Resiliency

[This is preliminary documentation and subject to change.]

Another benefit of installer component management capabilities is the added resiliency available to applications.

Because applications have become more complex, they can be prone to difficulties. For example, components can be deleted by a user or even another application, files can become corrupt, components can be replaced with incompatible versions. The installer provides applications the capability to recover gracefully from such situations without presenting the user with error messages. If the installer is unsuccessful at providing a component, the application can respond in the following ways:

• Attempt to repair the file if it has become corrupted.
• Reinstall the feature according to one of a number of possible criteria. For example, you can choose to reinstall a feature only if the current version is older than the version available on the source or only if the current file is corrupt in some manner. For information on installing, see Installing an Application. For information on reinstalling, see Reinstalling a Feature or Application.
• Reinstall aspects of an application, such as shortcuts and registry entries.

Another way that applications can fail is that the resources the applications require are unavailable. Applications that rely on network resources for just-in-time installation are especially susceptible to such failures. However, the installer has also implemented a resiliency feature known as the source list. The source list contains the locations that the installer searches for installation packages. The entries in this list can be network locations, Uniform Resource Locators (URLs), or compact discs. If one of these sources fails, the installer can quickly and seamlessly try the next.

If an application utilizes these resiliency methods, a great number of recovery methods must fail before the user is presented with an error message.