readme.md

Exercise 07 - Defining a second service

In this exercise you'll enhance the service definition by introducing a second service that sits on top of the same underlying data model. You'll see that you can define as many services on the same data model as you need, either in the same service definition file, or in separate files. In this exercise you'll define the second service in the same file.

The scenario is that a service is required to provide an analysis frontend with basic statistics on book orders, and nothing else - so there's no need (or desire) to expose the rest of the data model to this frontend.

Steps

At the end of these steps, you'll have two OData services both exposing different views on the same underlying data model.

1. Add a new service definition

Service definitions can live alongside each other in the same CDS file.

👉 In srv/service.cds, add a second service definition thus:

service Stats {
    @readonly entity OrderInfo as projection on my.Orders excluding {
        createdAt,
        createdBy,
        modifiedAt,
        modifiedBy,
        book,
        country
    }
}

Here the Stats service exposes the Orders entity in a read-only fashion as in the CatalogService, but uses the excluding clause to omit specific properties. These properties are not of interest to the analysis UI so are explicitly left out. Note that it also exposes the information as an entity called OrderInfo.

👉 Effect a deployment to the persistence layer so that the relevant artefact will be created:

user@host:~/bookshop
=> cds deploy

👉 Start the service up again via cds watch and check the root document at http://localhost:4004/. You should see something like this:

2. Create multiple orders

Now let's create a number of orders, and see what the OrderInfo entityset shows us. We can do this quickly using another Postman collection, and using Postman's "Collection Runner" feature.

Note: If you want to do this using curl, jump to the using curl on the command line section.

Using Postman

👉 Import another collection into Postman from the URL to this resource: postman-07.json.

This screenshot shows what the collection looks like (it contains multiple POST requests to create orders for various books) and also shows the extra options which allows all the requests in the collection to be executed in one go:

👉 After importing this collection, click the arrow to the right of the collection name to expand the options as shown, and select the small blue "Run" button which will open up the "Collection Runner" window:

👉 Use the large blue "Run ..." button to execute all the requests - a results window should appear.

Using curl on the command line

Order 2 copies of Wuthering Heights:

curl \
  -d '{"book_ID":201,"quantity":2}' \
  -H 'Content-Type: application/json' \
  http://localhost:4004/catalog/Orders

For Windows users:

curl ^
  -d "{\"book_ID\": 201, \"quantity\": 2}" ^
  -H "Content-Type: application/json" ^
  http://localhost:4004/catalog/Orders

Order 7 copies of Eleonora:

curl \
  -d '{"book_ID":252,"quantity":7}' \
  -H 'Content-Type: application/json' \
  http://localhost:4004/catalog/Orders

For Windows users:

curl ^
  -d "{\"book_ID\": 252, \"quantity\": 7}" ^
  -H "Content-Type: application/json" ^
  http://localhost:4004/catalog/Orders

Order 42 copies of The Hitch Hiker's Guide To The Galaxy (obviously!):

curl \
  -d '{"book_ID":421,"quantity":42}' \
  -H 'Content-Type: application/json' \
  http://localhost:4004/catalog/Orders

For Windows users:

curl ^
  -d "{\"book_ID\": 421, \"quantity\": 42}" ^
  -H "Content-Type: application/json" ^
  http://localhost:4004/catalog/Orders

Now it's time to take a look at what the service will show us for these orders. We know we can't look at the Orders entityset as it has a @insertonly annotation shortcut based restriction, so we turn to our new service Stats.

👉 Look at the http://localhost:4004/stats/OrderInfo entityset in the Stats service. You should see something like this:

It shows us only the quantity statistics for the orders, just what we want for our simple analysis frontend.

Summary

It's easy to explore building different views on the same underlying data model, views that are focused and appropriate for different consumers, whether they are user interfaces or API clients.

Questions

1. Why might it be better to specify properties to exclude, rather than properties to include?

2. What did the order creation HTTP requests look like - which service was used, and why?

3. What was the artefact created in the persistence layer in Step 1?