White Paper: Changes in RPM 3.0

Overview

The Red Hat Package Manager (RPM) is widely used throughout the Linux community. The ease with which software packages can be built, installed, and removed has convinced many developers and distributions to use RPM for their software. The design and implementation of RPM has proven to be a reliable method for software distribution.

RPM packaging can be viewed from several vantage points. At the heart of RPM is a well defined package format that permits some types of extensions while maintaining a degree of backward compatibility with older implementations. RPM 3.0 continues to use the same package format as previous versions and is backward compatible with packages produced by older versions of RPM. Another important component in RPM packaging is the installed package database. Each package has a header that describes the files within the package. Every installed package header is preserved in the database so that the files in a package may be verified at any time. The package header also includes information that describes what the package does and a URL where more information can be found. Finally, the package header contains prerequisite and dependency information that is used to identify components needed and/or supplied by each package.

From a system administrator's point of view there is little additional functionality in RPM 3.0 that is visible. The ability to install and/or upgrade, verify and erase packages is identical to previous versions. The command line syntax necessary to perform these functions is largely identical to previous versions.

RPM 3.0 does include internal changes of which installers and packagers will need to be aware. The changes and motivation for these changes is described in the following sections.