Subversion

Resources:
A. http://svnbook.red-bean.com/
B. http://subversion.tigris.org/
C. http://queens.db.toronto.edu/~nilesh/linux/subversion-howto/
D. http://www.linuxfromscratch.org/blfs/view/svn/server/svnserver.html

The resources listed above are pretty good and can help you figure out all you want to know about Subversion, but you may have to dig around a little to extract some of the minutia.

Subversion is a client/server version control application.

There is a choice of two servers; Apache + mod_dav_svn, and svnserve. The former is more complex and slower. The latter, svnserve is the one discussed here, and is installed automatically along with all the other pieces using aptitude install subversion subversion-tools

There is one client; svn and it is installed automatically with all the other pieces using aptitude install subversion subversion-tools

Subversion is a centralized system for sharing information. At its core is a repository, which is a central store of data. The repository lives on the real file system in the place you put it using mkdir /path/to/repository followed by svnadmin create /path/to/repository. Subversion's resources exist in a virtual filesystem which is rooted in the real file system repository directory (/path/to/repository in this example). A real world example is /var/svn/repos

Most of the reference material doesn't bother to explain that svnadmin create doesn't recursively create directories (like you do with mkdir -p). The path must already exist, or all but the last directory must exist for svnadmin create to work. If, for example, you follow the example given in http://svnbook.red-bean.com/en/1.1/ch05s02.html you will receive the error:
svnadmin: Repository creation failed
svnadmin: Could not create top-level directory
svnadmin: Can't create directory '/path/to/repos': No such file or directory
This can be very discouraging for a bottom-up type of person trying to understand what's going on. The last error message was especially confusing to me because it's telling me the path can't be created because it doesn't exist. Huh? It doesn't seem "an incredibly simple task" to me!

While perusing Resource A, Chapter 2, keep in mind:

• The directory created by svnadmin create holds all the subversion system files and will include all the versioned projects added with svn import and svn add but it will not add your project directories to the real file system so don't bother to go look for them. You have to use svn commands to see them, or something like svnlook tree
• The svn import command is a quick way to copy an unversioned tree of files into a repository creating intermediate directories as necessary, and can also be used to copy a single file (but not a group of files); but there's a difference in the syntax depending on whether you are copying a tree or just copying files. svn import my_file file:///var/svn/repo/files will _not_ put your file in /var/svn/repo/files/ but will, instead, put your file in /var/svn/repo/ with the name files. If, on the other hand, you have a directory /files/my_file somewhere in the file system, that command will do what you expect. i.e. put your file in /var/svn/repo/files/. In other words, it acts like cp except that cp won't creating intermediate directories.
• The original, unversioned tree is not, and will not become part of the versioning system and can be deleted once imported.
• To begin work on a project it must be checked out with svn co file:///var/svn/repo/files or whatever you named your tree. The checked out tree will be placed in the current working directory or the desired directory can be specified in the svn co command. You cannot check out a file; you can only check out a directory or tree of directories.

Some examples:

Install subversion: aptitude install subversion subversion-tools

Setup a repo: mkdir /var/svn; svnadmin create /var/svn/repo

Populate a repo: cd; svn import work file:///var/svn/repo/work

Checkout the project: cd ~/my_project; svn co file:///var/svn/repo/work

Commit the changes: svn commit

Start all over, erasing everything: rm -rf /var/svn/repo; rm -rf ~/my_project

There are three ways to invoke svnserve:

Invoke svnserve as a standalone "daemon" process:

Start the server: svnserve -r -d /var/svn

Check out a project: svn checkout svn://host.example.com/repo

Invoke svnserve using inetd:

Make entry in /etc/inetd.conf: svn stream tcp nowait root /usr/bin/svnserve svnserve -i -r /var/svn

Retart the inetd superserver: /etc/init.d/openbsd-inetd restart

Check out a project: svn checkout svn://host.example.com/repo

Invoke svnserve in tunnel mode over SSH:

Edit /var/svn/repo/conf/svnserve.conf:
[general]
anon-access = read
auth-access = write
realm = My First Repository
password-db = passwd

Check out a project: svn co ssh+svn://user@host.example.com/var/svn/repo

Working with files: Assume your project is in the repo directory www and you want to add a single file. svn import myfile file:///path/to/repo/www/myfile -m "importing myfile" should do it.
You cannot check out that single file; you can only check out directories but the svn checkout command has a -N option that prevents recursion so all you check out are the files in the specified directory: svn co -N file:///path/to/repo/www will do it.

Send mail to the Webmaster

	This site best viewed with a browser
	Warning: This is a Debian centric site and MAY contain peanuts.
	Many thanks to Debra Lynn and Ian Murdock for making Debian possible
	First created Apr 22, 2008 ~ Last revised August 23, 2010