e798119841
Flame needed pemanent adjustments to mathe the changes in the python-openstackclients. We now use openstacksdk or shade which will handle themselves the compatibility. We also made flame modular so that any-one can add features by implementing there own flame managers and adding their modules to the `openstack_flame` entry point. This new flame version is also fully compatible with python 3. Change-Id: I586a165b5022031963f504874bd50e1b11fe0d27
1.5 KiB
When developping on flameclient, do not forget to check code quality with the ./checkcode command, and then check you did not brake anything by running python -m unittest discover -v.
To create flame modules you need to create a module with a class which inheritates from flameclient.resources.ResourceManager
You need to implement the api_resources property and the get_hot_resources() method (read their docstring for more information). If you want you can also implement the add_arguments(parser=None) method to add your own module's command line arguments. You can also implement the post_process(), post_process_hot_resources(resources), post_process_heat_template(template) and/or post_process_adoption_data(adoption_data) methods to perform post processing after the generator's extract_data method was called (read their docstring for more information). This allows you to modify results before rendering the template. These post processing methods are not threaded and are executed in order of the managers' post_priority attribute (defaults to 100).
Then, you need to add in your package's setup.py or setup.cfg an 'openstack_flame' entry point pointing to the module file where your subclass of flameclient.resources.ResourceManager is defined.
Once your package installed, flame will automatically discover all 'openstack_flame' entry points to load the corresponding modules, and all loaded modules having a flameclient.resources.ResourceManager subclass will have this subclass detected and added tho the list of managers.