flagsmith / flagsmith-ruby-client Goto Github PK

Hello! I'm struggling to use version 4 of this gem.
I'm somewhat newish to ruby so please excuse me if I missed something.

I've added the gem to my application using bundle add flagsmith.
Afterwards the application fails to launch because the gem pry can't be found.

Looking the gemspec it seems that pry and pry-byebug are only defined as a dev dependency and thus should not be required.
However both gems are required in a way that seems like they are not optional by lib/flagsmith.rb

When the EnvironmentPollingManager thread fails to update the environment document in one instance, the thread exits and will not retry to update the document in the future if the error which caused the environment update to fail is resolved.

We wanted to call client.environment_data_polling_manager.stop manually in our code to prevent networking requests during tests. Doing this though causes networking requests do to the line at https://github.com/Flagsmith/flagsmith-ruby-client/blob/main/lib/flagsmith.rb#L106. This seems to make sense for initial calls to environment_data_polling_manager but I would expect subsequent calls to environment_data_polling_manager to return the initialized manager.

Reference PR: Flagsmith/flagsmith#3622

1. Extend the Environment model with the identity_overrides: List[IdentityModel] field.
2. On environment update in local evaluation mode, store overrides so they're efficiently accessed by identifier.
3. For build_indentity_model interface (it's now more appropriate to call it get_identity_model so please rename it to that to match the other changes in the other sdks), use the storage above to retrieve the identity overrides. Fall back to a new IdentityModel instance if not found. If found, update traits with user-provided traits.

Refer to the following existing implementations:

Flagsmith/flagsmith-python-client#72
Flagsmith/flagsmith-java-client#142
Flagsmith/flagsmith-nodejs-client#143

It seems as though, even in local evaluation mode, when evaluating the flags and having flag analytics enabled, an exception thrown when posting the analytics data to the server can cause the client to return an error / default flag.

Update the get_identity_flags interface to support

• An optional boolean transient argument to denote a transient identity
• An optional string list transient_traits argument when supplying traits

Update the SDK API calls to add values according to updated schema here Flagsmith/flagsmith#4279

Given an environment with a flag titled test_mv which has a control value and a variation which has a weighting of 100% running the following script should return the variation value each time, but seemingly returns the control value.

require "flagsmith"

flagsmith = Flagsmith::Client.new(
  environment_key: "<key>",
  enable_local_evaluation: true,
)

identifier = "{Random.rand}"
flags = flagsmith.get_identity_flags(identifier)

puts flags.get_feature_value("test_mv")

Are there plans to publish a version of the gem with the new name?

Hi, the source seems to suggest that when listing all flags, the respective segments will be returned as well https://github.com/BulletTrainHQ/bullet-train-ruby-client/blob/master/lib/bullet_train.rb#L26

In practice, it doesn't seem to be the case, is that expected? This is our setup:

And this is the partial response for get_flags, where segment is nil:

...
"test_segments" => {
        :enabled => false,
          :value => nil,
        :segment => nil
    }
...

Context

When the following conditions are met, the local environment of the rails server does not refresh the flag states...the environment document request actually happens, tough.

• rails app
• puma (on cluster mode - even with 1 worker 1 thread)
• local evaluation enabled
• booting with rails s command

Steps to reproduce

• Create a new rails app
• Configure puma with cluster mode (set a value above zero for workers on config/puma.rb)
• Create an initializer, defining the global variable, just like the documentation shows. Local evaluation must be enabled
• boot the server with rails s
• check if the flagsmith client is responding to flag state changes on the flagsmith dashboard between every environment document request

Useful info

• Works normally with puma on single mode
• Puma worker and thread count does not matter to the issue
• On puma cluster mode, it seems like there is something wrong with the puma boot using the bundle exec rails s command, since running bundle exec puma -t 1:1 -w 1:1 looks fine
• Worth mentioning that the local evaluation on the rails c console, the normal behavior occurs
• Tested on macOs Ventura 13.6.7 (M1 chip) and Ubuntu 22 (x64)

Example project

https://github.com/Guilherme2112/flagsmith-rails-example

(basic instructions at README)

Hi, I'm trying to use flagsmith but when I call Flagsmith::Client.new(....) throws me the next error

Traceback (most recent call last):
16: from (irb):8
15: from /bundle/ruby/2.7.0/gems/activesupport-6.1.3.2/lib/active_support/dependencies.rb:332:in require' 14: from /bundle/ruby/2.7.0/gems/activesupport-6.1.3.2/lib/active_support/dependencies.rb:299:in load_dependency'
13: from /bundle/ruby/2.7.0/gems/activesupport-6.1.3.2/lib/active_support/dependencies.rb:332:in block in require' 12: from /bundle/ruby/2.7.0/gems/zeitwerk-2.6.7/lib/zeitwerk/kernel.rb:38:in require'
11: from /bundle/ruby/2.7.0/gems/zeitwerk-2.6.7/lib/zeitwerk/kernel.rb:38:in require' 10: from /bundle/ruby/2.7.0/gems/flagsmith-3.1.1/lib/flagsmith.rb:4:in <top (required)>'
9: from /bundle/ruby/2.7.0/gems/activesupport-6.1.3.2/lib/active_support/dependencies.rb:332:in require' 8: from /bundle/ruby/2.7.0/gems/activesupport-6.1.3.2/lib/active_support/dependencies.rb:299:in load_dependency'
7: from /bundle/ruby/2.7.0/gems/activesupport-6.1.3.2/lib/active_support/dependencies.rb:332:in block in require' 6: from /bundle/ruby/2.7.0/gems/zeitwerk-2.6.7/lib/zeitwerk/kernel.rb:38:in require'
5: from /bundle/ruby/2.7.0/gems/zeitwerk-2.6.7/lib/zeitwerk/kernel.rb:38:in require' 4: from /bundle/ruby/2.7.0/gems/faraday-retry-1.0.3/lib/faraday/retry.rb:3:in <top (required)>'
3: from /bundle/ruby/2.7.0/gems/faraday-retry-1.0.3/lib/faraday/retry.rb:3:in require_relative' 2: from /bundle/ruby/2.7.0/gems/faraday-retry-1.0.3/lib/faraday/retriable_response.rb:4:in <top (required)>'
1: from /bundle/ruby/2.7.0/gems/faraday-retry-1.0.3/lib/faraday/retriable_response.rb:6:in `module:Faraday'
TypeError (superclass mismatch for class RetriableResponse)

Any thoughts? Thank you in advance for your help

Create a provider for the Ruby OpenFeature SDK.

As per reference implementations in Java, .NET and Go we should create a provider for the Ruby OpenFeature SDK.

The provider code should live in the ruby-sdk-contrib repository.