GENIE Quick SVN ref

What is Subversion?

Subversion (abbreviated to SVN) is the successor to CVS. It is closely related to CVS and so many of the commands are the same. Subversion still has the commands checkout, commit and update, except that, for example, cvs update is replaced with svn update. Happily this means that there is little re-learning required to use SVN instead of CVS. If CVS and SVN were entirely alike, we would not get the advantages that Subversion brings. Subversion addresses a number of the shortcomings of CVS. For example, the command svn move now exists, which allows you to rename a file without losing it's history, something not possible under CVS.

Using Subversion on your local Machine

First, just as it was with CVS, you will need a Subversion client installed on your machine. If you are using linux, then you will most likely have the command line client already installed. Try typing "which svn" to see if the clent is in your path. Command line and nice GUI clients for SVN are available for just about any operating system. tortoisesvn is popular for windows. SVNx is a good choice for the Mac. For more links see the links page on this wiki.

Once you have Subversion installed, you will need a username and password. If you haven't already them, email Gethin. You can get read-only access to the code by using the username "genie-user" and the password you would expect. Happily, access is considerably easier than for CVS. You will be prompted for your password when required. SVN caches your password, and so you often do not need to type it.

Another big difference between CVS and SVN is that repositories are accessed through a URL. For example, you can checkout GENIE from SVN by typing:

svn checkout http://source.ggy.bris.ac.uk/subversion/genie/tags/rel-2-1-0 --username <your_username> ~/genie

Quick Start

The above example showed how to obtain one of the stable releases of GENIE using the SVN command-line client. See the GENIE versions page for more information about what stable releases are available and what they have been used for. We will continue with more command-line examples below:

If you are developing GENIE, you will want to get hold of the HEAD version of the model:

svn co http://source.ggy.bris.ac.uk/subversion/genie/trunk --username <your-username> ~/genie

To update an existing checkout, simply cd to the highest-level directory (let's assume it's ~/genie) and issue the update command:

cd ~/genie
svn update

(You could update a sub-tree of your checkout by changing directory to, say, ~/genie/genie-embm and then calling "svn update".)

Now cd to the directory you will be working in.

If you have created a new file, you must first add it to your local checkout. (The action of adding it tells SVN that there is a new file which the repository must store, and schedules it for addition to the main repository when you next call "svn commit". More of svn commit in a moment):

svn add <filename>

or, for example:

svn add *.f90

To add a group of files en masse.

Similarly, to delete a file:

svn delete <filename>

or:

svn del *.f

Once you have made your local changes, you should test them before committing any of them back to the repository. Type:

make test
make testebgogs
make testbiogem

to run the IGCM- and EMBM-based short tests respectively (note that the BIOGEM test is based on an EBGOGS config').

You can check which files you have changed using:

svn status

Once your tests pass with an "**OK**", then you can go ahead and commit your changes. To commit a single file, replace <my_modified_file> with the appropriate filename and provide a meaningful message log in the following command:

svn commit -m "type your message here" <my_modified_file>

A successful commit will then advance the version number on the SVN repository by 1. Thus we can see that SVN uses version numbers differently to CVS. SVN gives the entire repository a version number and increments this whenever there has been a change. This is unlike CVS, where individual files have their own individual version numbers.

We can commit all modified files in the current directory if we omit the filename:

svn commit -m "type your message here"

A useful enhancement over CVS is the ability to move (or rename) a file without losing it's history:

svn move <before> <after>

To see the history of a given file, type

svn log <filename>

If things go a little pear-shaped with your checkout, you can go back to the last state obtained from the repository with:

svn revert --recursive

Note that --recursive will execute the command over a whole file-tree.

To see the current status of your checkout, e.g. to check what files you have changed:

svn status

To see the differences between your code and the 'head' code:

svn diff <filename>

To find out more about the various SVN commands type:

svn --help

or for the the usage of a particular command, e.g.:

svn copy --help

Further Documentation

Rather than reinvent the wheel and try and document all aspects of Subversion here, it would be better if you looked at the | Subversion user manual.