(#459) prevent colon added in realm.properties when auth_users hash is empty

[overag3]

Pull Request (PR) description

This changeset fix issue when @auth_config['file']['auth_users'] hash is empty, a colon added in realm.properties file when it shouldn't.
Fix also wrong indentation.

This Pull Request (PR) fixes the following issues

Fixes #459

[kenyon]

Thanks for the improvement. It would be even better if this had tests, to prevent such breakage from recurring.

[overag3]

Test already exists in auth_spec.rb file but i found the real problem...

In params manifest(L62, L168), the type of auth_users variable is hash but in spec tests and in real.properties.erb template file, the type of auth_users needs to be a array not hash.

Besides, is line 62 necessary ?

[kenyon]

Good finds. Yeah, looks like auth_users in params.pp line 62 is unused. The template also treats it as a hash, in addition to an array:

puppet-rundeck/templates/realm.properties.erb

Line 35 in 8b257e4

<%= @auth_config['file']['auth_users']['username'] -%>:<%= @auth_config['file']['auth_users']['password'] -%>

This module needs better data typing and better testing. 😒

[overag3]

Am I modifying the params.pp file in this PR ?

I have never written spec tests before. 😟

[overag3]

Good finds. Yeah, looks like auth_users in params.pp line 62 is unused. The template also treats it as a hash, in addition to an array:

puppet-rundeck/templates/realm.properties.erb

Line 35 in 8b257e4

<%= @auth_config['file']['auth_users']['username'] -%>:<%= @auth_config['file']['auth_users']['password'] -%>

This module needs better data typing and better testing. unamused

Yes, auth_users is a array of hashes. Example is available in :

puppet-rundeck/spec/classes/config/global/auth_spec.rb

Lines 56 to 67 in d50ff91

	'auth_users' => [
	{
	'username' => 'testuser',
	'password' => 'password',
	'roles' => %w[user deploy]
	},
	{
	'username' => 'anotheruser',
	'password' => 'anotherpassword',
	'roles' => ['user']
	}
	]

Tell me what I can do to solve this issue.

Regards

[kenyon]

I think this PR is fine as is since it fixes the problem. We can make other PRs for overhauling the data types and testing. I'm just waiting for anyone else to provide feedback here, otherwise I think we can merge this soon.