Design Pattern: Chain of Command or Chain of Responsibility

When programming, certain sections of code can sometimes be viewed as a workflow or preset sequence of tasks or commands. This can considered to be the design pattern called Chain of Command or alternatively, Chain of Responsibility. What this implies is that there is a container, the Context, which contains the objects processed by the workflow, which can be added and removed, by each Command in the Chain.

There are various models to define and implement these workflows, the easiest being writing the code as is (which isn’t much of a model). This introduces a level of formality in the workflow, any changes require source code changes and recompilation.

One alternative is to use a Java dynamic class loader (which contains the workflow) after some decision making process. With some small amount of creativity, this technique allows for new workflows to be created without having to modify and recompile the original source code, only the workflows. This may introduce undesirable runtime errors.

Another viable alternative is to use the Apache Chain library. The performance is very reasonable, the implementation, and configuration through XML files make it a treat to use. For performance, an empty Command (configured through the XML configuration file) was called for each word encountered in a set of 750 files through the use of a tokenizer, resulting in only an approximate 500ms processing delay (not counting startup costs or memory footprints, further results may follow).

About these ads