Make sure .valid? called only once

[mrwokkel]

I ran into an issue where the execute code changes the conditions for an interaction to be valid.

For instance uniqueness of something. When .valid? is called again, after execute, these conditions are not valid anymore.

[tfausak]

Good catch! If I understand you correctly, this example shows what you're talking about:

class ClearArray < ActiveInteraction::Base
  array :array

  validate { errors.add(:array, 'is empty') if array.empty? }

  def execute
    array.clear
  end
end

ClearArray.run!(array: [1, 2, 3])
# => ActiveInteraction::InvalidInteractionError: Array is empty

outcome = ClearArray.run(array: [1, 2, 3])
outcome.valid?
# => nil

interaction = ClearArray.new(array: [1, 2, 3])
interaction.valid?
# => true
interaction.execute
# => []
interaction.valid?
# => nil

[mrwokkel]

Yes. And your examples in the README should better use errors.empty? or something of the like.

[tfausak]

I think this is a bug. You should be calling outcome.valid? instead of outcome.errors.empty?.

[mrwokkel]

You are right. What I meant was a way of asking interaction_object.errors.empty?. Also when the outcome has an uniqueness validation it will fail after the run method.

The way I do it now is:

# Something model has an uniqueness validation
@create_something = CreateSomething.new(params[:create_something])
if @create_something.save # just calls private method .run internally like so: run.present?
  # outcome.valid? will return false if called after .run because something is not unique anymore
  redirect_to something_url(@create_something.outcome)
else
  render :new
end

I hope I make sense :).

[mrwokkel]

Ah language. I think you already caught my drift.

[tfausak]

Fixed in version 1.0.1. Thanks for reporting this!