Create an application that will list all available articles for the user to select and view in detail. The front end application will need to talk to a proxy API that caches data from Zendesk's FAQs API for an hour.
Ensure the backend is running before the frontend.
The application was developed on Ubuntu 18.04
Modules and Libraries used:
First clone repository.
git clone https://github.com/devadeka/zd_tickets_be.git
Install dependencies.
bundle install
Though the API no longer uses the database (see Design Decisions), the project started with a database so it needs to be initialized.
rails db:create
rails db:migrate
After cloning and installing dependencies. Ensure internet connection is available as the application makes calls to external APIs.
bundle exec rspec
After cloning and installing dependencies. Ensure internet connection is available as the application makes calls to external APIs.
rails s
The main design decision came in the form of the caching technique. Originally the API was built as a standard CRUD app with a PostgresDB- only allowing the controller action of index. The caching mechanism would have been to run a background cron-job to periodically request the data and use the Rails-Models to add/update in Postgres (this can be seen in the cron_caching branch of the repository). This method was abandoned after looking at the number of DB calls the background job would have to make to maintain the near 2000 entries.
The cron-caching method was abandoned in favour of redis-caching. Rails5 has support for Redis caching, so the set up was simple. Where the cron-caching method intended to keep 'models' of Articles/FAQs, the redis-caching method stored the JSON response of the page request (which contained the Articles and the metrics - page and page_count). If the cache expired, the controller will use an API wrapper to request the data.
A minor design decision was to filter out unnecessary fields in the JSON response. Because the purpose of this API was to cache and forward information to the frontend application, knowing exactly what information the frontend utilises allows this API to only return the needed fields. The advantage of this would be to reduce the payload for the frontend.
Reducing the payload was also a factor in deciding to hardcode (rather than allow the user to select) the per_page value when querying Zendesk's FAQs API. The per_page value can be set up to 100, but a value of 10 was chosen so that a response will be processed quicker - the user of the frontend application can make multiple small requests for page information.
An advantage of the cron-caching method was it could be tested with dummy data without relying on the external data source. A better testing strategy is needed for the API to not rely on having internet connection.
Since the database is no longer being used, all references to it needs to be removed.