Using the retry parameter results in method calls no longer raise errors

[Sokre95]

Example:

Without using retry config:

api_client.crm.contacts.basic_api.get_by_id(contact_id: 123)
  # => raises an `Hubspot::Crm::Contacts::ApiError` error with status code 404

Using retry config:

retry_config = {
    429 => { max_retries: 3, seconds_delay: 5 },
    500..530 => { max_retries: 2, seconds_delay: 10 }
  }

response = api_client.crm.contacts.basic_api.get_by_id(contact_id: 123, retry: retry_config)
  # => does not raise error but returns instance instead
response.is_a?(Hubspot::Crm::Contacts::ApiError)
=> true
response.code
=> 404

I don't think it is good that gem behaves like this. Just because you start using retries suddenly it no longer raises exceptions and that is not expected at all.

Example, method like this doesn't work anymore if you add retry block

  def get_contact_by_id(id:)
    api_client.crm.contacts.basic_api.get_by_id(contact_id: id)
  rescue Hubspot::Crm::Contacts::ApiError => e
    return nil if e.code == 404

    raise e
  end

I don't actually mind if gem returns error response or actually raises an error. I just expect that the interface will behave the same regardless of whether retry option is used or not.

Related to

hubspot-api-ruby/lib/hubspot/discovery/base_api_client.rb

Lines 101 to 125 in c1dbc58

	def call_api_with_retry(api_method, params_to_pass, retry_config, &block)
	call_api_with_rescue(api_method, params_to_pass) do |error|
	opts = retry_config.detect{ |k,v| k === error.code }&.last
	retries = opts&.dig(:max_retries) || 5
	block.call(error) unless block.nil?
	response = error
	while retries > 0 && opts
	sleep opts[:seconds_delay] if opts[:seconds_delay]
	response = call_api_with_rescue(api_method, params_to_pass) do |e|
	block.call(e) unless block.nil?
	e
	end
	return response unless response.respond_to?(:code)
	opts = retry_config.detect{ |k,v| k === response.code }&.last
	retries -= 1
	end
	response
	end
	end