Integrating by using a command-line interface

To set up this kind of integration, you do the following:

1. Install the Build Forge console on a host
2. Install the Build Forge agent on the application host (or a host that can access the application)
3. In the console, create a Server resource and Server authentication. Configure the Server resource to access the Build Forge agent you installed.
4. Configure the agent and host environments as necessary for commands to run on the application. This may be as simple as creating a user account for Build Forge to use, then assuring that the running agent has its PATH set up correctly. It may also require that you install and configure a client that is used to run commands in the application. For example, Rational ClearCase and ClearQuest require the use of client applications to run commands.

When the setup is complete, projects you create in Build Forge can contain steps that run application commands. You can control project execution based on pass or fail status of a step or set up log filters to scan the Build Forge log for patterns of output that the commands return. Applications may require additional setup in the project logic. For example, integrating with ClearCase would require that you include the creation, starting, and populating of views.

You can use condition steps and loop steps to control what happens in response to a particular commands' success or failure.

A typical use with source control applications is build avoidance. A command is run to query whether source code updates have been checked in since the last time a build was run:

• If there are no updates, no build is run.
• If there have been updates, a build is run

Module dependencies can be expressed in the execution logic of project steps. You can control whether a build of a particular module triggers a build of other modules or the entire software project.

You use notification templates to control how groups of project members are notified in response to a build success or failure. In a continuous integration environment it is important to notify on failure so that problem code can be fixed. Using command-line integration, it is possible only to notify an entire group of project members. The project members would need to inspect the step log for the project to determine what code caused the failure.