🔀 Solid::Result

Unleash a pragmatic and observable use of Result Pattern and Railway-Oriented Programming in Ruby.

It's a general-purpose result monad that allows you to create objects representing a success (Solid::Result::Success) or failure (Solid::Result::Failure).

What problem does it solve?

It allows you to consistently represent the concept of success and failure throughout your codebase.

Furthermore, this abstraction exposes several features that will be useful to make the application flow react cleanly and securely to the result represented by these objects.

Use it to enable the Railway Oriented Programming pattern (superpower) in your code.

• Supported Ruby
• Installation
• Usage
  • Solid::Result versus Result
• Reference
  • Basic methods
  • Checking types with result.is? or method missing
  • Checking types with result.success? or result.failure?
  • Result Hooks
  • result.on
  • result.on_type
  • result.on_success
  • result.on_failure
  • result.on_unknown
  • result.handle
  • Result Value
  • result.value_or
  • Result Data
  • result.data
  • Railway Oriented Programming
  • result.and_then
  • Solid::Result.mixin
    • Class example (Instance Methods)
    • Module example (Singleton Methods)
    • Important Requirement
    • Dependency Injection
    • Add-ons
  • Solid::Result::Expectations
  • Standalone versus Mixin mode
  • Type checking - Result Hooks
    • #success? and #failure?
    • #on and #on_type
    • #on_success and #on_failure
    • #handle
  • Type checking - Result Creation
    • Mixin mode
    • Standalone mode
  • Value checking - Result Creation
    • Success()
    • Failure()
    • Pattern Matching Support
  • Solid::Result::Expectations.mixin add-ons
  • Solid::Output
  • Defining successes and failures
  • Hash methods
  • Solid::Output.mixin
    • Class example (Instance Methods)
    • and_expose
    • Module example (Singleton Methods)
  • Solid::Output::Expectations
  • Mixin add-ons
• Pattern Matching
  • Solid::Result
  • Array/Find patterns
  • Hash patterns
  • Solid::Output
  • Array/Find patterns
  • Hash patterns
  • How to pattern match without the concept of success and failure
• Solid::Result.event_logs
  • metadata: {ids:}
  • Configuration
  • Turning on/off
  • Setting a trace_id fetcher
  • Setting a listener
  • Setting multiple listeners
• Solid::Result.configuration
  • Solid::Result.config
• Solid::Result#and_then!
  • Dependency Injection
  • Configuration
  • Analysis: Why is and_then! an Anti-pattern?
  • #and_then versus #and_then!
  • Analysis: Why is #and_then the antidote/standard?
• About
• Development
• Contributing
• License
• Code of Conduct

Supported Ruby

This library is tested against:

Version	2.7	3.0	3.1	3.2	3.3	Head
100% Coverage	✅	✅	✅	✅	✅	✅

Installation

Add this line to your application's Gemfile:

gem 'solid-result'

And then execute:

$ bundle install

If bundler is not being used to manage dependencies, install the gem by executing:

$ gem install solid-result

And require it in your code:

require 'solid/result'

⬆️  back to top

Usage

To create a result, you must define a type (symbol) and its value (any kind of object). e.g.,

Solid::Result::Success(:ok, :1)           #
                                          # The value can be any kind of object
Solid::Result::Failure(:err, 'the value') #

The reason for defining a type is that it is very common for a method/operation to return different types of successes or failures. Because of this, the type will always be required. e,g.,

Solid::Result::Success(:ok)  #
                             # The type is mandatory and the value is optional
Solid::Result::Failure(:err) #

⬆️  back to top

Solid::Result versus Result

This gem provides a way to create constant aliases for Solid::Result and other classes/modules.

To enable it, you must call the Solid::Result.configuration method and pass a block to it. You can turn the aliases you want on/off in this block.

Solid::Result.configuration do |config|
  config.constant_alias.enable!('Result')
end

So, instead of using Solid::Result everywhere, you can use Result as an alias/shortcut.

Result::Success(:ok) # <Solid::Result::Success type=:ok value=nil>

Result::Failure(:err) # <Solid::Result::Failure type=:err value=nil>

If you have enabled constant aliasing, all examples in this README that use Solid::Result can be implemented using Result.

There are other aliases and configurations available. Check the Solid::Result.configuration section for more information.

⬆️  back to top

Reference

Basic methods

Both Solid::Result::Success and Solid::Result::Failure are composed of the same methods. Look at the basic ones:

Solid::Result::Success

################
# With a value #
################
result = Solid::Result::Success(:ok, my: 'value')

result.success?   # true
result.failure?   # false
result.type?(:ok) # true
result.type       # :ok
result.value      # {:my => "value"}

###################
# Without a value #
###################
result = Solid::Result::Success(:yes)

result.success?    # true
result.failure?    # false
result.type?(:yes) # true
result.type        # :yes
result.value       # nil

Solid::Result::Failure

################
# With a value #
################
result = Solid::Result::Failure(:err, 'my_value')

result.success?    # false
result.failure?    # true
result.type?(:err) # true
result.type        # :err
result.value       # "my_value"

###################
# Without a value #
###################
result = Solid::Result::Failure(:no)

result.success?   # false
result.failure?   # true
result.type?(:no) # true
result.type       # :no
result.value      # nil

In both cases, the type must be a symbol, and the value can be any kind of object.

⬆️  back to top

Checking types with result.is? or method missing

Beyond the type? method, you can also use the is? method to check the result type. If you want to check the type directly, you can write the type using a method that ends with a question mark.

result = Solid::Result::Success(:ok)

result.is?(:ok) # true
result.ok?      # true

result = Solid::Result::Failure(:err)

result.is?(:err) # true
result.err?      # true

⬆️  back to top

Checking types with result.success? or result.failure?

Solid::Result#success? and Solid::Result#failure? are methods that allow you to check if the result is a success or a failure.

You can also check the result type by passing an argument to it. For example, result.success?(:ok) will check if the result is a success and if the type is :ok.

result = Solid::Result::Success(:ok)

result.success?(:ok)

# This is the same as:

result.success? && result.type == :ok

The same is valid for Solid::Result#failure?.

result = Solid::Result::Failure(:err)

result.failure?(:err)

# This is the same as:

result.failure? && result.type == :err

⬆️  back to top

Result Hooks

Result hooks are methods that allow you to execute a block of code based on the type of result obtained. To demonstrate their use, I will implement a method that can divide two numbers.

def divide(arg1, arg2)
  arg1.is_a?(::Numeric) or return Solid::Result::Failure(:invalid_arg, 'arg1 must be numeric')
  arg2.is_a?(::Numeric) or return Solid::Result::Failure(:invalid_arg, 'arg2 must be numeric')

  return Solid::Result::Failure(:division_by_zero, 'arg2 must not be zero') if arg2.zero?

  Solid::Result::Success(:division_completed, arg1 / arg2)
end

⬆️  back to top

result.on

When you use Solid::Result#on, the block will be executed only when the type matches the result type.

However, even if the block is executed, the method will always return the result itself.

The value of the result will be available as the first argument of the block.

result = divide(nil, 2)
#<Solid::Result::Failure type=:invalid_arg data='arg1 must be numeric'>

output =
  result
    .on(:invalid_arg) { |msg| puts msg }
    .on(:division_by_zero) { |msg| puts msg }
    .on(:division_completed) { |number| puts number }

# The code above will print 'arg1 must be numeric' and return the result itself.

result.object_id == output.object_id # true

You can define multiple types to be handled by the same hook/block

result = divide(4, 0)

output =
  result.on(:invalid_arg, :division_by_zero, :division_completed) { |value| puts value }

# The code above will print 'arg2 must not be zero' and return the result itself.

result.object_id == output.object_id # true

PS: The divide() implementation is here.

⬆️  back to top

result.on_type

Solid::Result#on_type is an alias of Solid::Result#on.

result = divide(nil, 2)
#<Solid::Result::Failure type=:invalid_arg data='arg1 must be numeric'>

output =
  result
    .on_type(:invalid_arg, :division_by_zero) { |msg| puts msg }
    .on_type(:division_completed) { |number| puts number }

# The code above will print 'arg1 must be numeric' and return the result itself.

result.object_id == output.object_id # true

PS: The divide() implementation is here.

⬆️  back to top

result.on_success

The Solid::Result#on_success method is quite similar to the Solid::Result#on hook, but with a few key differences:

1. It will only execute the block of code if the result is a success.
2. If the type declaration is not included, the method will execute the block for any successful result, regardless of its type.

# In both examples, it executes the block and returns the result itself.

divide(4, 2).on_success { |number| puts number }

divide(4, 2).on_success(:division_completed) { |number| puts number }

# It doesn't execute the block as the type is different.

divide(4, 4).on_success(:ok) { |value| puts value }

# It doesn't execute the block, as the result is a success, but the hook expects a failure.

divide(4, 4).on_failure { |error| puts error }

PS: The divide() implementation is here.

⬆️  back to top

result.on_failure

It is the opposite of Result#on_success:

1. It will only execute the block of code if the result is a failure.
2. If the type declaration is not included, the method will execute the block for any failed result, regardless of its type.

# In both examples, it executes the block and returns the result itself.

divide(nil, 2).on_failure { |error| puts error }

divide(4, 0).on_failure(:division_by_zero) { |error| puts error }

# It doesn't execute the block as the type is different.

divide(4, 0).on_failure(:invalid_arg) { |error| puts error }

# It doesn't execute the block, as the result is a failure, but the hook expects a success.

divide(4, 0).on_success { |number| puts number }

PS: The divide() implementation is here.

⬆️  back to top

result.on_unknown

Solid::Result#on_unknown will execute the block when no other hook (#on, #on_type, #on_failure, #on_success) has been executed.

Regardless of the block being executed, the method will always return the result itself.

The value of the result will be available as the first argument of the block.

divide(4, 2)
  .on(:invalid_arg) { |msg| puts msg }
  .on(:division_by_zero) { |msg| puts msg }
  .on_unknown { |value, type| puts [type, value].inspect }

# The code above will print '[:division_completed, 2]' and return the result itself.

PS: The divide() implementation is here.

⬆️  back to top

result.handle

This method lets you define blocks for each hook (type, failure, or success), but instead of returning itself, it will return the output of the first match/block execution.

divide(4, 2).handle do |result|
  result.success { |number| number }
  result.failure(:invalid_arg) { |err| puts err }
  result.type(:division_by_zero) { raise ZeroDivisionError }
  result.unknown { raise NotImplementedError }
end

#or

divide(4, 2).handle do |on|
  on.success { |number| number }
  on.failure { |err| puts err }
  on.unknown { raise NotImplementedError }
end

#or

divide(4, 2).handle do |on|
  on.type(:invalid_arg) { |err| puts err }
  on.type(:division_by_zero) { raise ZeroDivisionError }
  on.type(:division_completed) { |number| number }
  on.unknown { raise NotImplementedError }
end

# or

divide(4, 2).handle do |on|
  on[:invalid_arg] { |err| puts err }
  on[:division_by_zero] { raise ZeroDivisionError }
  on[:division_completed] { |number| number }
  on.unknown { raise NotImplementedError }
end

# The [] syntax 👆 is an alias of #type.

Notes:

• You can define multiple types to be handled by the same hook/block
• If the type is missing, it will execute the block for any success or failure handler.
• The #type and #[] handlers require at least one type/argument.

PS: The divide() implementation is here.

⬆️  back to top

Result Value

To access the result value, you can simply call Solid::Result#value.

However, there may be instances where you need to retrieve the value of a successful result or a default value if the result is a failure. In such cases, you can make use of Solid::Result#value_or.

result.value_or

BCCD::Result#value_or returns the value when the result is a success. However, if it is a failure, the given block will be executed, and its outcome will be returned.

def divide(arg1, arg2)
  arg1.is_a?(::Numeric) or return Solid::Result::Failure(:invalid_arg)
  arg2.is_a?(::Numeric) or return Solid::Result::Failure(:invalid_arg)

  return Solid::Result::Failure(:division_by_zero) if arg2.zero?

  Solid::Result::Success(:division_completed, arg1 / arg2)
end

# When the result is success
divide(4, 2).value_or { 0 } # 2

# When the result is failure
divide('4', 2).value_or { 0 } # 0
divide(4, '2').value_or { 0 } # 0
divide(100, 0).value_or { 0 } # 0

PS: The divide() implementation is here.

⬆️  back to top

Result Data

result.data

The Solid::Result#data exposes the result attributes (kind, type, value) directly and as a hash (to_h/to_hash) and array (to_a/to_ary).

This is helpful if you need to access the result attributes generically or want to use Ruby features like splat (*) and double splat (**) operators.

See the examples below to understand how to use it.

result = Solid::Result::Success(:ok, 1)

success_data = result.data # #<Solid::Result::Data kind=:success type=:ok value=1>

success_data.kind  # :success
success_data.type  # :ok
success_data.value # 1

success_data.to_h  # {:kind=>:success, :type=>:ok, :value=>1}
success_data.to_a  # [:success, :ok, 1]

kind, type, value = success_data

[kind, type, value] # [:success, :ok, 1]

def print_to_ary(kind, type, value)
  puts [kind, type, value].inspect
end

def print_to_hash(kind:, type:, value:)
  puts [kind, type, value].inspect
end

print_to_ary(*success_data)   # [:success, :ok, 1]

print_to_hash(**success_data) # [:success, :ok, 1]

NOTE: The example above uses a success result, but the same is valid for a failure result.

⬆️  back to top

Railway Oriented Programming

"Railway Oriented Programming (ROP)" is a programming technique that involves linking blocks together to form a sequence of operations, also known as a pipeline. If a failure occurs in any of the blocks, the pipeline is interrupted and subsequent blocks are skipped.

The ROP technique allows you to structure your code in a way that expresses your logic as a series of operations, with the added benefit of stopping the process at the first detection of failure.

If all blocks successfully execute, the final result of the pipeline will be a success.

result.and_then

module Divide
  extend self

  def call(arg1, arg2)
    validate_numbers(arg1, arg2)
      .and_then { |numbers| validate_nonzero(numbers) }
      .and_then { |numbers| divide(numbers) }
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Solid::Result::Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Solid::Result::Failure(:invalid_arg, 'arg2 must be numeric')

    Solid::Result::Success(:ok, [arg1, arg2])
  end

  def validate_nonzero(numbers)
    return Solid::Result::Success(:ok, numbers) if numbers.last.nonzero?

    Solid::Result::Failure(:division_by_zero, 'arg2 must not be zero')
  end

  def divide((number1, number2))
    Solid::Result::Success(:division_completed, number1 / number2)
  end
end

Example of outputs:

Divide.call('4', 2)
#<Solid::Result::Failure type=:invalid_arg data="arg1 must be numeric">

Divide.call(2, '2')
#<Solid::Result::Failure type=:invalid_arg data="arg2 must be numeric">

Divide.call(2, 0)
#<Solid::Result::Failure type=:division_by_zero data="arg2 must not be zero">

Divide.call(2, 2)
#<Solid::Result::Success type=:division_completed data=1>

⬆️  back to top

Solid::Result.mixin

This method generates a module that any object can include or extend. It adds two methods to the target object: Success() and Failure().

The main difference between these methods and Solid::Result::Success()/Solid::Result::Failure() is that the former will utilize the target object (which has received the include/extend) as the result's source.

Because the result has a source, the #and_then method can call methods from it.

Class example (Instance Methods)

class Divide
  include Solid::Result.mixin

  attr_reader :arg1, :arg2

  def initialize(arg1, arg2)
    @arg1 = arg1
    @arg2 = arg2
  end

  def call
    validate_numbers
      .and_then(:validate_nonzero)
      .and_then(:divide)
  end

  private

  def validate_numbers
    arg1.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    # As arg1 and arg2 are instance methods, they will be available in the instance scope.
    # So, in this case, I'm passing them as an array to show how the next method can receive the value as its argument.
    Success(:ok, [arg1, arg2])
  end

  def validate_nonzero(numbers)
    return Success(:ok, numbers) unless numbers.last.zero?

    Failure(:division_by_zero, 'arg2 must not be zero')
  end

  def divide((number1, number2))
    Success(:division_completed, number1 / number2)
  end
end

Divide.new(4, 2).call #<Solid::Result::Success type=:division_completed value=2>

Divide.new(4, 0).call   #<Solid::Result::Failure type=:division_by_zero value="arg2 must not be zero">
Divide.new('4', 2).call #<Solid::Result::Failure type=:invalid_arg value="arg1 must be numeric">
Divide.new(4, '2').call #<Solid::Result::Failure type=:invalid_arg value="arg2 must be numeric">

Module example (Singleton Methods)

module Divide
  extend self, Solid::Result.mixin

  def call(arg1, arg2)
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero)
      .and_then(:divide)
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    Success(:ok, [arg1, arg2])
  end

  def validate_nonzero(numbers)
    return Success(:ok, numbers) unless numbers.last.zero?

    Failure(:division_by_zero, 'arg2 must not be zero')
  end

  def divide((number1, number2))
    Success(:division_completed, number1 / number2)
  end
end

Divide.call(4, 2) #<Solid::Result::Success type=:division_completed value=2>

Divide.call(4, 0)   #<Solid::Result::Failure type=:division_by_zero value="arg2 must not be zero">
Divide.call('4', 2) #<Solid::Result::Failure type=:invalid_arg value="arg1 must be numeric">
Divide.call(4, '2') #<Solid::Result::Failure type=:invalid_arg value="arg2 must be numeric">

⬆️  back to top

Important Requirement

To use the #and_then method to call methods, they must use Success() and Failure() to produce the results.

If you try to use Solid::Result::Success()/Solid::Result::Failure(), or results from another Solid::Result.mixin instance with #and_then, it will raise an error because the sources are different.

Note: You can still use the block syntax, but all the results must be produced by the source's Success() and Failure() methods.

module ValidateNonzero
  extend self, Solid::Result.mixin

  def call(numbers)
    return Success(:ok, numbers) unless numbers.last.zero?

    Failure(:division_by_zero, 'arg2 must not be zero')
  end
end

module Divide
  extend self, Solid::Result.mixin

  def call(arg1, arg2)
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero)
      .and_then(:divide)
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    Success(:ok, [arg1, arg2])
  end

  def validate_nonzero(numbers)
    ValidateNonzero.call(numbers) # This will raise an error
  end

  def divide((number1, number2))
    Success(:division_completed, number1 / number2)
  end
end

Look at the error produced by the code above:

Divide.call(2, 0)

# You cannot call #and_then and return a result that does not belong to the same source! (Solid::Result::Error::InvalidResultSource)
# Expected source: Divide
# Given source: ValidateNonzero
# Given result: #<Solid::Result::Failure type=:division_by_zero value="arg2 must not be zero">

In order to fix this, you must handle the result produced by ValidateNonzero.call() and return a result that belongs to the same source.

module ValidateNonzero
  extend self, Solid::Result.mixin

  def call(numbers)
    return Success(:ok, numbers) unless numbers.last.zero?

    Failure(:division_by_zero, 'arg2 must not be zero')
  end
end

module Divide
  extend self, Solid::Result.mixin

  def call(arg1, arg2)
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero)
      .and_then(:divide)
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    Success(:ok, [arg1, arg2])
  end

  def validate_nonzero(numbers)
    # In this case we are handling the result from other source
    # and returning our own
    ValidateNonzero.call(numbers).handle do |on|
      on.success { |numbers| Success(:ok, numbers) }

      on.failure { |err| Failure(:division_by_zero, err) }
    end
  end

  def divide((number1, number2))
    Success(:division_completed, number1 / number2)
  end
end

Look at the output of the code above:

Divide.call(2, 0)

#<Solid::Result::Failure type=:division_by_zero value="arg2 must not be zero">

⬆️  back to top

Dependency Injection

The Solid::Result#and_then accepts a second argument that will be used to share a value with the source's method. To receive this argument, the source's method must have an arity of two, where the first argument will be the result value and the second will be the injected value.

require 'logger'

module Divide
  extend self, Solid::Result.mixin

  def call(arg1, arg2, logger: ::Logger.new(STDOUT))
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero, logger)
      .and_then(:divide, logger)
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    Success(:ok, [arg1, arg2])
  end

  def validate_nonzero(numbers, logger)
    if numbers.last.zero?
      logger.error('arg2 must not be zero')

      Failure(:division_by_zero, 'arg2 must not be zero')
    else
      logger.info('The numbers are valid')

      Success(:ok, numbers)
    end
  end

  def divide((number1, number2), logger)
    division = number1 / number2

    logger.info("The division result is #{division}")

    Success(:division_completed, division)
  end
end

Divide.call(4, 2)
# I, [2023-10-11T00:08:05.546237 #18139]  INFO -- : The numbers are valid
# I, [2023-10-11T00:08:05.546337 #18139]  INFO -- : The division result is 2
#=> #<Solid::Result::Success type=:division_completed value=2>

Divide.call(4, 2, logger: Logger.new(IO::NULL))
#=> #<Solid::Result::Success type=:division_completed value=2>

⬆️  back to top

Add-ons

The Solid::Result.mixin also accepts the config: argument. It is a hash that will be used to define custom behaviors for the mixin.

given

This addon is enabled by default. It will create the Given(value) method. Use it to add a value to the result chain and invoke the next step (through and_then).

You can turn it off by passing given: false to the config: argument or using the Solid::Result.configuration.

continue

This addon will create the Continue(value) method and change the Success() behavior to terminate the step chain.

So, if you want to advance to the next step, you must use Continue(value) instead of Success(type, value). Otherwise, the step chain will be terminated.

In this example below, the validate_nonzero will return a Success(:division_completed, 0) and terminate the chain if the first number is zero.

module Divide
  extend self, Solid::Result.mixin(config: { addon: { continue: true } })

  def call(arg1, arg2)
    Given([arg1, arg2])
      .and_then(:validate_numbers)
      .and_then(:validate_nonzero)
      .and_then(:divide)
  end

  private

  def validate_numbers(numbers)
    number1, number2 = numbers

    number1.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    number2.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    Continue(numbers)
  end

  def validate_nonzero(numbers)
    return Failure(:division_by_zero, 'arg2 must not be zero') if numbers.last.zero?

    return Success(:division_completed, 0) if numbers.first.zero?

    Continue(numbers)
  end

  def divide((number1, number2))
    Success(:division_completed, number1 / number2)
  end
end

⬆️  back to top

Solid::Result::Expectations

This feature lets you define contracts for your results' types and values. There are two ways to use it: the standalone (Solid::Result::Expectations.new) and the mixin (Solid::Result::Expectations.mixin) mode.

It was designed to ensure all the aspects of the result's type and value. So, an error will be raised if you try to create or handle a result with an unexpected type or value.

Standalone versus Mixin mode

The standalone mode creates an object that knows how to produce and validate results based on the defined expectations. Look at the example below:

module Divide
  Result = Solid::Result::Expectations.new(
    success: %i[numbers division_completed],
    failure: %i[invalid_arg division_by_zero]
  )

  def self.call(arg1, arg2)
    arg1.is_a?(Numeric) or return Result::Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(Numeric) or return Result::Failure(:invalid_arg, 'arg2 must be numeric')

    arg2.zero? and return Result::Failure(:division_by_zero, 'arg2 must not be zero')

    Result::Success(:division_completed, arg1 / arg2)
  end
end

In the code above, we define a constant Divide::Result. And because of this (it is a constant), we can use it inside and outside the module.

Look what happens if you try to create a result without one of the expected types.

Divide::Result::Success(:ok)
# type :ok is not allowed. Allowed types: :numbers, :division_completed
# (Solid::Result::Contract::Error::UnexpectedType)

Divide::Result::Failure(:err)
# type :err is not allowed. Allowed types: :invalid_arg, :division_by_zero
# (Solid::Result::Contract::Error::UnexpectedType)

The mixin mode is similar to Solid::Result::Mixin, but it also defines the expectations for the result's types and values.

class Divide
  include Solid::Result::Expectations.mixin(
    success: %i[numbers division_completed],
    failure: %i[invalid_arg division_by_zero]
  )

  def call(arg1, arg2)
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero)
      .and_then(:divide)
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    Success(:numbers, [arg1, arg2])
  end

  def validate_nonzero(numbers)
    return Success(:numbers, numbers) unless numbers.last.zero?

    Failure(:division_by_zero, 'arg2 must not be zero')
  end

  def divide((number1, number2))
    Success(:division_completed, number1 / number2)
  end
end

This mode also defines an Result constant to be used inside and outside the module.

PROTIP: You can use the Result constant to mock the result's type and value in your tests. As they will have the exact expectations, your tests will check if the result clients are handling the result correctly.

Now that you know the two modes, let's understand how expectations can be beneficial and powerful for defining contracts.

⬆️  back to top

Type checking - Result Hooks

The Solid::Result::Expectations will check if the type of the result is valid. This checking will be performed in all methods that depend on the result’s type, such as #success?, #failure?, #on, #on_type, #on_success, #on_failure, and #handle.

#success? and #failure?

When checking whether a result is a success or failure, Solid::Result::Expectations will also verify if the result type is valid/expected. In case of an invalid type, an error will be raised.

Success example:

result = Divide.new.call(10, 2)

result.success?                      # true
result.success?(:numbers)            # false
result.success?(:division_completed) # true

result.success?(:ok)
# type :ok is not allowed. Allowed types: :numbers, :division_completed
# (Solid::Result::Contract::Error::UnexpectedType)

Failure example:

result = Divide.new.call(10, '2')

result.failure?                    # true
result.failure?(:invalid_arg)      # true
result.failure?(:division_by_zero) # false

result.failure?(:err)
# type :err is not allowed. Allowed types: :invalid_arg, :division_by_zero
# (Solid::Result::Contract::Error::UnexpectedType)

PS: The Divide implementation is here.

⬆️  back to top

#on and #on_type

If you use #on or #on_type to execute a block, Solid::Result::Expectations will check whether the result type is valid/expected. Otherwise, an error will be raised.

result = Divide.new.call(10, 2)

result
  .on(:invalid_arg, :division_by_zero) { |msg| puts msg }
  .on(:division_completed) { |number| puts "The result is #{number}" }

# The code above will print 'The result is 5'

result.on(:number) { |_| :this_type_does_not_exist }
# type :number is not allowed. Allowed types: :numbers, :division_completed, :invalid_arg, :division_by_zero
# (Solid::Result::Contract::Error::UnexpectedType)

PS: The Divide implementation is here.

⬆️  back to top

#on_success and #on_failure

If you use #on_success or #on_failure to execute a block, Solid::Result::Expectations will check whether the result type is valid/expected. Otherwise, an error will be raised.

result = Divide.new.call(10, '2')

result
  .on_failure(:invalid_arg, :division_by_zero) { |msg| puts msg }
  .on_success(:division_completed) { |number| puts "The result is #{number}" }

result
  .on_success { |number| puts "The result is #{number}" }
  .on_failure { |msg| puts msg }

# Both codes above will print 'arg2 must be numeric'

result.on_success(:ok) { |_| :this_type_does_not_exist }
# type :ok is not allowed. Allowed types: :numbers, :division_completed
# (Solid::Result::Contract::Error::UnexpectedType)

result.on_failure(:err) { |_| :this_type_does_not_exist }
# type :err is not allowed. Allowed types: :invalid_arg, :division_by_zero
# (Solid::Result::Contract::Error::UnexpectedType)

PS: The Divide implementation is here.

⬆️  back to top

#handle

The Solid::Result::Expectations will also be applied on all the handlers defined by the #handle method/block.

result = Divide.call(10, 2)

result.handle do |on|
  on.type(:ok) { |_| :this_type_does_not_exist }
end
# type :ok is not allowed. Allowed types: :numbers, :division_completed, :invalid_arg, :division_by_zero (Solid::Result::Contract::Error::UnexpectedType)

result.handle do |on|
  on.success(:ok) { |_| :this_type_does_not_exist }
end
# type :ok is not allowed. Allowed types: :numbers, :division_completed (Solid::Result::Contract::Error::UnexpectedType)

result.handle do |on|
  on.failure(:err) { |_| :this_type_does_not_exist }
end
# type :err is not allowed. Allowed types: :invalid_arg, :division_by_zero (Solid::Result::Contract::Error::UnexpectedType)

PS: The Divide implementation is here.

⬆️  back to top

Type checking - Result Creation

The Solid::Result::Expectations will be used on the result creation Success() and Failure() methods. So, when the result type is valid/expected, the result will be created. Otherwise, an error will be raised.

This works for both modes (standalone and mixin).

Mixin mode

module Divide
  extend Solid::Result::Expectations.mixin(success: :ok, failure: :err)

  def self.call(arg1, arg2)
    arg1.is_a?(Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    arg2.zero? and return Failure(:division_by_zero, 'arg2 must not be zero')

    Success(:division_completed, arg1 / arg2)
  end
end

Divide.call('4', 2)
# type :invalid_arg is not allowed. Allowed types: :err
# (Solid::Result::Contract::Error::UnexpectedType)

Divide.call(4, 2)
# type :division_completed is not allowed. Allowed types: :ok
# (Solid::Result::Contract::Error::UnexpectedType)

⬆️  back to top

Standalone mode

module Divide
  Result = Solid::Result::Expectations.new(success: :ok, failure: :err)

  def self.call(arg1, arg2)
    arg1.is_a?(Numeric) or return Result::Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(Numeric) or return Result::Failure(:invalid_arg, 'arg2 must be numeric')

    arg2.zero? and return Result::Failure(:division_by_zero, 'arg2 must not be zero')

    Result::Success(:division_completed, arg1 / arg2)
  end
end

Divide.call('4', 2)
# type :invalid_arg is not allowed. Allowed types: :err
# (Solid::Result::Contract::Error::UnexpectedType)

Divide.call(4, 2)
# type :division_completed is not allowed. Allowed types: :ok
# (Solid::Result::Contract::Error::UnexpectedType)

⬆️  back to top

Value checking - Result Creation

The Result::Expectations supports two types of validations. The first is the type checking only, and the second is the type and value checking.

To define expectations for your result's values, you must declare a Hash with the type as the key and the value as the value. A value validator is any object that responds to #=== (case equality operator).

Mixin mode:

module Divide
  extend Solid::Result::Expectations.mixin(
    success: {
      numbers: ->(value) { value.is_a?(Array) && value.size == 2 && value.all?(Numeric) },
      division_completed: Numeric
    },
    failure: {
      invalid_arg: String,
      division_by_zero: String
    }
  )

  def self.call(arg1, arg2)
    arg1.is_a?(Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    arg2.zero? and return Failure(:division_by_zero, 'arg2 must not be zero')

    Success(:division_completed, arg1 / arg2)
  end
end

Standalone mode:

module Divide
  Result = Solid::Result::Expectations.new(
    success: {
      numbers: ->(value) { value.is_a?(Array) && value.size == 2 && value.all?(Numeric) },
      division_completed: Numeric
    },
    failure: {
      invalid_arg: String,
      division_by_zero: String
    }
  )

  def self.call(arg1, arg2)
    arg1.is_a?(Numeric) or return Result::Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(Numeric) or return Result::Failure(:invalid_arg, 'arg2 must be numeric')

    arg2.zero? and return Result::Failure(:division_by_zero, 'arg2 must not be zero')

    Result::Success(:division_completed, arg1 / arg2)
  end
end

The value validation will only be performed through the methods Success() and Failure().

⬆️  back to top

Success()

Divide::Result::Success(:ok)
# type :ok is not allowed. Allowed types: :numbers, :division_completed (Solid::Result::Contract::Error::UnexpectedType)

Divide::Result::Success(:numbers, [1])
# value [1] is not allowed for :numbers type (Solid::Result::Contract::Error::UnexpectedValue)

Divide::Result::Success(:division_completed, '2')
# value "2" is not allowed for :division_completed type (Solid::Result::Contract::Error::UnexpectedValue)

Failure()

Divide::Result::Failure(:err)
# type :err is not allowed. Allowed types: :invalid_arg, :division_by_zero (Solid::Result::Contract::Error::UnexpectedType)

Divide::Result::Failure(:invalid_arg, :arg1_must_be_numeric)
# value :arg1_must_be_numeric is not allowed for :invalid_arg type (Solid::Result::Contract::Error::UnexpectedValue)

Divide::Result::Failure(:division_by_zero, msg: 'arg2 must not be zero')
# value {:msg=>"arg2 must not be zero"} is not allowed for :division_by_zero type (Solid::Result::Contract::Error::UnexpectedValue)

⬆️  back to top

Pattern Matching Support

The value checking has support for handling pattern-matching errors, and the cleanest way to do it is using the one-line pattern matching operators (=> since Ruby 3.0) and (in Ruby 2.7).

How does this operator work? They raise an error when the pattern does not match but returns nil when it matches.

Because of this, you will need to enable nil as a valid value checking. You can do it through the Solid::Result.configuration or by allowing it directly on the mixin config.

module Divide
  extend Solid::Result::Expectations.mixin(
    config: {
      pattern_matching: { nil_as_valid_value_checking: true }
    },
    success: {
      division_completed: ->(value) { value => (Integer | Float) }
    },
    failure: { invalid_arg: String, division_by_zero: String }
  )

  def self.call(arg1, arg2)
    arg1.is_a?(Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    arg2.zero? and return Failure(:division_by_zero, 'arg2 must not be zero')

    Success(:division_completed, String(arg1 / arg2))
  end
end

Divide.call(10, 5)
# value "2" is not allowed for :division_completed type ("2": Float === "2" does not return true) (Solid::Result::Contract::Error::UnexpectedValue)

⬆️  back to top

Solid::Result::Expectations.mixin add-ons

The Solid::Result::Expectations.mixin also accepts the config: argument. It is a hash that can be used to define custom behaviors for the mixin.

Continue

It is similar to Solid::Result.mixin(config: { addon: { continue: true } }). The key difference is that the expectations will ignore the Continue(value).

Based on this, use the Success() to produce a terminal result and Continue() to produce a result that will be used in the next step.

class Divide
  include Solid::Result::Expectations.mixin(
    config: { addon: { continue: true } },
    success: :division_completed,
    failure: %i[invalid_arg division_by_zero]
  )

  def call(arg1, arg2)
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero)
      .and_then(:divide)
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:invalid_arg, 'arg2 must be numeric')

    Continue([arg1, arg2])
  end

  def validate_nonzero(numbers)
    return Continue(numbers) unless numbers.last.zero?

    Failure(:division_by_zero, 'arg2 must not be zero')
  end

  def divide((number1, number2))
    Success(:division_completed, number1 / number2)
  end
end

result = Divide.new.call(4, 2)
# => #<Solid::Result::Success type=:division_completed value=2>

# The example below shows an error because the :ok type is not allowed.
# But look at the allowed types have only one type (:division_completed).
# This is because the :_continue_ type is ignored by the expectations.
#
result.success?(:ok)
# type :ok is not allowed. Allowed types: :division_completed (Solid::Result::Contract::Error::UnexpectedType)

⬆️  back to top

Solid::Output

The Solid::Output is a Solid::Result, meaning it has all the features of the Solid::Result. The main difference is that it only accepts keyword arguments as a value, which applies to the and_then: The called methods must receive keyword arguments, and the dependency injection will be performed through keyword arguments.

As the input/output are hashes, the results of each and_then call will automatically accumulate. This is useful in operations chaining, as the result of the previous operations will be automatically available for the next one. Because of this behavior, the Solid::Output has the #and_expose method to expose only the desired keys from the accumulated result.

Defining successes and failures

As the Solid::Result, you can declare success and failures directly from Solid::Output.

Solid::Output::Success(:ok, a: 1, b: 2)
#<Solid::Output::Success type=:ok value={:a=>1, :b=>2}>

Solid::Output::Failure(:err, message: 'something went wrong')
#<Solid::Output::Failure type=:err value={:message=>"something went wrong"}>

But different from Solid::Result that accepts any value, the Solid::Output only takes keyword arguments.

Solid::Output::Success(:ok, [1, 2])
# wrong number of arguments (given 2, expected 1) (ArgumentError)

Solid::Output::Failure(:err, { message: 'something went wrong' })
# wrong number of arguments (given 2, expected 1) (ArgumentError)

#
# Use ** to convert a hash to keyword arguments
#
Solid::Output::Success(:ok, **{ message: 'hashes can be converted to keyword arguments' })
#<Solid::Output::Success type=:ok value={:message=>"hashes can be converted to keyword arguments"}>

⬆️  back to top

Hash methods

The Solid::Output only accepts hashes as its values. Because of this, its instances have some Hash's methods to query/access the values. The available methods are:

• #slice to extract only the desired keys.
• #[], #dig, #fetch to access the values.
• #values_at and #fetch_values to get the values of the desired keys.

result = Solid::Output::Success(:ok, a: 1, b: 2, c: {d: 4})

result[:a]         # 1
result.fetch(:a)   # 1
result.dig(:c, :d) # 4

result.slice(:a, :b) # {:a=>1, :b=>2}

result.values_at(:a, :b) # [1, 2]
result.fetch_values(:a, :b) # [1, 2]

These methods are available for Solid::Output::Success and Solid::Output::Failure instances.

⬆️  back to top

Solid::Output.mixin

As in the Solid::Result, you can use the Solid::Output.mixin to add the Success() and Failure() methods to your classes/modules.

Let's see this feature and the data accumulation in action:

Class example (Instance Methods)

require 'logger'

class Divide
  include Solid::Output.mixin

  def call(arg1, arg2, logger: ::Logger.new(STDOUT))
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero)
      .and_then(:divide, logger: logger)
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Failure(:err, message: 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:err, message: 'arg2 must be numeric')

    Success(:ok, number1: arg1, number2: arg2)
  end

  def validate_nonzero(number2:, **)
    return Success(:ok) if number2.nonzero?

    Failure(:err, message: 'arg2 must not be zero')
  end

  #
  # The logger was injected via #and_then and keyword arguments
  #
  def divide(number1:, number2:, logger:)
    result = number1 / number2

    logger.info("The division result is #{result}")

    Success(:ok, number: result)
  end
end

Divide.new.call(10, 5)
# I, [2023-10-27T01:51:46.905004 #76915]  INFO -- : The division result is 2
#<Solid::Output::Success type=:ok value={:number=>2}>

Divide.new.call('10', 5)
#<Solid::Output::Failure type=:err value={:message=>"arg1 must be numeric"}>

Divide.new.call(10, '5')
#<Solid::Output::Failure type=:err value={:message=>"arg2 must be numeric"}>

Divide.new.call(10, 0)
#<Solid::Output::Failure type=:err value={:message=>"arg2 must not be zero"}>

⬆️  back to top

and_expose

This allows you to expose only the desired keys from the accumulated result. It can be used with any Solid::Output object.

Let's add it to the previous example:

class Divide
  include Solid::Output.mixin

  def call(arg1, arg2)
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero)
      .and_then(:divide)
      .and_expose(:division_completed, [:number])
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Failure(:err, message: 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:err, message: 'arg2 must be numeric')

    Success(:ok, number1: arg1, number2: arg2)
  end

  def validate_nonzero(number2:, **)
    return Success(:ok) if number2.nonzero?

    Failure(:err, message: 'arg2 must not be zero')
  end

  def divide(**input)
    Success(:ok, number: input.values.reduce(:/), **input)
  end
end

Divide.new.call(10, 5)
#<Solid::Output::Success type=:division_completed value={:number=>2}>

As you can see, even with divide success exposing the division number with all the accumulated data (**input), the #and_expose could generate a new success with a new type and only with the desired keys.

Remove the #and_expose call to see the difference. This will be the outcome:

Divide.new.call(10, 5)
#<Solid::Output::Success type=:ok value={:number=>2, :number1=>10, :number2=>5}>

PS: The #and_expose produces a terminal success by default. This means the next step will not be executed even if you call #and_then after #and_expose. To change this behavior, you can pass terminal: false to #and_expose.

⬆️  back to top

Module example (Singleton Methods)

Solid::Output.mixin can also produce singleton methods. Below is an example using a module (but it could be a class, too).

module Divide
  extend self, Solid::Output.mixin

  def call(arg1, arg2)
    validate_numbers(arg1, arg2)
      .and_then(:validate_nonzero)
      .and_then(:divide)
      .and_expose(:division_completed, [:number])
  end

  private

  def validate_numbers(arg1, arg2)
    arg1.is_a?(::Numeric) or return Failure(:err, message: 'arg1 must be numeric')
    arg2.is_a?(::Numeric) or return Failure(:err, message: 'arg2 must be numeric')

    Success(:ok, number1: arg1, number2: arg2)
  end

  def validate_nonzero(number2:, **)
    return Success(:ok) if number2.nonzero?

    Failure(:err, message: 'arg2 must not be zero')
  end

  def divide(number1:, number2:)
    Success(:ok, number: number1 / number2)
  end
end

Divide.call(10, 5)
#<Solid::Output::Success type=:division_completed value={:number=>2}>

Divide.call('10', 5)
#<Solid::Output::Failure type=:err value={:message=>"arg1 must be numeric"}>

Divide.call(10, '5')
#<Solid::Output::Failure type=:err value={:message=>"arg2 must be numeric"}>

Divide.call(10, 0)
#<Solid::Output::Failure type=:err value={:message=>"arg2 must not be zero"}>

⬆️  back to top

Solid::Output::Expectations

The Solid::Output::Expectations is a Solid::Result::Expectations with the Solid::Output features.

This is an example using the mixin mode, but the standalone mode is also supported.

class Divide
  include Solid::Output::Expectations.mixin(
    config: {
      pattern_matching: { nil_as_valid_value_checking: true }
    },
    success: {
      division_completed: ->(value) { value => { number: Numeric } }
    },
    failure: {
      invalid_arg:      ->(value) { value => { message: String } },
      division_by_zero: ->(value) { value => { message: String } }
    }
  )

  def call(arg1, arg2)
    arg1.is_a?(Numeric) or return Failure(:invalid_arg, message: 'arg1 must be numeric')
    arg2.is_a?(Numeric) or return Failure(:invalid_arg, message: 'arg2 must be numeric')

    arg2.zero? and return Failure(:division_by_zero, message: 'arg2 must not be zero')

    Success(:division_completed, number: (arg1 / arg2))
  end
end

Divide.new.call(10, 5)
#<Solid::Output::Success type=:division_completed value={:number=>2}>

As in the Solid::Result::Expectations.mixin, the Solid::Output::Expectations.mixin will add a Result constant in the target class. It can generate success/failure results, which ensure the mixin expectations.

Let's see this using the previous example:

Divide::Result::Success(:division_completed, number: 2)
#<Solid::Output::Success type=:division_completed value={:number=>2}>

Divide::Result::Success(:division_completed, number: '2')
# value {:number=>"2"} is not allowed for :division_completed type ({:number=>"2"}: Numeric === "2" does not return true) (Solid::Result::Contract::Error::UnexpectedValue)

⬆️  back to top

Mixin add-ons

The Solid::Output.mixin and Solid::Output::Expectations.mixin also accepts the config: argument. And it works the same way as the Solid::Result mixins.

given

This addon is enabled by default. It will create the Given(*value) method. Use it to add a value to the result chain and invoke the next step (through and_then).

You can turn it off by passing given: false to the config: argument or using the Solid::Result.configuration.

The Given() addon for a Solid::Output can be called with one or more arguments. The arguments will be converted to a hash (to_h) and merged to define the first value of the result chain.

continue

The Solid::Output.mixin(config: { addon: { continue: true } }) or Solid::Output::Expectations.mixin(config: { addon: { continue: true } }) creates the Continue(value) method and change the Success() behavior to terminate the step chain.

So, if you want to advance to the next step, you must use Continue(**value) instead of Success(type, **value). Otherwise, the step chain will be terminated.

Let's use a mix of Solid::Output features to see in action with this add-on:

module Division
  require 'logger'

  extend self, Solid::Output::Expectations.mixin(
    config: {
      addon:            { continue: true },
      pattern_matching: { nil_as_valid_value_checking: true }
    },
    success: {
      division_completed: ->(value) { value => { number: Numeric } }
    },
    failure: {
      invalid_arg:      ->(value) { value => { message: String } },
      division_by_zero: ->(value) { value => { message: String } }
    }
  )

  def call(arg1, arg2, logger: ::Logger.new(STDOUT))
    Given(number1: arg1, number2: arg2)
      .and_then(:require_numbers)
      .and_then(:check_for_zeros)
      .and_then(:divide, logger: logger)
      .and_expose(:division_completed, [:number])
  end

  private

  def require_numbers(number1:, number2:)
    number1.is_a?(::Numeric) or return Failure(:invalid_arg, message: 'arg1 must be numeric')
    number2.is_a?(::Numeric) or return Failure(:invalid_arg, message: 'arg2 must be numeric')

    Continue()
  end

  def check_for_zeros(number1:, number2:)
    return Failure(:division_by_zero, message: 'arg2 must not be zero') if number2.zero?

    return Success(:division_completed, number: 0) if number1.zero?

    Continue()
  end

  def divide(number1:, number2:, logger:)
    result = number1 / number2

    logger.info("The division result is #{result}")

    Continue(number: result)
  end
end

Division.call(14, 2)
# I, [2023-10-27T02:01:05.812388 #77823]  INFO -- : The division result is 7
#<Solid::Output::Success type=:division_completed value={:number=>7}>

Division.call(0, 2)
##<Solid::Output::Success type=:division_completed value={:number=>0}>

Division.call('14', 2)
#<Solid::Output::Failure type=:invalid_arg value={:message=>"arg1 must be numeric"}>

Division.call(14, '2')
#<Solid::Output::Failure type=:invalid_arg value={:message=>"arg2 must be numeric"}>

Division.call(14, 0)
#<Solid::Output::Failure type=:division_by_zero value={:message=>"arg2 must not be zero"}>

⬆️  back to top

Pattern Matching

The Solid::Result and Solid::Output also provides support to pattern matching.

Solid::Result

In the further examples, I will use the Divide lambda to exemplify its usage.

Divide = lambda do |arg1, arg2|
  arg1.is_a?(::Numeric) or return Solid::Result::Failure(:invalid_arg, 'arg1 must be numeric')
  arg2.is_a?(::Numeric) or return Solid::Result::Failure(:invalid_arg, 'arg2 must be numeric')

  return Solid::Result::Failure(:division_by_zero, 'arg2 must not be zero') if arg2.zero?

  Solid::Result::Success(:division_completed, arg1 / arg2)
end

Array/Find patterns

case Divide.call(4, 2)
in Solid::Failure[:invalid_arg, msg] then puts msg
in Solid::Failure[:division_by_zero, msg] then puts msg
in Solid::Success[:division_completed, num] then puts num
end

# The code above will print: 2

case Divide.call(4, 0)
in Solid::Failure[:invalid_arg, msg] then puts msg
in Solid::Failure[:division_by_zero, msg] then puts msg
in Solid::Success[:division_completed, num] then puts num
end

# The code above will print: arg2 must not be zero

⬆️  back to top

Hash patterns

case Divide.call(10, 2)
in Solid::Failure(type: :invalid_arg, value: msg) then puts msg
in Solid::Failure(type: :division_by_zero, value: msg) then puts msg
in Solid::Success(type: :division_completed, value: num) then puts num
end

# The code above will print: 5

case Divide.call('10', 2)
in Solid::Failure(type: :invalid_arg, value: msg) then puts msg
in Solid::Failure(type: :division_by_zero, value: msg) then puts msg
in Solid::Success(type: :division_completed, value: num) then puts num
end

# The code above will print: arg1 must be numeric

You can also use Solid::Result::Success and Solid::Result::Failure as patterns.

⬆️  back to top

Solid::Output

In the further examples, I will use the Divide lambda to exemplify its usage.

Divide = lambda do |arg1, arg2|
  arg1.is_a?(::Numeric) or return Solid::Output::Failure(:invalid_arg, err: 'arg1 must be numeric')
  arg2.is_a?(::Numeric) or return Solid::Output::Failure(:invalid_arg, err: 'arg2 must be numeric')

  return Solid::Output::Failure(:division_by_zero, err: 'arg2 must not be zero') if arg2.zero?

  Solid::Output::Success(:division_completed, num: arg1 / arg2)
end

Array/Find patterns

case Divide.call(4, 2)
in Solid::Failure[:invalid_arg, {msg:}] then puts msg
in Solid::Failure[:division_by_zero, {msg:}] then puts msg
in Solid::Success[:division_completed, {num:}] then puts num
end

# The code above will print: 2

case Divide.call(4, 0)
in Solid::Failure[:invalid_arg, {msg:}] then puts msg
in Solid::Failure[:division_by_zero, {msg:}] then puts msg
in Solid::Success[:division_completed, {num:}] then puts num
end

# The code above will print: arg2 must not be zero

⬆️  back to top

Hash patterns

If you don't provide the keys :type and :value, the pattern will match the result value.

case Divide.call(10, 2)
in Solid::Failure({msg:}) then puts msg
in Solid::Success({num:}) then puts num
end

case Divide.call(10, 2)
in Solid::Failure(type: :invalid_arg, value: {msg:}) then puts msg
in Solid::Failure(type: :division_by_zero, value: {msg:}) then puts msg
in Solid::Success(type: :division_completed, value: {num:}) then puts num
end

# The code above will print: 5

case Divide.call('10', 2)
in Solid::Failure(type: :invalid_arg, value: {msg:}) then puts {msg:}
in Solid::Failure(type: :division_by_zero, value: {msg:}) then puts msg
in Solid::Success(type: :division_completed, value: {num:}) then puts num
end

# The code above will print: arg1 must be numeric

You can also use Solid::Output::Success and Solid::Output::Failure as patterns.

⬆️  back to top

How to pattern match without the concept of success and failure

You can use the classes Solid::Result and Solid::Output as patterns, and the pattern matching will work without the concept of success and failure.

case Divide.call(10, 2)
in Solid::Output(:invalid_arg, {msg:}) then puts msg
in Solid::Output(:division_by_zero, {msg:}) then puts msg
in Solid::Output(:division_completed, {num:}) then puts num
end

case Divide.call(10, 2)
in Solid::Result(:invalid_arg, msg) then puts msg
in Solid::Result(:division_by_zero, msg) then puts msg
in Solid::Result(:division_completed, num) then puts num
end

The Solid::Result will also work with the Solid::Output, but the opposite won't.

⬆️  back to top

Solid::Result.event_logs

Use Solid::Result.event_logs(&block) to track all the results produced in the same or between different operations (it works with Solid::Result and Solid::Output). When there is a nesting of event_logs blocks, this mechanism will be able to correlate parent and child blocks and present the duration of all operations in milliseconds.

When you wrap the creation of the result with Solid::Result.event_logs, the final one will expose all the event log records through the Solid::Result#event_logs method.

class Division
  include Solid::Result.mixin(config: { addon: { continue: true } })

  def call(arg1, arg2)
    Solid::Result.event_logs(name: 'Division', desc: 'divide two numbers') do
      Given([arg1, arg2])
        .and_then(:require_numbers)
        .and_then(:check_for_zeros)
        .and_then(:divide)
    end
  end

  private

  ValidNumber = ->(arg) { arg.is_a?(Numeric) && (!arg.respond_to?(:finite?) || arg.finite?) }

  def require_numbers((arg1, arg2))
    ValidNumber[arg1] or return Failure(:invalid_arg, 'arg1 must be a valid number')
    ValidNumber[arg2] or return Failure(:invalid_arg, 'arg2 must be a valid number')

    Continue([arg1, arg2])
  end

  def check_for_zeros(numbers)
    num1, num2 = numbers

    return Failure(:division_by_zero, 'num2 cannot be zero') if num2.zero?

    num1.zero? ? Success(:division_completed, 0) : Continue(numbers)
  end

  def divide((num1, num2))
    Success(:division_completed, num1 / num2)
  end
end

module SumDivisionsByTwo
  extend self, Solid::Result.mixin

  def call(*numbers)
    Solid::Result.event_logs(name: 'SumDivisionsByTwo') do
      divisions = numbers.map { |number| Division.new.call(number, 2) }

      if divisions.any?(&:failure?)
        Failure(:errors, divisions.select(&:failure?).map(&:value))
      else
        Success(:sum, divisions.sum(&:value))
      end
    end
  end
end

Let's see the result of the SumDivisionsByTwo call:

result = SumDivisionsByTwo.call(20, 10)
# => #<Solid::Result::Success type=:sum value=15>

result.event_logs
{
  :version => 1,
  :metadata => {
    :duration => 0,   # milliseconds
    :trace_id => nil, # can be set through configuration
    :ids => {
      :tree => [0, [[1, []], [2, []]]],
      :matrix => { 0 => [0, 0], 1 => [1, 1], 2 => [2, 1]},
      :level_parent => { 0 => [0, 0], 1 => [1, 0], 2 => [1, 0]}
    }
  },
  :records=> [
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>1, :name=>"Division", :desc=>"divide two numbers"},
      :result => {:kind=>:success, :type=>:_given_, :value=>[20, 2], :source=><Division:0x0000000102fd7ed0>},
      :and_then => {},
      :time => 2024-01-26 02:53:11.310346 UTC
    },
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>1, :name=>"Division", :desc=>"divide two numbers"},
      :result => {:kind=>:success, :type=>:_continue_, :value=>[20, 2], :source=><Division:0x0000000102fd7ed0>},
      :and_then => {:type=>:method, :arg=>nil, :method_name=>:require_numbers},
      :time => 2024-01-26 02:53:11.310392 UTC
    },
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>1, :name=>"Division", :desc=>"divide two numbers"},
      :result => {:kind=>:success, :type=>:_continue_, :value=>[20, 2], :source=><Division:0x0000000102fd7ed0>},
      :and_then => {:type=>:method, :arg=>nil, :method_name=>:check_for_zeros},
      :time=>2024-01-26 02:53:11.310403 UTC
    },
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>1, :name=>"Division", :desc=>"divide two numbers"},
      :result => {:kind=>:success, :type=>:division_completed, :value=>10, :source=><Division:0x0000000102fd7ed0>},
      :and_then => {:type=>:method, :arg=>nil, :method_name=>:divide},
      :time => 2024-01-26 02:53:11.310409 UTC
    },
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>2, :name=>"Division", :desc=>"divide two numbers"},
      :result => {:kind=>:success, :type=>:_given_, :value=>[10, 2], :source=><Division:0x0000000102fd6378>},
      :and_then => {},
      :time => 2024-01-26 02:53:11.310424 UTC
    },
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>2, :name=>"Division", :desc=>"divide two numbers"},
      :result => {:kind=>:success, :type=>:_continue_, :value=>[10, 2], :source=><Division:0x0000000102fd6378>},
      :and_then => {:type=>:method, :arg=>nil, :method_name=>:require_numbers},
      :time => 2024-01-26 02:53:11.310428 UTC
    },
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>2, :name=>"Division", :desc=>"divide two numbers"},
      :result => {:kind=>:success, :type=>:_continue_, :value=>[10, 2], :source=><Division:0x0000000102fd6378>},
      :and_then => {:type=>:method, :arg=>nil, :method_name=>:check_for_zeros},
      :time => 2024-01-26 02:53:11.310431 UTC
    },
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>2, :name=>"Division", :desc=>"divide two numbers"},
      :result => {:kind=>:success, :type=>:division_completed, :value=>5, :source=><Division:0x0000000102fd6378>},
      :and_then => {:type=>:method, :arg=>nil, :method_name=>:divide},
      :time => 2024-01-26 02:53:11.310434 UTC
    },
    {
      :root => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :parent => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :current => {:id=>0, :name=>"SumDivisionsByTwo", :desc=>nil},
      :result => {:kind=>:success, :type=>:sum, :value=>15, :source=>SumDivisionsByTwo},
      :and_then => {},
      :time => 2024-01-26 02:53:11.310444 UTC
    }
  ]
}

⬆️  back to top

metadata: {ids:}

The :ids metadata property is a hash with three properties:

• :tree, a graph/tree representation of the id of each event_logs block.
• :level_parent, a hash with the level (depth) of each block and its parent id.
• :matrix, a matrix representation of the event logs ids. It is a simplification of the :tree property.

Use these data structures to build your own visualization.

Check out Event Logs Listener example to see how a listener can be used to build a STDOUT visualization, using these properties.

# tree:
# A graph representation (array of arrays) of the each event logs block id.
#
0                  # [0, [
|- 1               #   [1, [[2, []]]],
|  |- 2            #   [3, []],
|- 3               #   [4, [
|- 4               #     [5, []],
|  |- 5            #     [6, [[7, []]]]
|  |- 6            #   ]],
|     |- 7         #   [8, []]
|- 8               # ]]

# level_parent:
# The event logs ids are the keys, and the level (depth) and parent id the values.
                   # {
0                  #   0 => [0, 0],
|- 1               #   1 => [1, 0],
|  |- 2            #   2 => [2, 1],
|- 3               #   3 => [1, 0],
|- 4               #   4 => [1, 0],
|  |- 5            #   5 => [2, 4],
|  |- 6            #   6 => [2, 4],
|     |- 7         #   7 => [3, 6],
|- 8               #   8 => [1, 0]
                   # }

# matrix:
# The rows are the direct blocks from the root block,
# and the columns are the nested blocks from the direct ones.
                   # {
0 | 1 | 2 | 3 | 4  #   0 => [0, 0],
- | - | - | - | -  #   1 => [1, 1],
0 |   |   |   |    #   2 => [1, 2],
1 | 1 | 2 |   |    #   3 => [2, 1],
2 | 3 |   |   |    #   4 => [3, 1],
3 | 4 | 5 | 6 | 7  #   5 => [3, 2],
4 | 8 |   |   |    #   6 => [3, 3],
                   #   7 => [3, 4],
                   #   8 => [4, 1]
                   # }

⬆️  back to top

Configuration

Turning on/off

You can use Solid::Result.config.feature.disable!(event_logs) and Solid::Result.config.feature.enable!(event_logs) to turn on/off the Solid::Result.event_logs feature.

Solid::Result.configuration do |config|
  config.feature.disable!(event_logs)
end

result = SumDivisionsByTwo.call(20, 10)
# => #<Solid::Result::Success type=:sum value=15>

result.event_logs
{
  :version=>1,
  :records=>[],
  :metadata=>{
    :duration=>0,
    :ids=>{:tree=>[], :matrix=>{}, :level_parent=>{}}, :trace_id=>nil
  }
}

⬆️  back to top

Setting a trace_id fetcher

You can define a lambda (arity 0) to fetch the trace_id. This lambda will be called before the first event logs block and will be used to set the :trace_id in the :metadata property.

Use to correlate different or the same operation (executed multiple times).

Solid::Result.config.event_logs.trace_id = -> { Thread.current[:solid_result_event_logs_trace_id] }

⬆️  back to top

Setting a listener

You can define a listener to be called during the event logs tracking (check out this example). It must be a class that includes Solid::Result::EventLogs::Listener.

Use it to build your additional logic on top of the tracking. Examples:

• Log the event logs.
• Perform the tracing.
• Instrument the event logs (measure/report).
• Build a visualization (Diagrams, using the records: + metadata: {ids:} properties).

After implementing your listener, you can set it to the Solid::Result.config.event_logs.listener=:

Solid::Result.config.event_logs.listener = MyEventLogsListener

See the example below to understand how to implement one:

class MyEventLogsListener
  include Solid::Result::EventLogs::Listener

  # A listener will be initialized before the first event logs block, and it is discarded after the last one.
  def initialize
  end

  # This method will be called before each event logs block.
  # The parent block will be called first in the case of nested ones.
  #
  # @param scope: {:id=>1, :name=>"SomeOperation", :desc=>"Optional description"}
  def on_start(scope:)
  end

  # This method will wrap all the event logs in the same block.
  # It can be used to perform an instrumentation (measure/report).
  #
  # @param scope: {:id=>1, :name=>"SomeOperation", :desc=>"Optional description"}
  def around_event_logs(scope:)
    yield
  end

  # This method will wrap each and_then call.
  # It can be used to perform an instrumentation of the and_then calls.
  #
  # @param scope: {:id=>1, :name=>"SomeOperation", :desc=>"Optional description"}
  # @param and_then:
  #  {:type=>:block, :arg=>:some_injected_value}
  #  {:type=>:method, :arg=>:some_injected_value, :method_name=>:some_method_name}
  def around_and_then(scope:, and_then:)
    yield
  end

  # This method will be called after each result recording/tracking.
  #
  # @param record:
  # {
  #   :root => {:id=>0, :name=>"RootOperation", :desc=>nil},
  #   :parent => {:id=>0, :name=>"RootOperation", :desc=>nil},
  #   :current => {:id=>1, :name=>"SomeOperation", :desc=>nil},
  #   :result => {:kind=>:success, :type=>:_continue_, :value=>{some: :thing}, :source=><MyProcess:0x0000000102fd6378>},
  #   :and_then => {:type=>:method, :arg=>nil, :method_name=>:some_method},
  #   :time => 2024-01-26 02:53:11.310431 UTC
  # }
  def on_record(record:)
  end

  # This method will be called at the end of the event logs tracking.
  #
  # @param event_logs:
  # {
  #   :version => 1,
  #   :metadata => {
  #     :duration => 0,
  #     :trace_id => nil,
  #     :ids => {
  #       :tree => [0, [[1, []], [2, []]]],
  #       :matrix => { 0 => [0, 0], 1 => [1, 1], 2 => [2, 1]},
  #       :level_parent => { 0 => [0, 0], 1 => [1, 0], 2 => [1, 0]}
  #     }
  #   },
  #   :records => [
  #     # ...
  #   ]
  # }
  def on_finish(event_logs:)
  end

  # This method will be called when an exception is raised during the event logs tracking.
  #
  # @param exception: Exception
  # @param event_logs: Hash
  def before_interruption(exception:, event_logs:)
  end
end

⬆️  back to top

Setting multiple listeners

You can use Solid::Result::EventLogs::Listeners[] to creates a listener of listeners (check out this example), which will be called in the order they were added.

Attention: It only allows one listener to handle around_and_then and another around_event_logs records.

The example below defines different listeners to handle around_and_then and around_event_logs, but it is also possible to define a listener to handle both.

class AroundAndThenListener
  include Solid::Result::EventLogs::Listener

  # It must be a static/singleton method.
  def self.around_and_then?
    true
  end

  def around_and_then(scope:, and_then:)
    #...
  end
end

class AroundEventLogsListener
  include Solid::Result::EventLogs::Listener

  # It must be a static/singleton method.
  def self.around_event_logs?
    true
  end

  def around_event_logs(scope:)
    #...
  end
end

class MyEventLogsListener
  include Solid::Result::EventLogs::Listener
end

How to use it:

# The listeners will be called in the order they were added.
Solid::Result.config.event_logs.listener = Solid::Result::EventLogs::Listeners[
  MyEventLogsListener,
  AroundAndThenListener,
  AroundEventLogsListener
]

Check out this example to see a listener to print the event logs and another to store them in the database.

⬆️  back to top

Solid::Result.configuration

The Solid::Result.configuration allows you to configure default behaviors for Solid::Result and Solid::Output through a configuration block. After using it, the configuration is frozen, ensuring the expected behaviors for your application.

Note: You can use Solid::Result.configuration(freeze: false) {} to avoid the freezing. This can be useful in tests. Please be sure to use it with caution.

Solid::Result.configuration do |config|
  config.addon.enable!(:given, :continue)

  config.constant_alias.enable!('Result', 'Solid::Output')

  config.pattern_matching.disable!(:nil_as_valid_value_checking)

  # config.feature.disable!(:expectations) if ::Rails.env.production?
end

Use disable! to disable a feature and enable! to enable it.

Let's see what each configuration in the example above does:

config.addon.enable!(:given, :continue) <!-- omit in toc -->

This configuration enables the Continue() method for Solid::Result.mixin, Solid::Output.mixin, Solid::Result::Expectation.mixin, and Solid::Output::Expectation.mixin. Link to documentations: (1) (2).

It is also enabling the Given() which is already enabled by default. Link to documentation: (1) (2).

config.constant_alias.enable!('Result', 'Solid::Output') <!-- omit in toc -->

This configuration make Result a constant alias for Solid::Result, and Solid::Output a constant alias for Solid::Output.

Link to documentations:

• Result alias

config.pattern_matching.disable!(:nil_as_valid_value_checking) <!-- omit in toc -->

This configuration disables the nil_as_valid_value_checking for Solid::Result and Solid::Output. Link to documentation.

config.feature.disable!(:expectations) <!-- omit in toc -->

This configuration turns off the expectations for Solid::Result and Solid::Output. The expectations are helpful in development and test environments, but they can be disabled in production environments for performance gain.

PS: I'm using ::Rails.env.production? to check the environment, but you can use any logic you want.

⬆️  back to top

Solid::Result.config

The Solid::Result.config allows you to access the current configuration.

Solid::Result.config.addon <!-- omit in toc -->

Solid::Result.config.addon.enabled?(:continue)
Solid::Result.config.addon.enabled?(:given)

Solid::Result.config.addon.options
# {
#   :continue=>{
#     :enabled=>false,
#     :affects=>[
#       "Solid::Result.mixin",
#       "Solid::Output.mixin",
#       "Solid::Result::Expectations.mixin",
#       "Solid::Output::Expectations.mixin"
#     ]
#   },
#   :given=>{
#     :enabled=>true,
#     :affects=>[
#       "Solid::Result.mixin",
#       "Solid::Output.mixin",
#       "Solid::Result::Expectations.mixin",
#       "Solid::Output::Expectations.mixin"
#     ]
#   }
# }

Solid::Result.config.constant_alias <!-- omit in toc -->

Solid::Result.config.constant_alias.enabled?('Result')

Solid::Result.config.constant_alias.options
# {
#   "Result"=>{:enabled=>false, :affects=>["Object"]}
# }

Solid::Result.config.pattern_matching <!-- omit in toc -->

Solid::Result.config.pattern_matching.enabled?(:nil_as_valid_value_checking)

Solid::Result.config.pattern_matching.options
# {
#   :nil_as_valid_value_checking=>{
#     :enabled=>false,
#     :affects=>[
#       "Solid::Result::Expectations,
#       "Solid::Output::Expectations"
#     ]
#   }
# }

Solid::Result.config.feature <!-- omit in toc -->

Solid::Result.config.feature.enabled?(:expectations)

Solid::Result.config.feature.options
# {
#   :expectations=>{
#     :enabled=>true,
#     :affects=>[
#       "Solid::Result::Expectations,
#       "Solid::Output::Expectations"
#     ]
#   },
#   event_logs=>{
#     :enabled=>true,
#     :affects=>[
#       "Solid::Result",
#       "Solid::Output"
#     ]
#   },
#   :and_then!=>{
#     :enabled=>false,
#     :affects=>[
#       "Solid::Result",
#       "Solid::Output"
#     ]
#   },
# }

⬆️  back to top

Solid::Result#and_then!

In the Ruby ecosystem, several gems facilitate operation composition using classes and modules. Two notable examples are the interactor gem and the u-case gem.

interactor gem example

class PlaceOrder
  include Interactor::Organizer

  organize CreateOrder,
           PayOrder,
           SendOrderConfirmation,
           NotifyAdmins
end

u-case gem example

class PlaceOrder < Micro::Case
  flow CreateOrder, PayOrder, SendOrderConfirmation, NotifyAdmins
end

# Alternative approach
class PlaceOrder < Micro::Case
  def call!
    call(CreateOrder)
      .then(PayOrder)
      .then(SendOrderConfirmation)
      .then(NotifyAdmins)
  end
end

To facilitate migration for users accustomed to the above approaches, solid-result includes the Solid::Result#and_then!/Solid::Output#and_then! methods, which will invoke the method call of the given operation and expect it to return a Solid::Result/Solid::Output object.

Solid::Result.configure do |config|
  config.feature.enable!(:and_then!)
end

class PlaceOrder
  include Solid::Output.mixin

  def call(**input)
    Given(input)
      .and_then!(CreateOrder.new)
      .and_then!(PayOrder.new)
      .and_then!(SendOrderConfirmation.new)
      .and_then!(NotifyAdmins.new)
  end
end

⬆️  back to top

Dependency Injection

Like #and_then, #and_then! also supports an additional argument for dependency injection.

In Solid::Result

class PlaceOrder
  include Solid::Result.mixin

  def call(input, logger:)
    Given(input)
      .and_then!(CreateOrder.new, logger)
      # Further method chaining...
  end
end

In Solid::Output

class PlaceOrder
  include Solid::Output.mixin

  def call(logger:, **input)
    Given(input)
      .and_then!(CreateOrder.new, logger:)
      # Further method chaining...
  end
end

⬆️  back to top

Configuration

Solid::Result.configure do |config|
  config.feature.enable!(:and_then!)

  config.and_then!.default_method_name_to_call = :perform
end

Explanation:

• enable!(:and_then!): Activates the and_then! feature.

• default_method_name_to_call: Sets a default method other than :call for and_then!.

⬆️  back to top

Analysis: Why is and_then! an Anti-pattern?

The and_then! approach, despite its brevity, introduces several issues:

• Lack of Clarity: The input/output relationship between the steps is not apparent.

• Steps Coupling: Each operation becomes interdependent (high coupling), complicating implementation and compromising the reusability of these operations.

We recommend cautious use of #and_then!. Due to these issues, it is turned off by default and considered an antipattern.

It should be a temporary solution, primarily for assisting in migration from another to gem to solid-result.

⬆️  back to top

#and_then versus #and_then!

The main difference between the #and_then and #and_then! is that the latter does not check the result source. However, as a drawback, the result source will change.

Attention: to ensure the correct behavior, do not mix #and_then and #and_then! in the same result chain.

⬆️  back to top

Analysis: Why is #and_then the antidote/standard?

The Solid::Result#and_then/Solid::Output#and_then methods diverge from the above approach by requiring explicit invocation and mapping of the outcomes at each process step. This approach has the following advantages:

• Clarity: The input/output relationship between the steps is apparent and highly understandable.

• Steps uncoupling: Each operation becomes independent (low coupling). You can even map a failure result to a success (and vice versa).

See this example to understand what your code should look like:

class PlaceOrder
  include Solid::Output.mixin(config: { addon: { continue: true } })

  def call(**input)
    Given(input)
      .and_then(:create_order)
      .and_then(:pay_order)
      .and_then(:send_order_confirmation)
      .and_then(:notify_admins)
      .and_expose(:order_placed, %i[order])
  end

  private

  def create_order(customer:, products:)
    CreateOrder.new.call(customer:, products:).handle do |on|
      on.success { |output| Continue(order: output[:order]) }
      on.failure { |error| Failure(:order_creation_failed, error:) }
    end
  end

  def pay_order(customer:, order:, payment_method:, **)
    PayOrder.new.call(customer:, payment_method:, order:).handle do |on|
      on.success { |output| Continue(payment: output[:payment]) }
      on.failure { |error| Failure(:order_payment_failed, error:) }
    end
  end

  def send_order_confirmation(customer:, order:, payment:, **)
    SendOrderConfirmation.new.call(customer:, order:, payment:).handle do |on|
      on.success { Continue() }
      on.failure { |error| Failure(:order_confirmation_failed, error:) }
    end
  end

  def notify_admins(customer:, order:, payment:, **)
    NotifyAdmins.new.call(customer:, order:, payment:)

    Continue()
  end
end

⬆️  back to top

About

Rodrigo Serradura created this project. He is the Solid Process creator and has already made similar gems like the u-case and kind. This gem can be used independently, but it also contains essential features that facilitate the adoption of Solid Process (the method) in code.

⬆️  back to top

Development

After checking out the repo, run bin/setup to install dependencies. Then, run bundle exec rake to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and the created tag, and push the .gem file to rubygems.org.

⬆️  back to top

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/solid-process/solid-result. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the code of conduct.

⬆️  back to top

License

The gem is available as open source under the terms of the MIT License.

⬆️  back to top

Code of Conduct

Everyone interacting in the Solid::Result project's codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.