NAME

cvsintro —

introduction to Concurrent Versioning System

DESCRIPTION

Concurrent Versions System is a popular open source version control system. Although it is mostly used to track changes to software source code for development teams, there are very few limitations to the kind of data that can be managed by the system, and it can be used for document archival purposes as well.

This document provides an introduction to using OpenCVS to manage repositories. It will first cover some of the aspects of using and managing CVS and, where appropriate, explain the main differences between OpenCVS and other distributions.

USING CVS TO TRACK SOURCE CHANGES

One of the most common uses of cvs(1) is to track changes to a collection of source files (a module) contained within a certain, defined, location (a repository). This allows the user to keep a set of local sources in sync with a set of remote sources. It also allows the user to view information about changes to the repository (such as what the exact change was, who made it, and when), to view and compare information about different versions of a file, and possibly to make local changes to files.

As an example, we'll look at a user wishing to track source changes to the OpenBSD tree, maintaining a local (personal) copy of the source on their own machine. The user will have to decide:

The CVS server to use. There may be only one server or, for larger projects, a number of mirror servers.
Where the source module(s) will be kept on the local machine.

In our example, the OpenBSD project, there are a large number of servers to choose from. It's generally better to use a server that is geographically closer since this tends to be faster and less of a strain for the master server. Utilities such as ping(8) and traceroute(8) can be used to decide which host will be best to use. Secondly, the local repository has to be stored somewhere. OpenBSD uses /usr/src as the default location for its source tree (the “src” module), but this is largely arbitrary.

This example shows how a user initially checks out a copy of the source tree from local mirror anoncvs.nl.openbsd.org:

$ cd /usr
$ cvs -d anoncvs@anoncvs.nl.openbsd.org:/cvs checkout -P src

In this case it was first necessary to add ourselves to the “wsrc” group, since /usr/src is writable only by user “root” and/or group “wsrc”. The -d option was necessary to tell cvs(1) the location of the remote server. Note the “:/cvs” string appended to the server's address: the path to the repository must be specified. Finally the checkout command was used to obtain a copy of the module “src”.

Note that the above example simply checked out OpenBSD-current: often different software versions are available: use an identifier (tag) to specify which version to check out.

Thereafter the user is free to manipulate the source tree using the cvs(1) utility itself. For example, the following would update (sync) the local copy of “src” with the remote copy:

$ cd /usr/src
$ cvs -d anoncvs@anoncvs.nl.openbsd.org:/cvs update -Pd

General users may wish to use cvs(1) simply to keep a copy of sources up to date with a development tree; developers of software projects can also use cvs(1) to make their own changes to a set of remote source files, and to view changes made by other software developers.

See cvs(1) for more information on the different commands available. See the entry for $HOME/.cvsrc in cvs(5) for details on configuring cvs(1) itself.

USING CVS TO MANAGE A REPOSITORY

Software developers may wish to use cvs(1) to manage their own software projects. Here we look at an example usage: providing remote access to a small group of developers working on project “foo”, located in /cvs/projects. First of all a repository has to be created. A repository is just the location of the group of files to be managed. Within the repository, files may be organised into collections of files, called modules, which are just logical groupings of files.

In our example, module “foo” is located at /cvs/projects/foo on machine “cvs.example.org”. Therefore /cvs/projects is both the root directory ($CVSROOT) of our repository and the name of our repository.

If a set of source files already exist, possibly already under revision control, they can be added to a repository using the cvs(1) command import. This is a useful way of adding modules to a pre-existing repository. In our example we are starting from scratch, so the repository has to be initialised using the init command:

# cd /cvs/projects
# cvs -d /cvs/projects init

This will create a default administrative directory, $CVSROOT/CVSROOT, filled with files concerned with the management of the repository.

Thereafter, access will have to be arranged for developers participating in the project. This typically means providing SSH access via sshd(8) for remote access, unless the development network is local.

HISTORY

The OpenCVS project is a -licensedBSD rewrite of the original Concurrent Versioning System written by Jean-Francois Brousseau. The original CVS code was written in large parts by Dick Grune, Brian Berliner, and Jeff Polk.

AUTHORS

Jean-Francois Brousseau

CAVEATS

This CVS implementation does not fully conform to the GNU CVS version. In some cases, this was done explicitly because GNU CVS has inconsistencies or ambiguous behaviour. Some things have also been left out or modified to enhance the overall security of the system.

Among other things, support for the pserver connection mechanism has been dropped because of security issues with the authentication mechanism.