in v2.2.2 of the event source gem. --------- Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: hroederld <hroeder@launchdarkly.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> Co-authored-by: Louis Chan <lchan@launchdarkly.com>
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Apr 3, 2023
* make type coercion behavior consistent with earlier versions for now * whitespace * break up Evaluator tests further * make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) * Drop support for EOL ruby versions (#196) Ruby 2.5 was EOL 2021-04-05 As of June 27th, 2022, the latest jRuby is Ruby 2.6 compatible. * Remove alias support (#195) * Add polling support for contract test service (#198) * Update rubocop and enable in CI (#197) Several of the Rubocop cop definitions have been renamed or moved to entirely other gems. This brings the configuration up to date with the latest naming conventions. * Add windows tests in circleci (#199) At some point in the past, we were experimenting with using Azure to verify Window builds. Now that CircleCI supports Windows, we should keep everything on a single CI provider. * Add application info support (#194) * reuse EvaluationDetail instances by precomputing results * rubocop reformatting * add super constructor calls * disable rubocop Rails rules and fix some remaining syntax offenses * fix super calls * Add big segment support to contract tests (#201) * Initial creation of LDContext (#206) This introduces the initial structure and usage of the LDContext class. Instances of this class are expected to be created through static factory methods: ```ruby LaunchDarkly::LDContext.create({...}) LaunchDarkly::LDContext.create_multi([...]) ``` This class is not completed yet. Rather, this initial commit is focused on the creation patterns and the most basic operations. Subsequent commits will continue fleshing out this class and its operation. The `get_value` method will see significant changes as we introduce attribute reference support. Its current more simplistic implementation exists only to serve some interim unit tests. * Add reference based value retrieval (#207) This commit introduces the References type used for targeting complex attributes in the new LDContexts. References are expected to be created through static factory methods: ```ruby LaunchDarkly::Reference.create("/a/b") LaunchDarkly::Reference.create_literal("/a/b") ``` These references can be used to retrieve values from an existing LDContext ```ruby ref = LaunchDarkly::Reference.create("/a/b") result = context.get_value_for_reference(ref) ``` * Basic changes to use contexts in evaluations instead of users (#208) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. This replaces `LDUser` with `LDContext` in the parameters for evaluations, and makes the minimum necessary adjustments to allow evaluations to keep working as before as long as the context kind is "user". None of the new behavior defined in the U2C spec is implemented yet. Generation of evaluation events is temporarily disabled because the event logic hasn't been updated yet. U2C contract tests for evaluations are partially enabled; a lot of functionality is still missing, but all the tests that only cover previously-existing evaluation behavior are passing. [pr]: launchdarkly/php-server-sdk-private#103 * Support ContextKind in Clauses (#209) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. The main features of this commit are: - introduction of `individual_context` and `individual_context_count` methods - context kind matching in clauses [pr]: launchdarkly/php-server-sdk-private#108 * Support included / excluded contexts in segments (#210) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. Segments are now able to provide `includedContext` and `excludedContext` properties which can target values within a specific context kind. ```json { "includedContexts": [ { "contextKind": "org", "values": ["orgkey1", "orgkey2"] } ] } ``` [pr]: launchdarkly/php-server-sdk-private#111 * Add contextKind support for rollouts & experiements (#211) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. [pr]: launchdarkly/php-server-sdk-private#110 * Style and test matcher improvements (#212) This commit enables several rubocop rules that were previously disabled. Once enabled, `rubocop -A` was run to automatically apply these fixes. There are a couple of additional changes that were made by hand: - I added the rubocop and rubocop-performance gems as dev packages. This should help address the original installation issue we ran into when I introduced these tools. - By default, new rubocop rules are disabled. This was the default before, but if you don't explicitly set this value, each run generates a ton of warning noise. This quiets that down. - Updates some LDContext tests to be more strict in their expectations of truth. * Remove support for secondary attribute (#213) As decided in the [spec], we are removing the special behavior of the secondary attribute. Going forward, secondary will be treated like any other attribute, and will no longer be included when determining the bucket for a context. [spec]: https://launchdarkly.atlassian.net/wiki/spaces/ENG/pages/2165212563/Consistent+and+Transparent+Rollout+Behavior+Unifying+Percent+Rollout+and+Traffic+Allocation * Remove deprecated APIs (#214) Since the users to context change requires a version break, this is the perfect time to remove previously deprecated bits of functionality. This includes: - Removing `update_processor*` config entries - `FileDataSource` entry point - `RedisFeatureStore` entry point - `Redis::sadd?` warning in unit tests * store data model with classes that aren't Hash * lint * remove [] override methods in places where we don't need them * comments * migrate some more of the model to be non-hash classes * lint * Anonymous cannot be nil in new context format (#216) The legacy user format allowed anonymous to be missing or explicitly provided but set to nil. The new context format requires anonymous to either not be set, or if it is explicitly set, it must be a boolean value. * Tweak error message language and style (#217) Our previous error messages suffered from a couple drawbacks: - The messages were complete sentences, limiting our ability to compose error messages - The messages were overly broad in many cases - The messages unnecessarily required string interpolation that rarely provided much value These new messages are more succinct and are written as small clauses which can be used in conjunction with other error messages more easily. * copyedit Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> * Implement prerequisite cycle detection (#219) * Support attribute reference lookups (#215) This adds support for slash-delimited paths in clause attributes, bucketBy, etc. It does not do anything related to private attribute redaction because none of the U2C event logic is implemented yet. * Implement segment recursion and cycle detection (#220) Clauses in segment rules are now allowed to reference segments. To prevent an infinite recursion edge case, we implement a similar cycle detection mechanism as used on prerequisites. * Update event logic to support users to context change (#221) * Add legacy user-type support to the contract tests (#222) * Remove inline user configuration option (#223) * Add context_ configuration options (#224) These new context_ configuration options are meant to replace the historic user_ options. If both are provided, the context_ variant will take precedence. * Add support for flag context targets (#225) * Bump diplomat * Bump redis * Remove oga * Bump connection_pool * Favor set for faster target lookups (#228) A few of our internal models maintain arrays of values. These arrays can frequently be checked to see if they contain specific values. Since set lookups are much faster than array lookups, this commit changes the internal structure to a set for the values stored in Target and SegmentTarget. * Add secure mode hash to contract tests (#229) * Update big segment support for users to context (#226) To support the users to context change for big segments, this commit makes the following changes: - Introduces a new `Segment.unboundedContextKind` attribute. This will default to `LDContext::KIND_DEFAULT` and is only referenced when `Segment.unbounded` is true. - With the creation of multi-kind contexts, a single evaluation may result in multiple queries to the big segment store. This is reflected in the changes to the `EvalResult` processing. * Drop support for ruby 2.6 (#227) Ruby 2.6 went EOL in March 2022. We originally didn't drop support for it as doing so would require dropping support for jRuby as well. However, jRuby recently released 9.4 which is Ruby 2.7+ compatible. * Update remaining references from user to contexts (#231) There are multiple places throughout the code where we are still referencing users. I have tried to update all the places where a rename seems reasonable or appropriate. There is still some work to do in the test flag builders, but that will be done in a subsequent commit. * Remove new relic integration (#233) The new relic integration was removed many versions ago but a small trace remained behind. * Rename config option private_attribute_names (#234) Co-authored-by: Eli Bishop <eli@launchdarkly.com> * Update test data integration to support contexts (#232) * improve data model validation logging; allow missing/empty attribute for segmentMatch (#236) * Fix JSON serialization failure (#237) When we introduced models for the flag and segment data, we added to_json methods which proxy to the underlying to_json method provided by the `json` gem. We defined the method with a variadic parameter, but we failed to unpack it when passing it on to the underlying implementation. This resulted in a serialization failure which prevented the redis data store from initializing. * fix: Bump eventsource to resolve header parsing (#239) The underlying event source library had an issue where, in certain environments, content-type header detection was failing. This was resolved in v2.2.2 of the event source gem. * Add key to error log for invalid context during variation call for easier debugging (#214) --------- Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: hroederld <hroeder@launchdarkly.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> Co-authored-by: Louis Chan <lchan@launchdarkly.com> Co-authored-by: Matt Hooks <46452201+matt-dutchie@users.noreply.github.com>
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
Apr 13, 2023
* make EvaluationReason an immutable class * FrozenError doesn't exist in older Ruby, use more general RuntimeError * precompute evaluation reasons when we receive a flag * rm unused * rename FeatureStore to DataStore * remove references to UpdateProcessor (now DataSource) * add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) * Drop support for EOL ruby versions (#196) Ruby 2.5 was EOL 2021-04-05 As of June 27th, 2022, the latest jRuby is Ruby 2.6 compatible. * Remove alias support (#195) * Add polling support for contract test service (#198) * Update rubocop and enable in CI (#197) Several of the Rubocop cop definitions have been renamed or moved to entirely other gems. This brings the configuration up to date with the latest naming conventions. * Add windows tests in circleci (#199) At some point in the past, we were experimenting with using Azure to verify Window builds. Now that CircleCI supports Windows, we should keep everything on a single CI provider. * Add application info support (#194) * reuse EvaluationDetail instances by precomputing results * rubocop reformatting * add super constructor calls * disable rubocop Rails rules and fix some remaining syntax offenses * fix super calls * Add big segment support to contract tests (#201) * Initial creation of LDContext (#206) This introduces the initial structure and usage of the LDContext class. Instances of this class are expected to be created through static factory methods: ```ruby LaunchDarkly::LDContext.create({...}) LaunchDarkly::LDContext.create_multi([...]) ``` This class is not completed yet. Rather, this initial commit is focused on the creation patterns and the most basic operations. Subsequent commits will continue fleshing out this class and its operation. The `get_value` method will see significant changes as we introduce attribute reference support. Its current more simplistic implementation exists only to serve some interim unit tests. * Add reference based value retrieval (#207) This commit introduces the References type used for targeting complex attributes in the new LDContexts. References are expected to be created through static factory methods: ```ruby LaunchDarkly::Reference.create("/a/b") LaunchDarkly::Reference.create_literal("/a/b") ``` These references can be used to retrieve values from an existing LDContext ```ruby ref = LaunchDarkly::Reference.create("/a/b") result = context.get_value_for_reference(ref) ``` * Basic changes to use contexts in evaluations instead of users (#208) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. This replaces `LDUser` with `LDContext` in the parameters for evaluations, and makes the minimum necessary adjustments to allow evaluations to keep working as before as long as the context kind is "user". None of the new behavior defined in the U2C spec is implemented yet. Generation of evaluation events is temporarily disabled because the event logic hasn't been updated yet. U2C contract tests for evaluations are partially enabled; a lot of functionality is still missing, but all the tests that only cover previously-existing evaluation behavior are passing. [pr]: launchdarkly/php-server-sdk-private#103 * Support ContextKind in Clauses (#209) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. The main features of this commit are: - introduction of `individual_context` and `individual_context_count` methods - context kind matching in clauses [pr]: launchdarkly/php-server-sdk-private#108 * Support included / excluded contexts in segments (#210) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. Segments are now able to provide `includedContext` and `excludedContext` properties which can target values within a specific context kind. ```json { "includedContexts": [ { "contextKind": "org", "values": ["orgkey1", "orgkey2"] } ] } ``` [pr]: launchdarkly/php-server-sdk-private#111 * Add contextKind support for rollouts & experiements (#211) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. [pr]: launchdarkly/php-server-sdk-private#110 * Style and test matcher improvements (#212) This commit enables several rubocop rules that were previously disabled. Once enabled, `rubocop -A` was run to automatically apply these fixes. There are a couple of additional changes that were made by hand: - I added the rubocop and rubocop-performance gems as dev packages. This should help address the original installation issue we ran into when I introduced these tools. - By default, new rubocop rules are disabled. This was the default before, but if you don't explicitly set this value, each run generates a ton of warning noise. This quiets that down. - Updates some LDContext tests to be more strict in their expectations of truth. * Remove support for secondary attribute (#213) As decided in the [spec], we are removing the special behavior of the secondary attribute. Going forward, secondary will be treated like any other attribute, and will no longer be included when determining the bucket for a context. [spec]: https://launchdarkly.atlassian.net/wiki/spaces/ENG/pages/2165212563/Consistent+and+Transparent+Rollout+Behavior+Unifying+Percent+Rollout+and+Traffic+Allocation * Remove deprecated APIs (#214) Since the users to context change requires a version break, this is the perfect time to remove previously deprecated bits of functionality. This includes: - Removing `update_processor*` config entries - `FileDataSource` entry point - `RedisFeatureStore` entry point - `Redis::sadd?` warning in unit tests * store data model with classes that aren't Hash * lint * remove [] override methods in places where we don't need them * comments * migrate some more of the model to be non-hash classes * lint * Anonymous cannot be nil in new context format (#216) The legacy user format allowed anonymous to be missing or explicitly provided but set to nil. The new context format requires anonymous to either not be set, or if it is explicitly set, it must be a boolean value. * Tweak error message language and style (#217) Our previous error messages suffered from a couple drawbacks: - The messages were complete sentences, limiting our ability to compose error messages - The messages were overly broad in many cases - The messages unnecessarily required string interpolation that rarely provided much value These new messages are more succinct and are written as small clauses which can be used in conjunction with other error messages more easily. * copyedit Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> * Implement prerequisite cycle detection (#219) * Support attribute reference lookups (#215) This adds support for slash-delimited paths in clause attributes, bucketBy, etc. It does not do anything related to private attribute redaction because none of the U2C event logic is implemented yet. * Implement segment recursion and cycle detection (#220) Clauses in segment rules are now allowed to reference segments. To prevent an infinite recursion edge case, we implement a similar cycle detection mechanism as used on prerequisites. * Update event logic to support users to context change (#221) * Add legacy user-type support to the contract tests (#222) * Remove inline user configuration option (#223) * Add context_ configuration options (#224) These new context_ configuration options are meant to replace the historic user_ options. If both are provided, the context_ variant will take precedence. * Add support for flag context targets (#225) * Bump diplomat * Bump redis * Remove oga * Bump connection_pool * Favor set for faster target lookups (#228) A few of our internal models maintain arrays of values. These arrays can frequently be checked to see if they contain specific values. Since set lookups are much faster than array lookups, this commit changes the internal structure to a set for the values stored in Target and SegmentTarget. * Add secure mode hash to contract tests (#229) * Update big segment support for users to context (#226) To support the users to context change for big segments, this commit makes the following changes: - Introduces a new `Segment.unboundedContextKind` attribute. This will default to `LDContext::KIND_DEFAULT` and is only referenced when `Segment.unbounded` is true. - With the creation of multi-kind contexts, a single evaluation may result in multiple queries to the big segment store. This is reflected in the changes to the `EvalResult` processing. * Drop support for ruby 2.6 (#227) Ruby 2.6 went EOL in March 2022. We originally didn't drop support for it as doing so would require dropping support for jRuby as well. However, jRuby recently released 9.4 which is Ruby 2.7+ compatible. * Update remaining references from user to contexts (#231) There are multiple places throughout the code where we are still referencing users. I have tried to update all the places where a rename seems reasonable or appropriate. There is still some work to do in the test flag builders, but that will be done in a subsequent commit. * Remove new relic integration (#233) The new relic integration was removed many versions ago but a small trace remained behind. * Rename config option private_attribute_names (#234) Co-authored-by: Eli Bishop <eli@launchdarkly.com> * Update test data integration to support contexts (#232) * improve data model validation logging; allow missing/empty attribute for segmentMatch (#236) * Fix JSON serialization failure (#237) When we introduced models for the flag and segment data, we added to_json methods which proxy to the underlying to_json method provided by the `json` gem. We defined the method with a variadic parameter, but we failed to unpack it when passing it on to the underlying implementation. This resulted in a serialization failure which prevented the redis data store from initializing. * fix: Bump eventsource to resolve header parsing (#239) The underlying event source library had an issue where, in certain environments, content-type header detection was failing. This was resolved in v2.2.2 of the event source gem. * Add key to error log for invalid context during variation call for easier debugging (#214) * feat: Add support for payload filtering (#238) --------- Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: hroederld <hroeder@launchdarkly.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> Co-authored-by: Louis Chan <lchan@launchdarkly.com> Co-authored-by: Matt Hooks <46452201+matt-dutchie@users.noreply.github.com>
LaunchDarklyReleaseBot
added a commit
that referenced
this issue
May 4, 2023
* add event payload ID header * (6.0) drop support for old Ruby versions * add Ruby version constraint to gemspec * remove Rake dependency * update ld-eventsource to 1.0.2 which doesn't have Rake dependency * implement diagnostic events in Ruby (#130) * update ruby-eventsource to 1.0.3 for backoff bug * fix incorrect initialization of EventProcessor * remove install-time openssl check that breaks if you don't have rake * treat comparison with wrong data type as a non-match, not an exception (#134) * fail fast for nil SDK key when appropriate * tolerate nil value for user.custom (#137) * Only shutdown the Redis pool if it is owned by the SDK (#158) * Only shutdown a Redis pool created by SDK * Make pool shutdown behavior an option * improve doc comment * remove support for indirect/patch and indirect/put (#138) * update to json 2.3.1 (#139) * update json dep to 2.3.x to fix CVE * add publication of API docs on GitHub Pages (#143) * try fixing release metadata * update the default base url (#144) * revert renames of feature_store & update_processor * [ch92483] Use http gem and add socket factory support (#142) * update dependencies and add CI for ruby 3 (#141) * reference eventsource 2.0 in gemspec * add 5.x releasable branch for releaser * use Ruby 2.6.6 in releases * Removed the guides link * [ch99757] add alias method (#147) * don't send event for nil user evaluation * remove lockfile (#148) * rm redundant nil check * Experiment Allocation Changes (#150) * WIP - from sam's pairing session * starting sdk changes * adding tests and making sure everything works * adding more tests * removing the singleton for fallthrough * Revert "removing the singleton for fallthrough" This reverts commit dff7adbb809ecc63118d0fbff9742a88a039c679. * taking a different approach to keep things immutable * adding tests for untracked * remove unnecessary comment * making sure to return two values in all code paths Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> * Use camelCase for JSON property names (#151) The in_experiment attribute was added to reasons as part of #150 but it doesn't appear to be received in events. I think that's because it's sending it in JSON as "in_experiment" rather than "inExperiment" as we expect to parse it. * fixing ruby logic causing ih failures (#152) * fixing ruby logic * adding missing spec * Apply suggestions from code review Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * pr tweaks * making spec language consistent Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> * add log warning for missing user key (#153) * add log warnings for nil/empty user key * rm warning for empty string key * fix test * diagnostic events should respect HTTPS_PROXY (#154) * minor test simplification (#155) * allow higher minor versions of json and http gems * allow v5.x of http gem (#157) * use Bundler 2.2.10 + modernize CI config (#158) * enable verbose rspec output * fix socket factory tests * restore log suppression * Replacing deprecated circleci image usage (#159) * use Releaser v2 config (#161) * Updates docs URLs * Update lib/ldclient-rb/ldclient.rb Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> * remove reliance on git in gemspec (#163) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 (#164) * use ruby-eventsource 2.1.1 for fix of sc-123850 and sc-125504 * comment phrasing * Start work on flag builder. * Add user targeting and rule builder * Add datasource implementation * Convert the current_flags hash to use symbols instead of strings as keys * Fix typo on FlagRuleBuilder copy constructor * minor refactoring of impl; Added use of new Clause struct instead of Hash in FlagRuleBuilder; Moved TestData.factory out of Impl namespace and renamed Impl to TestDataImpl * Add the doc comments * (big segments 1) add public config/interface/reason types (#167) * Cleanup docstrings to be YARD docs * Added Util.is_bool helper function to clean up the check for whether an object is a boolean; Removed the DeepCopyHash/DeepCopyArray objects in favor of deep_copy_hash and deep_copy_array functions * Move public classes out of Impl namespace. Most of it is in public namespace except for the data source now. * Move require of concurrent/atomics to the correct module * (big segments 2) implement Big Segments evaluation & status APIs (#168) * improve CONTRIBUTING.md with notes on code organization * add note about doc comments * Cleanup YARD warnings and cleanup docs * Address PR feedback: Move is_bool back to Impl namespace to avoid confusion; Remove unnecessary nil check on variations in build function; fixup comments * (big segments 3) implement Redis & DynamoDB big segment stores (#169) * add missing import * fix stale calculation * fix big segments user hash algorithm to use SHA256 * improve & refactor client/evaluation tests * more cleanup/DRY * add use_preconfigured_flag and use_preconfigured_segment to TestData (#173) * always cache big segment query result even if it's nil * comments * add test for cache expiration * use TestData in our own tests (#174) * use TestData in our own tests * fix test * replace LaunchDarkly::FileDataSource with LaunchDarkly::Integrations::FileData * update ruby-eventsource version for recent SSE fixes * Bump bundler version (#184) * Add ability to to set initial reconnect delay (#183) * Treat secondary as a built-in attribute (#180) * all_flags_state is invalid if store isn't initialized (#182) * identify should not emit events if user key is "" (#181) * Account for traffic allocation on all flags (#185) * Add contract tests (#178) * Fix string interpolation in log message (#187) * Default opts to empty hash when creating persistent feature store (#186) * Remove Hakiri badge from README (#188) Hakiri was sunset on January 31st, 2022 at which time our badge stopped working. * detect http/https proxy env vars when creating HTTP clients * rever accidental change * fix nil safety in test service config * master -> main (#190) * master -> main * update ruby-eventsource version for parsing efficiency fix * miscellaneous optimizations for event processing (#193) * Drop support for EOL ruby versions (#196) Ruby 2.5 was EOL 2021-04-05 As of June 27th, 2022, the latest jRuby is Ruby 2.6 compatible. * Remove alias support (#195) * Add polling support for contract test service (#198) * Update rubocop and enable in CI (#197) Several of the Rubocop cop definitions have been renamed or moved to entirely other gems. This brings the configuration up to date with the latest naming conventions. * Add windows tests in circleci (#199) At some point in the past, we were experimenting with using Azure to verify Window builds. Now that CircleCI supports Windows, we should keep everything on a single CI provider. * Add application info support (#194) * reuse EvaluationDetail instances by precomputing results * rubocop reformatting * add super constructor calls * disable rubocop Rails rules and fix some remaining syntax offenses * fix super calls * Add big segment support to contract tests (#201) * Initial creation of LDContext (#206) This introduces the initial structure and usage of the LDContext class. Instances of this class are expected to be created through static factory methods: ```ruby LaunchDarkly::LDContext.create({...}) LaunchDarkly::LDContext.create_multi([...]) ``` This class is not completed yet. Rather, this initial commit is focused on the creation patterns and the most basic operations. Subsequent commits will continue fleshing out this class and its operation. The `get_value` method will see significant changes as we introduce attribute reference support. Its current more simplistic implementation exists only to serve some interim unit tests. * Add reference based value retrieval (#207) This commit introduces the References type used for targeting complex attributes in the new LDContexts. References are expected to be created through static factory methods: ```ruby LaunchDarkly::Reference.create("/a/b") LaunchDarkly::Reference.create_literal("/a/b") ``` These references can be used to retrieve values from an existing LDContext ```ruby ref = LaunchDarkly::Reference.create("/a/b") result = context.get_value_for_reference(ref) ``` * Basic changes to use contexts in evaluations instead of users (#208) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. This replaces `LDUser` with `LDContext` in the parameters for evaluations, and makes the minimum necessary adjustments to allow evaluations to keep working as before as long as the context kind is "user". None of the new behavior defined in the U2C spec is implemented yet. Generation of evaluation events is temporarily disabled because the event logic hasn't been updated yet. U2C contract tests for evaluations are partially enabled; a lot of functionality is still missing, but all the tests that only cover previously-existing evaluation behavior are passing. [pr]: launchdarkly/php-server-sdk-private#103 * Support ContextKind in Clauses (#209) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. The main features of this commit are: - introduction of `individual_context` and `individual_context_count` methods - context kind matching in clauses [pr]: launchdarkly/php-server-sdk-private#108 * Support included / excluded contexts in segments (#210) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. Segments are now able to provide `includedContext` and `excludedContext` properties which can target values within a specific context kind. ```json { "includedContexts": [ { "contextKind": "org", "values": ["orgkey1", "orgkey2"] } ] } ``` [pr]: launchdarkly/php-server-sdk-private#111 * Add contextKind support for rollouts & experiements (#211) This commit follows the general approach of the [equivalent PHP SDK PR][pr]. [pr]: launchdarkly/php-server-sdk-private#110 * Style and test matcher improvements (#212) This commit enables several rubocop rules that were previously disabled. Once enabled, `rubocop -A` was run to automatically apply these fixes. There are a couple of additional changes that were made by hand: - I added the rubocop and rubocop-performance gems as dev packages. This should help address the original installation issue we ran into when I introduced these tools. - By default, new rubocop rules are disabled. This was the default before, but if you don't explicitly set this value, each run generates a ton of warning noise. This quiets that down. - Updates some LDContext tests to be more strict in their expectations of truth. * Remove support for secondary attribute (#213) As decided in the [spec], we are removing the special behavior of the secondary attribute. Going forward, secondary will be treated like any other attribute, and will no longer be included when determining the bucket for a context. [spec]: https://launchdarkly.atlassian.net/wiki/spaces/ENG/pages/2165212563/Consistent+and+Transparent+Rollout+Behavior+Unifying+Percent+Rollout+and+Traffic+Allocation * Remove deprecated APIs (#214) Since the users to context change requires a version break, this is the perfect time to remove previously deprecated bits of functionality. This includes: - Removing `update_processor*` config entries - `FileDataSource` entry point - `RedisFeatureStore` entry point - `Redis::sadd?` warning in unit tests * store data model with classes that aren't Hash * lint * remove [] override methods in places where we don't need them * comments * migrate some more of the model to be non-hash classes * lint * Anonymous cannot be nil in new context format (#216) The legacy user format allowed anonymous to be missing or explicitly provided but set to nil. The new context format requires anonymous to either not be set, or if it is explicitly set, it must be a boolean value. * Tweak error message language and style (#217) Our previous error messages suffered from a couple drawbacks: - The messages were complete sentences, limiting our ability to compose error messages - The messages were overly broad in many cases - The messages unnecessarily required string interpolation that rarely provided much value These new messages are more succinct and are written as small clauses which can be used in conjunction with other error messages more easily. * copyedit Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> * Implement prerequisite cycle detection (#219) * Support attribute reference lookups (#215) This adds support for slash-delimited paths in clause attributes, bucketBy, etc. It does not do anything related to private attribute redaction because none of the U2C event logic is implemented yet. * Implement segment recursion and cycle detection (#220) Clauses in segment rules are now allowed to reference segments. To prevent an infinite recursion edge case, we implement a similar cycle detection mechanism as used on prerequisites. * Update event logic to support users to context change (#221) * Add legacy user-type support to the contract tests (#222) * Remove inline user configuration option (#223) * Add context_ configuration options (#224) These new context_ configuration options are meant to replace the historic user_ options. If both are provided, the context_ variant will take precedence. * Add support for flag context targets (#225) * Bump diplomat * Bump redis * Remove oga * Bump connection_pool * Favor set for faster target lookups (#228) A few of our internal models maintain arrays of values. These arrays can frequently be checked to see if they contain specific values. Since set lookups are much faster than array lookups, this commit changes the internal structure to a set for the values stored in Target and SegmentTarget. * Add secure mode hash to contract tests (#229) * Update big segment support for users to context (#226) To support the users to context change for big segments, this commit makes the following changes: - Introduces a new `Segment.unboundedContextKind` attribute. This will default to `LDContext::KIND_DEFAULT` and is only referenced when `Segment.unbounded` is true. - With the creation of multi-kind contexts, a single evaluation may result in multiple queries to the big segment store. This is reflected in the changes to the `EvalResult` processing. * Drop support for ruby 2.6 (#227) Ruby 2.6 went EOL in March 2022. We originally didn't drop support for it as doing so would require dropping support for jRuby as well. However, jRuby recently released 9.4 which is Ruby 2.7+ compatible. * Update remaining references from user to contexts (#231) There are multiple places throughout the code where we are still referencing users. I have tried to update all the places where a rename seems reasonable or appropriate. There is still some work to do in the test flag builders, but that will be done in a subsequent commit. * Remove new relic integration (#233) The new relic integration was removed many versions ago but a small trace remained behind. * Rename config option private_attribute_names (#234) Co-authored-by: Eli Bishop <eli@launchdarkly.com> * Update test data integration to support contexts (#232) * improve data model validation logging; allow missing/empty attribute for segmentMatch (#236) * Fix JSON serialization failure (#237) When we introduced models for the flag and segment data, we added to_json methods which proxy to the underlying to_json method provided by the `json` gem. We defined the method with a variadic parameter, but we failed to unpack it when passing it on to the underlying implementation. This resulted in a serialization failure which prevented the redis data store from initializing. * fix: Bump eventsource to resolve header parsing (#239) The underlying event source library had an issue where, in certain environments, content-type header detection was failing. This was resolved in v2.2.2 of the event source gem. * Add key to error log for invalid context during variation call for easier debugging (#214) * feat: Add support for payload filtering (#238) * feat: Add data source status provider support (#240) The client instance will now provide access to a `data_source_status_provider`. This provider allows developers to retrieve the status of the SDK on demand, or asynchronously by registering listeners. * test: Skip database integration tests by default (#241) The full unit test suite includes tests relying on externally instances of Redis, Consult, and DynamoDB. By default, we do not want to run these tests. Rather, they should be opt-in by setting the environment variable `LD_SKIP_DATABASE_TESTS=0`. * feat: Introduce flag change tracker api (#242) The client instance will now provide access to a `flag_tracker`. This tracker allows developers to be notified when a flag configuration changes (or optionally when the /value/ of a flag changes for a particular context). * ci: Add code coverage generation (#244) * feat: Add support for data store status monitoring (#243) The client instance will now provide access to a `data_store_status_provider`. This provider allows developers to retrieve the data store status of the SDK on demand, or asynchronously by registering listeners. * Fix monitoring_enabled? access --------- Co-authored-by: Eli Bishop <eli@launchdarkly.com> Co-authored-by: LaunchDarklyCI <dev@launchdarkly.com> Co-authored-by: Jacob Smith <jacob@jacobsmith.io> Co-authored-by: Elliot <35050275+Apache-HB@users.noreply.github.com> Co-authored-by: Ben Woskow <bwoskow@launchdarkly.com> Co-authored-by: Ben Woskow <48036130+bwoskow-ld@users.noreply.github.com> Co-authored-by: hroederld <hroeder@launchdarkly.com> Co-authored-by: Kerrie Martinez <kyee@launchdarkly.com> Co-authored-by: pellyg-ld <gpelly@launchdarkly.com> Co-authored-by: Sam Stokes <sstokes@launchdarkly.com> Co-authored-by: LaunchDarklyReleaseBot <launchdarklyreleasebot@launchdarkly.com> Co-authored-by: Ember Stevens <ember.stevens@launchdarkly.com> Co-authored-by: ember-stevens <79482775+ember-stevens@users.noreply.github.com> Co-authored-by: Louis Chan <91093020+louis-launchdarkly@users.noreply.github.com> Co-authored-by: Matthew M. Keeler <keelerm84@gmail.com> Co-authored-by: Ben Levy <benjaminlevy007@gmail.com> Co-authored-by: Ben Levy <ben@foxhound.systems> Co-authored-by: Matthew M. Keeler <mkeeler@launchdarkly.com> Co-authored-by: Louis Chan <lchan@launchdarkly.com> Co-authored-by: Matt Hooks <46452201+matt-dutchie@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Is this a support request?
No.
Describe the bug
hi, I have nearly burnt (physically) my mac because of launch-darkly. How it happened:
Lucky it didn’t happen inside the case…..
To reproduce
Guess switch off internet totally (or block DNS) but keep wifi connected.
Expected behavior
Don't burn CPU
Logs
SDK version
ldclient-rb (5.5.6)
OS/platform
Mac Catalina
The text was updated successfully, but these errors were encountered: