home | tech | misc | code | bookmarks (broken) | contact | README


CVS notes

CVS setup

This file shows the set of the commands to create a CVS repository in a local computer and start to work with it. Useful information can be found in the cvs(1) manual page and here.

First, export your CVSROOT environment variable for the repository root directory. For example:

$ export CVSROOT=/home/user/cvs

Then, create the repository with the following command:

$ cvs -d /home/user/cvs init

After creating it, cd your project root directory and use the command cvs import to put files there:

$ cvs import -m "Log Message" project_name_in_cvs/ vendortag releasetag

I don't know, yet, what vendortag or release tag mean.

To download information from CVS to your working directory, run the following command on the a directory where you put project directories (See that, unlike cvs import, you don't run it in the project root directory, but one level above.):

$ cvs checkout project_name_in_cvs/

Now, suppose you changed a file and need to update the CVS repository. Just commit it:

$ cvs commit -m "commentary for changelog" filename

See that you need to checkout the directory that the file you are editing is in before doing the commit.

Now, if you need to add a new file to the repository, use the add command and commit it:

$ cvs add filename
$ cvs commit -m "commentary for changelog" filename

I got some errors when I tried to add a file whose filename is the same as the project name. Don't know why... maybe I was doing something wrong.

Ok. It might be enough to set and use a simple cvs local server. The next steps will show us how to configure the system to allow remote use, i. e., allow users of other machines (on a LAN or not) commit, checkout, add and delete files.

On BSD systems (at least on NetBSD) it is not necessary to set the CVS_RSH environment variable (according to a friend), but it is on other systems. It is made to use the ssh protocol. Steps for using other protocols are different.

$ export CVS_RSH=ssh

Making a commit remotelly is exactly the same as making it locally. Also is other operations, but the repository address is diferent, so it is necessary to pass the -d flag to the cvs command, or set the CVSROOT environment variable, with the following content:

export CVSROOT=:ext:user@host:/path/to/repository

It is enough, for now.

CVS tips

Syncing trunk and branch

Sometimes it is necessary to get a change made in HEAD and add it in a Branch. To make this sync, do:

  1. checkout the branch
  2. update -r HEAD the files
  3. commit the new files in the branch

Sometimes it is necessary to do the opposite: sync changes in branch to the trunk. This is a good reference that I got to make it and type the commands above:

$ cvs checkout $module
$ cd $module
$ cvs update -j HEAD -j $branch
$ cvs commit

Tags and branches

To tag a repository, enter at the workdir and type:

$ cvs tag <tagname>

To create a branch from a tag, type:

$ cvs tag -r <existing tag> -b <branchname>

To delete a tag use the -d flag:

$ cvs tag -d <tagname>