IR extension for hierarchical attributes and variables

[joaotgoncalves]

No description provided.

[Nfsaavedra]

I think you should add some test that considers the case being handled here. Maybe it is a good time to add the first unit test for the parsing portion. I am not sure if the integration tests cover this case for every technology. Also, please add a more descriptive title and if necessary a description to the PR. It can be valuable for future contributions. Thanks!

[jff]

I left some comments. See also @Nfsaavedra's comments.

[Nfsaavedra]

We are close :) I will open a new issue with the discussion about the handling of hashes in Puppet and Chef in order to let us proceed. The only thing I see missing now is this case.

$my_hash = {
	'key1' => {
    		'test1' => '1',
    		'test2' => '2',
    	},
	'key2' => 'value2',
	'key3' => 'value3',
}

$my_hash['key4'] = 'value4'

The last assignment should be parsed as the first one but with only key4 as another variable. I think this is the behavior for Chef, however in Puppet it is not working. Also I think Chef is merging the assignments in a single variable.

default[:zabbix][:database][:password] = nil
default[:zabbix][:test][:name] = "something"

This translates to the IR to only one default[:zabbix]. I am actually not sure how this works (and if it was you that programmed it), but it would be nice to have the same behavior in Puppet.

Let me know if anything does not make sense

[joaotgoncalves]

@Nfsaavedra , is the following example also possible? I would like to know your opinion about it.

$my_hash = {
	'key1' => {
    		'test1' => '1',
    		'test2' => '2',
    	},
	'key2' => 'value2',
	'key3' => 'value3',
}

$my_hash['key4']['key'5] = 'value4'

In this case, 'key4' has also an hash value right? So we should have '$my_hash' with keyvalues ['key1', 'key2', 'key3', 'key4'], 'key4' with keyvalues ['key5'] and 'key5' with value 'value4'?

[Nfsaavedra]

LGTM. I will merge

Comments were already addressed