TWiki > Know > CVS	TWiki webs: Main | TWiki | Know | Sandbox
	Know . { Readme | Changes | Index | Search | Go }

• For All Users: Accessing the CVS Archive
  • Preliminary work to start fetching files from the CVS archive
  • Fetch the files and start working on a project
  • Additional information
    • Graphical interfaces for CVS
    • CVS Environment variables
    • Miscellaneous
• For Project Managers: Creating a New CVS Project Repository in the CVS Archive
• For Project Participants: Creating a New Module in a Project Repository
  • Creating a new module based on existing source code
  • Creating a new empty module called modname
• CVS Command Reference
  • Command Summary
  • Branching
  • Merging
  • Retrieve previous revision
  • Creating a patch
  • Tagging
• CVS Bibliography

For All Users: Accessing the CVS Archive

Preliminary work to start fetching files from the CVS archive

• You need to have login account at mammoth.ctie.monash.edu.au to access the archive. The CVS archive is located at mammoth.
• Make sure that the environment variable CVS_RSH is set to ssh :
  If your login shell is tcsh, add the following line to your .cshrc: setenv CVS_RSH ssh
  If your login shell is bash, add the following line to your .bashrc: export CVS_RSH=ssh
• If you don't want to enter your password every time you do a commit or update, generate your key pair and install your public key. Here are the commands for generating and installing key pairs.

Fetch the files and start working on a project

A CVS Project may contain multiple CVS Modules. Following diagram shows an example.

• Do these just once:
  1. login to supreme.
  2. Create a directory called project_foo and cd to this directory.
  3. Issue the following command from inside your shell prompt:
     cvs -d yasar@mammoth.ctie.monash.edu.au:/home/cvs/project_foo co module_bar. This command will create the directory module_bar in the current working directory. From now on you do not have to issue this command since it is remembered inside the module_bar (if the above doesn't work read about CVS_RSH). environment variable.
  4. You are now ready to work on the module_bar.
• Do these whenever you feel like you want to work on the module_bar:
  1. login to supreme .
  2. cd module_bar.
  3. cvs update -d command will fetch the latest versions of all the files from the repository into this directory.
  4. Start editing stuff.
  5. Issue cvs update to detect whether someone else worked on the files simultaneously while you were editing the stuff.
  6. See whether there are conflicts.
  7. cvs commit will upload your changes to the CVS repository.

Additional information

Graphical interfaces for CVS

• For linux users: cervisia is good one. Its help pages are very useful.
• For Windows users: wincvs

CVS Environment variables

• CVSIGNORE can contain many patterns of filenames that cvs will ignore. e.g. export CVSIGNORE="*_n.cc *.o"
  
• CVS_RSH The remote shell to execute cvs commands with. ssh is the only accepted setting on mammoth for now.
  
• CVSROOT The default repository to use when "-d :ext:user@mammoth.ctie.monash.edu.au:/home/cvs/projectfoo" is not given on the command line and you are not in a working directory. Same format as the -d flag.
  

Miscellaneous

• .cvsignore Each line contains a filename. These files are ignored by cvs. This file is only effective for the directory in which it resides in.
• Emacs has a good mode for cvs called PCL-CVS. Usually its installed already under the Tools menu. It lists at a glance differences in your working directory from the repository. It also contains hooks to editing changelogs for modified files so that when you commit the file it will conveniently put the releveant entries from the changelog into the cvs comment.

For Project Managers: Creating a New CVS Project Repository in the CVS Archive

Assuming you want to create a repository called projname on mammoth granting access to [user1 user2 ... usern]. Note that the specified users must already exist on the system.

• Execute the script /usr/local/sbin/mkcvsproj as root

• mkcvsproj projname [username ...]

• This script will perform the sequence of commands listed above AND add all the members of the ctie group to the projname group
• groupadd on mammoth only allows lowercase - so create lowercase projectnames!! To create a project with uppercase characters - manually modify directory name. [could edit /etc/group and /etc/gshadow as well]
• The username list should specify student accounts to have access to this repository
• Afterwards do not forget to run /root/bin/distribute....

Removing a project

• The script /usr/local/sbin/rmcvsproj will delete a repository
• The script takes one parameter, projname
• This should not be run unless you make a mistake in creating a repository.
• Afterwards do not forget to run /root/bin/distribute....

For Project Participants: Creating a New Module in a Project Repository

Creating a new module based on existing source code

We will now create a new module based on existing source code in directory /home/user/modname. The module to be added to the repository entitled projname.

1. Change to the source directory that contains the files of the module, its name is also modname:
   cd /home/user/modname
2. Import only text files to the repository first:
   cvs -d username@mammoth.ctie.monash.edu.au:/home/cvs/projname import -m "The initial log message for the project" modname ctie start
   (binary files will be added with -kb keyword later).
3. Change to a test directory to checkout the module
   cd /home/user/temp
4. Checkout the module
   =cvs -d username@mammoth.ctie.monash.edu.au:/home/cvs/projname checkout modname=
   This should create a new directory /home/user/temp/modname. This directory will contain the checkout version of the module. Inside this directory is a subdirectory called CVS. this allows you to ignore specifying the -d username@mammoth... parameter in future calls to CVS as long as this is the current working directory. If everything is looking OK you can go back to the cd /home/user/modname directory and add the binary files.

Creating a new empty module called modname

1. Create the source directory that will contain the files of the module, its name will be also modname.
2. Go to that directory, say
   cd /home/user/modname
3. Import this empty module into the project projname:
   cvs -d username@mammoth.ctie.monash.edu.au:/home/cvs/projname import -m "The initial log message for the project" modname ctie start
   
4. Go to the parent directory: cd ..
5. Checkout the new module:
   cvs -d username@mammoth:/home/cvs/projname co modname
6. You can now create new files or copy files into the checked-out directory.
7. Import the new text files into the module repository:
   cvs add textfilename1 textfilename2 ....
8. If you have binary files (gif, jpeg images for example), use this command:
   cvs add -kb binaryfilename1 binaryfilename2 ...

CVS Command Reference

Command Summary

cvs -H <command> Display options for command

• co - checkout the latest revision or previous revisions by date or tag name
  • cvs co -d <dir> <module path> checks out the files under module path and places them into dir (you may wish to include the -P switch as well. Every large project should be checked out with -P switch. It means "prune": -P switch does not check out empty directories.)
  • cvs co -D <yyyymmddhhmmss> <module path> checks out the files from that date and time.
• import - create new module and place in repository
• add - add a file or directory
  • use cvs add -m "LOGTEXT" filename to add filename to the repository.
    Note: if the "-m LOGTEXT" option is omitted, vi or a text editor specified by the EDITOR environment variable will be started automatically to allow a description to be added.
• remove - remove a file or directory (file has to be deleted first)
• ci - commit changes permanently including adds and removes
  • use cvs ci -m "LOGTEXT" filename to commit the changes made to filename .
    Note: if the "-m LOGTEXT" option is omitted, vi or a text editor specified by the EDITOR environment variable will be started automatically to allow a description to be added.
• diff - compare against repository
  • cvs diff -r <revision no.> <file>
• tag - associate a name with the current revision
• log - browse the CVS log for the file
• status - shows the current revision number compared to revision stored in CVS
• update - merge changes from repository in your working directory
  • use cvs update to get other people's changes.
  • cvs update -d will update the working directory with new directories that may have been added by others
• export - same as checkout except without the CVS subdirectories. Useful for creating archives and patches
• rdiff - Create a patch
• rtag - Similar to tag but works on repository directly

Branching

Make sure the module is in stable state and no one is going to check things in.

1. cvs tag -b <branch_name>
2. cvs co -d <new dir> -r <branch_name> <module path> This will check out a copy at the branch point so you can start working in that branch. Any commits you do in <new dir> will increment the branch's revision. Others can continue working on the trunk.
   Alternatively cvs -q update -r <branch_name> will modify the current directory to be part of the branch.

Another possible branch command is cvs rtag <branch_name> <module_name>. This does not require a working directory at all and will create a branch out of the latest sources in the repository.

Merging

Once you've decided the code needs merging back to the trunk do in the latest working copy not the branch working dir

1. cvs update -d -j <branch_name> This merges all changes on that branch onto the trunk. Conflicts will have to be resolved before committing the merges back onto the trunk.
2. cvs tag <tag_name> in the branches copy. This is required if you want to continue working from the same branch after the merge. This way you can merge the changes from this point to the trunk. You need two -j options with the previous command to effect such a merge the first -j is the <tag_name> with the second been the <branch_name>.

Retrieve previous revision

1. cvs -Q update -p -r <revision no> <file> > <new filename>
2. cvs update -A <file> Revert back to the latest (HEAD revision) after applying the above command(removes the sticky tag).

Creating a patch

The following command does not require a working directory (sandbox in cervisia's terminology).

1. cvs rdiff -u -D 20011030 -D 20020130 <module> > diffoutput.txt Creates the patch file diffoutput.txt. This can be applied with patch -p1 < diffoutput.txt in the directory with version as at 20011030 and will bring it up to the version on this date 20020130.
2. cvs diff -r <Tag> Execute this from the working directory to create a patch against a branch or tag name. Useful to compare current changes from a previous reference point in development.

Tagging

1. cvs tag <tagname>

CVS Bibliography

• Extensive documentation and tutorials can be found at CVS home page.

Attachment:	Action:	Size:	Date:	Who:	Comment:
mkcvsproj	action	5040	29 Jul 2003 - 08:14	BrettPentland	Removed code that adds CTIE staff to project
rmcvsproj	action	2969	31 Mar 2002 - 06:55	AhmetSekercioglu