Starting with PostgreSQL 9.4 and MySQL 5.7, we can create JSON and array columns. The main benefit of storing these values as structured data (as opposed to “stringifying” them) is that we can query INTO the values via more sophisticated mechanisms than “does this string match”.

Create a new database migration to add a tags array column to the Product table. This should allow us to do some non-hierarchical categorization on the product list (i.e., “Sauces”, “Bakery”, “Beverages”).

Create a new database migration to add a preferences column for json values to the Customer table. Present the information on the customer’s page.

A publish-subscribe (pubsub) system is a software architecture pattern where publishers push messages into “channels”, and subscribers who have an interest in particular channels receive them. Publishers and subscribers have no direct knowledge of each other.

Whenever a new order is created, use the existing web socket mechanism with LISTEN and NOTIFY calls to trigger a page refresh (if users are viewing the dashboard).

When implementing a search-engine-like feature on a web application, typically the results must very closely match the search term. In the past, this limitation was countered by adding new system components like Apache Solr and [Lucene]( [Apache Lucene - Apache Lucene Core]https://lucene.apache.org/core/). Setting these up is a daunting task, to say the least, and is absolutely overkill for many use cases.

Thankfully, modern versions of PostgreSQL and MySQL feature simplified versions of this technology. We can perform a search against multiple fields, specifying how much “weight” should be given to each field.

There’s currently a “global search” feature on our workshop app, which uses an overly simplistic and narrow mechanism to find relevant results. Upgrade this feature using our database’s full text search feature set.