12.3. Source Code Management Systems There are several source code management systems used in the Unix community:
SCCS The Source Code Control System. SCCS is the original Unix source code management system. It was developed in the late 1970s for the Programmer's Workbench (PWB) Unix systems within Bell Labs. It is still in use at a few large longtime Unix sites. However, for a long time it was not available as a standard part of most commercial or BSD Unix systems, and it did not achieve the widespread popularity of other, later systems. (It is still available with Solaris.) SCCS uses a file storage format that allows it to retrieve any version of a source file in constant time.
RCS The Revision Control System. RCS was developed in the early 1980s at Purdue University by Walter F. Tichy. It became popular in the Unix world when it was shipped with 4.2 BSD in 1983. At the time, Berkeley Unix was the most widely used Unix variant, even though to get it, a site had to have a Unix license from AT&T. RCS is easier to use than SCCS. Although it has a number of related commands, only three or four are needed for day-to-day use, and they are quickly mastered. A central repository is easy to use: you first create a directory for the sandbox. In the sandbox, you make a symbolic link to the repository named RCS, and then all the developers can share the repository. RCS uses a file format that is optimized for retrieving the most recent version of a file.
CVS The Concurrent Versions System. CVS was initially built as a series of shell scripts sitting atop RCS. Later it was rewritten in C for robustness, although still using RCS commands to manage the storage of files. However, for quite some time, CVS has had the RCS functionality built into it, and it no longer requires that RCS be available. The file format continues to be the same. CVS was the first distributed source code management system and is currently the standard one for Unix systemsin particular for collaborative, distributed, Free and open source development projects. The repository is named when you create a sandbox and is then stored in the files in the sandbox, so that it need not be provided every time you run a CVS command. Unlike SCCS and RCS, which provide multiple commands, CVS has one main command (named cvs), which you use for just about every operation.
Subversion With increasing use, it became clear that CVS lacked some fundamental capabilities. The Subversion project was started by several longtime CVS users and developers with the explicit goal to "build a better CVS," not necessarily to explore uncharted territory in source code management systems. Subversion is thus intentionally easy to learn for CVS users. Subversion uses its own format for data storage, based on the Berkeley DB in-process data library. Distributed use was designed in from day one, providing useful facilities that leverage the capabilities of the well-known Apache HTTP server.
RCS, CVS, and Subversion represent a progression, each one building on the features of its predecessors. For example, all three share a large subset of the same keyword substitutions, and command names are similar or identical in all three. They also demonstrate the progression from centralized, locking-based development to distributed, conflict-resolution-based development. |
