3. Setting up CVS

First you need to install the CVS package. On Redhat Linux use:

cd /mnt/cdrom/Redhat/RPMS
rpm -i rcs*.rpm
rpm -i cvs*.rpm
To see the list of files installed do -
rpm -qpl cvs*.rpm | less

On other flavors of Unix, you may need to download the RCS and CVS tar balls and follow the README, INSTALL files to setup CVS. Visit http://www.cyclic.com and http://www.loria.fr/~molli/cvs-index.html

3.1 Environment variables

The following environment variables need to be setup in /etc/profile - default values required for all users. If not set in /etc/profile, then you should add these to your local profile file  /.bash_profile.

export EDITOR=/bin/vi
export CVSROOT=/home/cvsroot
export CVSREAD=yes

# File ~/.bash_profile
# Overriding env variables by resetting
export EDITOR=/usr/bin/emacs
export CVSROOT=/home/anotherdir/java/cvsroot

Create a directory to store the source code repository and give read, write access to Unix group/user. Also make sure that the directory name of CVSROOT does not contain any blank spaces. For example CVSROOT should not be like '/home/my rootcvs'.

bash$ su - root
bash# export CVSROOT=/home/cvsroot
bash# groupadd --help
bash# groupadd cvs
bash# useradd --help
bash# useradd -g cvs -d /home/cvsroot cvs

bash# ls -ld $CVSROOT   ... (you should see the listing)
bash# chmod o-rwx $CVSROOT
bash# chmod ug+rwx $CVSROOT

#To initialize the CVS repository and to put in source code files do:
bash# cvs init

# Add the unix users to the cvs group. Create supplementary groups for users.
# Note that you MUST not put any blank spaces after comma seperating the
# group names in -G option.
# In example below user tom belongs to groups cvs, users and staff and user
# johnson belongs to group cvs only.
bash# usermod --help
bash# usermod -G cvs some_unix_username
bash# usermod -G cvs,users,staff tom
bash# usermod -G cvs,users,staroffice billclinton
bash# usermod -G cvs johnson
bash# exit    .... (logout of root superuser mode)

# Login as a user and import files into cvs....
bash$ su - billclinton
bash$ export EDITOR=/bin/vi
bash$ export CVSROOT=/home/cvsroot
bash$ export CVSREAD=yes

# Change directory is a must
bash$ cd $HOME/somedir/anotherdir/directory/my_source_code_dir

# Must give vendor tag and revision tag
cvs import somedir/anotherdir/directory/my_source_code_dir Vendor1_0 Rev1_0  

# Also note that it is very important to give the directory tree starting
# from the $HOME, that is, in above example starting from somedir.
# For example I did:
bash$ cd $HOME/howto/foobar
bash$ cvs import howto/foobar Vendor1_0 Rev1_0

# Another example is:
bash$ cd $HOME/javafilesdir
bash$ cvs import javafilesdir Vendor1_0 Rev1_0

# A sample testing and verification:
bash$ cd $HOME/howto/foobar
bash$ cvs checkout myfoo.java

TROUBLESHOOTING: When doing checkout it says module is unknown. It is a common mistake not to change directory while doing cvs import. You MUST change directory to the source-code-directory and then do cvs import. For example:

bash$ cd $HOME/somedirectory/foobardir
bash$ cvs import somedirectory/foobardir Vendor1_0 Rev1_0

3.2 Migrate RCS to CVS

To migrate the existing RCS files to CVS, use the following script. Make sure that you installed the Korn shell package pdksh*.rpm from the Linux contrib cdrom.

NOTE : Get the Korn shell /bin/ksh by installing pdksh*.rpm from the Linux contrib cdrom

#!/bin/ksh

#############################################################
# Program to Migrate the existing source code in RCS to CVS 
#
# Needs the korn shell RPM package  pdksh*.rpm from Linux 
# contrib cdrom
#############################################################

#
# rcs2cvs - convert source tree from RCS to CVS
#

# project to convert
PROJECT='project'

# current RCS root
RCSROOT="$HOME/rcs"

if cd "$RCSROOT/$PROJECT"
then
        cd "$RCSROOT"
else
        echo >&2 "`basename "$0"`: can't change to RCS directory '$RCSROOT/$PROJECT'."
        exit 1
fi

# current CVS root
CVSROOT="$HOME/cvs"

# create new CVS directory for project 'project'
if mkdir "$CVSROOT/$PROJECT"
then
        :
else
        echo >&2 "`basename "$0"`: can't create CVS directory '$CVSROOT/$PROJECT'."
        exit 2
fi

# create CVS project tree from RCS tree
find "$PROJECT" -type d -name RCS -print |
while read RCS
do
        CVS="`dirname "$RCS"`"
        (if cd "$RCS"
        then
#               if find . -type f -name '*,v' -print | cpio -pdmv "$CVSROOT/$CVS"
                if find . -type f -print | cpio -pdmv "$CVSROOT/$CVS"
                then
                        :
                else
                        echo >&2 "`basename "$0"`: can't convert RCS subdirectory '$RCSROOT/$RCS' to CVS subdirectory '$CVSROOT/$CVS'."
                fi
        else
                echo >&2 "`basename "$0"`: can't change to RCS subdirectory '$RCSROOT/$RCS'."
        fi)
done