Spring is an open-source framework created to address the complexity of Java enterprise application development. One of Spring's goals is to help developers write simple, testable and loosely coupled systems while reducing the amount of scaffolding code required. In this respect, Groovy has a common goal. So, for simple systems, Groovy alone may be sufficient for your needs. However, as your system grows in size and complexity, and especially in hybrid Java/Groovy environments, you might find Spring's facilities provide great value to your Groovy system development.
Here we look at using Spring's Bean Factory mechanisms within Groovy. These facilities allow beans to be managed within a Spring container. The beans are normally Java objects, but since Groovy objects are Java objects, Spring can just as easily manage Groovy objects for you. In particular, in mixed Java/Groovy environments, you can leverage any existing domain objects or services already have defined in your Spring wiring from the Java part of your application. You can immediately start using those within your Groovy scripts and code, allowing a great mix of strongly-typed Java domain objects and dynamically typed Groovy code.
Let's start by exploring a simple calculator application.
Bare Bones Approach
Suppose we have the following implementation class:
We can make use of that class from a script as follows:
This script relies on no external wiring files. Everything is configured in the script itself. If we wish to alter our system at a later time, we simply alter the configuration inside the
Classical Spring Approach
Probably the most common way to use Spring is to use an XML wiring file. As systems grow larger, wiring files allow configuration to be centralised in easy to change 'groupings' of beans. Let's assume our calculator needs to eventually be expanded to have additional functionality. An approach to handling complexity as the system grows is to delegate functionality off to other components. Here is how we might code up an adder component:
Then, applying the delegate design pattern we mentioned earlier would result in the following refactored calculator:
To capture our software system configuration, we will use an XML wiring file (we have two beans):
Now, our script code looks like:
If we wish, we can remove the need for an XML file by using annotations. Note that we need to think carefully before using this approach extensively in a large system for two reasons. Firstly, we are more tightly coupling our system to Spring. Secondly, we are associating configuration and deployment information with our source code. Perhaps separating those concerns will have enormous benefits for large systems. The good news is that Spring lets you take on board just those annotations that you are happy to use.
Here is how we might code up our adder component:
Here is our modified calculator:
And here is our script code (note no XML file is required):