How To Compile and Install a 2.6.x Series Linux Kernel

The Linux kernel is the core component in any Linux distribution. Without a kernel, your computer would be essentially useless. It is the piece of software which allows interaction between you, your computer's applications, and your computer's hardware. With such a powerful role in your computing experience, it is important to keep your kernel up-to-date. Each new release provides more hardware support and many performance enhancements. It is also important to keep your kernel up-to-date for security purposes.

Let's upgrade our Linux kernels together. I will walk you through each of the steps I take, from beginning to end, to upgrade my kernel. Just as a warning, I prefer to do the whole process on the command line, so you might want to pull up a terminal, konsole, xterm or whatever you prefer to use for your command line operations.

First you need to download the kernel source code. Many Linux distributions provide specialized editions of the Linux kernel. Typically, you don't want to manually compile and install a custom kernel for these distributions. This does not mean that you can't, it simply means that you might be better off using the "official" kernels for your distribution, which can usually be obtained through your distribution's package manager. You can get the official, 100% free, and complete Linux kernel source code from http://www.kernel.org/. Look for "The latest stable version of the Linux kernel is:" and click the link on the F on the same line. Currently, the latest stable version is 2.6.20, and that's what I'll be using for this tutorial. Please note that commands which begin with a dollar sign ($) are executed as a regular user and commands beginning with a pound sign (#) are executed as a superuser.

$ cd /home/user/download
$ wget http://www.kernel.org/pub/linux/kernel/v2.6/linux-2.6.20.tar.bz2

Now login as the superuser, and navigate to the /usr/src directory. Then extract the kernel source into that directory.

$ su -
# cd /usr/src
# tar jxf /home/user/download/linux-2.6.20.tar.bz2

You probably already have a symlink or shortcut called linux which points to your most recent kernel. If you do, delete the link and create another link to the new source tree. Then go into your kernel source tree.

# rm /usr/src/linux
# ln -s /usr/src/linux-2.6.20 /usr/src/linux
# cd /usr/src/linux

I like to identify each compile of my kernel uniquely, to make sure that I'm using the right one. To do that, you have to modify your Makefile

# vi Makefile

You will see the following lines, or something similar, at the very top of the file:

VERSION = 2
PATCHLEVEL = 6
SUBLEVEL = 20
EXTRAVERSION =
NAME = Homicidal Dwarf Hamster

Change the EXTRAVERSION property to something you want to use to identify this kernel. I will use -jcv1

EXTRAVERSION = -jcv1

The rest of the Makefile should be fine. In fact, I discourage editing Makefiles unless you know what you're doing. This next step is totally optional, but I like to do it to save some time. You can copy your existing kernel's configuration file in order to have a very similar kernel configuration. My previous kernel version was 2.6.19.1, so this is the command I use:

# cp /usr/src/linux-2.6.19.1/.config /usr/src/linux/

Then I run make oldconfig or make silentoldconfig to update my older kernel configuration file to be able to handle newer features. If you use oldconfig you are required to specify whether or not you want the new features included in your kernel, whereas silentoldconfig will use the defaults determined by kernel developers (they usually know best), asking for minimal input. Let's update our configuration file and then customize it by running make menuconfig (there are several options here, such as make xconfig and make gconfig, but I prefer the text-based menuconfig; there is another you can run by using make config, which runs through each and every option available--it's scary).

# make silentoldconfig
# make menuconfig

menuconfig is a graphical command line application which lets you navigate the features offered by the kernel. Each computer is considerably different from the next, so it really does no good to provide a list of things that I tweak. However, it is important to note what some of the symbols are in the menuconfig utility:

  • M = Module. Modules are loaded when they are required and can contribute to the speed of your system
  • * = built into the kernel. These are typically things which are necessary for your machine to function properly, such as support for your root file system.
  • X = exclusively selected. You'll see this when you select what type of processor you have, for example.

One thing to note before we go further is MAKE SURE YOU KERNEL HAS BUILT-IN SUPPORT FOR YOUR ROOT FILE SYSTEM!!!! My root file system is reiserfs. In my configuration, I made sure that reiserfs was marked with a star. If you don't do this, your kernel won't boot and you will be very frustrated. Trust me.

Your computer is probably quite different than mine, so you might want to just poke around and see if you recognize things that deal with your computer's hardware. Once you are done tweaking your kernel configuration, exit the configuration utility and make sure the configuration is stored in /usr/src/linux/.config

Next we get to build and install the kernel. After that, we have to add an entry to our boot manager so that we can try out our new kernel. The compilation part usually takes just about a half hour on my 2.2Ghz Turion64 processor with 1.25GB of RAM. It takes about 6 hours on my 300Mhz Pentium 2 with 32MB of RAM. Let's find out how long it takes for you to compile your kernel!

# time make
...
real    27m29.663s
user    23m34.476s
sys     2m56.575s

Now let's install the modules and install the appropriate files in the boot area:

# make modules_install
# make install

This is the part that always used to mess me up. I use Slackware Linux, which is more UNIX-ish than most distributions. It's actually the oldest surviving Linux distribution to date, but that's another story. For some reason, the make install command doesn't always work with Slackware. There is a process I use to setup my boot directory when I compile a new kernel. I wrote a simple shell script called fixkernelinstall to take care of it for me:

#!/bin/bash
# Configure my computer for a new kernel
# Author: Josh VanderLinden
# Assisted By: Dan Purcell

# if the user didn't supply a kernel number, ask for it
if [ $# -eq 0 ]; then
    echo -n "Kernel: "
    read kernel
else
    kernel=$1
fi

# determine root partition
echo "Determining root partition..."
rootpart=`mount -l | grep ' / ' | cut -f 1 -d\ `
echo "Root partition is $rootpart"

# copy kernel configuration file
cp /usr/src/linux/.config ./config-$kernel

# now rename everything
echo "Renaming files..."
mv System.map System.map-$kernel
mv vmlinuz vmlinuz-$kernel

# if the config file exists and it's a symlink, remove it
if [ -f 'config' -a `stat config | grep -c 'symbolic link'` = '1' ]; then
    echo "Removing link to configuration file"
    rm config
else
    # otherwise it might be important
    echo "Renaming configuration file"
    mv config config.bak
fi

# Link files
echo "Creating symlinks..."
ln -s System.map-$kernel System.map
ln -s config-$kernel config
ln -s vmlinuz-$kernel vmlinuz

# Update lilo
echo "Adding entry to /etc/lilo.conf for $kernel"
echo "image = /boot/vmlinuz-$kernel" >> /etc/lilo.conf
echo "  root = $rootpart" >> /etc/lilo.conf
echo "  label = $kernel" >> /etc/lilo.conf
echo "  read-only" >> /etc/lilo.conf
echo "Linux kernel $kernel has been configured."
echo "Please check your lilo configuration and run lilo before rebooting"

I'm not an expert on shell scripts, so please feel free to offer suggestions for doing things better if you know how. This script uses the kernel version (given by the user) to setup by /boot directory properly. In my case, I run the script as such

# cd /boot
# fixkernelinstall 2.6.20-jcv1

And the output is something like:

Determining root partition...
Root partition is /dev/hda5
Renaming files...
Renaming configuration file
Creating symlinks...
Adding entry to /etc/lilo.conf for 2.6.20-jcv1
Linux kernel 2.6.20-jcv1 has been configured.
Please check your lilo configuration and run lilo before rebooting

As you can see from the script, I use LILO instead of the arguably more popular GRUB. Either one works for me, but LILO is sufficient for my needs. If you want to use the same kind of script for a GRUB installation, just change the LILO part at the end to something like:

echo 'Adding entry to /boot/grub/menu.lst for $kernel'
echo '  title Linux on ($rootpart)' >> /boot/grub/menu.lst
echo '  root (hd0,4)' >> /boot/grub/menu.lst
echo '  kernel /boot/vmlinuz-$kernel root=$rootpart ro vga=normal' >> /boot/grub/menu.lst

Make sure you change the line with root (hd0,4) to fit your setup. With GRUB, you don't have to worry about applying changes to see the menu entry at boot. It's automatically there. With LILO, however, you have to actually apply changes each time you make them. You do this by running the lilo command as the superuser:

# lilo
Added Windows
Added Linux
Added 2.6.20-jcv1 *

The star (*) signifies the default kernel to boot. Make sure that your root partition is correctly specified in your boot loader configuration. My root partition is on /dev/hda5, but yours may be (and probably is) on a different partition. If you fail to specify the correct root partition, your system will not boot that kernel until the configuration is fixed. GRUB makes this a lot easier than LILO.

And this is the point when you start to cross your figures and hope that your computer doesn't blow up... We get to reboot our computer and hope that our configuration file plays well with our computer. So, let's do that! See you in a few minutes (hopefully).

# shutdown -r now

So here I am, back on Linux on my freshly-rolled kernel. I hope you are as successful as I have been this time around. Keep in mind that you have to reinstall custom kernel modules if you installed others while you were on your other kernel. For example, I use ndiswrapper to access wireless Internet. I have to recompile and reinstall the ndiswrapper module and device drivers before I can use wireless. Likewise, I have VMWare Server on my laptop, which installed special modules. I have to run vmware-config.pl to reconfigure VMWare Server for my new kernel before I can run any virtual machines.

To summarize, here are the commands that I used in this tutorial. Remember that lines beginning with a dollar sign ($) are executed as a non-privileged user, while lines beginning with the pound sign (#) are executed as the superuser (root).

$ cd /home/user/download
$ wget http://www.kernel.org/pub/linux/kernel/v2.6/linux-2.6.20.tar.bz2
$ su -
# cd /usr/src
# tar jxf /home/user/download/linux-2.6.20.tar.bz2
# rm /usr/src/linux
# ln -s /usr/src/linux-2.6.20 /usr/src/linux
# cd /usr/src/linux
# make clean
# vi Makefile (to change EXTRAVERSION to -jcv1)
# cp ../linux-2.6.19.1/.config .
# make silentoldconfig
# make menuconfig (just to ensure settings were good)
# time make
# make modules_install
# make install
# cd /boot
# fixkernelinstall 2.6.20-jcv1
# vi /etc/lilo.conf (to make sure things were good)
# lilo
# shutdown -r now

I hope that you are able to use this tutorial to successfully install or upgrade your kernel. Good luck! Any comments or suggestions are welcome!

How to Upgrade a NetBeans Project to JDK 6

NOTE: I am currently using NetBeans 5.5.

Here are the simple steps:

  1. Download and install JDK 6 if you haven't already
  2. Add the JDK 6 platform to NetBeans
    1. Tools > Java Platform Manager
    2. Add Platform
    3. Navigate to JDK 6's installation root directory. It will probably have a fancy overlay icon over it.
    4. Click Next
    5. Give the platform a name and continue onward
  3. Open a project in NetBeans
  4. Right-click on the project and open its properties
  5. Open the Libraries item on the left
  6. Change the Java Platform to your new 1.6
  7. Open the Sources item on the left
  8. Change the source level at the bottom to 1.6
  9. Enjoy!

There really are quite a few nice features in this release. So far I've found time to play with the built-in JTable sorting methods and the system tray icon stuff. It's so much easier than it used to be!!

MySQL on Slackware

Many of us who have installed Slackware on our machines in the past few years have noticed something annoying on the first boot: the MySQL service fails to start!!

In response to this, I would like to offer this simple tutorial. Right now I am doing this blindly, meaning I don't have a fresh install to work with. Please bear with me if you notice some errors in the tutorial, and please tell me about them!

  1. Log in as root: # su -
  2. Make sure the mysql package is installed. You can do this by running pkgtool and selecting the view option. Hit n and it will take you to the packages in the list that begin with n. MySQL should be right above the n packages. If you don't have mysql already installed, you can download it from http://www.linuxpackages.net/ or wherever you'd like. Once you have a copy of it, install it by using the installpkg mysql......tgz command
  3. Setup the databases: # mysql_install_db
  4. Apply the proper permissions: # chown -R mysql.mysql /var/lib/mysql
  5. Start the database: # mysqld_safe &
  6. Setup the root user:
# mysqladmin -u root password 'newpassword'
# mysqladmin -p -u root -h localhost password 'newpassword'

Finally, test your installation:

# mysql -p

That should do it! Please comment if you notice any errors with this posting.

Installing SuSE Linux Enterprise Desktop 10

This will be the second time for me putting SLED 10 on my laptop, but I've also put SLES 10, SuSE Linux 10.1, and various others on this laptop several times before. It _has_ been a few months since I last installed Linux on this laptop, so we'll see how well I remember how to do it. I will be installing the 64-bit DVD version, so as to take better advantage of the 64-bit capabilities of my processor. This shouldn't have a drastic effect on the overall procedure, compared to that of installing the 32-bit CD/DVD version.

Here are some specs of my laptop:

  • Make/Model: HP Pavilion dv8000
  • CPU: AMD Turion 64 ML-40 (2.2Ghz)
  • RAM: 1.25GB PC2700 DDR333 SODIMM
  • HDD: 5400RPM 80GB
  • Video: ATI Radeon Xpress 200M (128MB dedicated RAM, up to an additional 128MB shared RAM)

Note: I make a few assumptions in the writing of this article. One is that you are on a machine running Windows XP. If your computer can't handle Windows XP, you probably don't want to be running SLED 10. Another assumption is that you don't yet have your hard drive partitioned into more than one partition. I also assume that you already have the installation media in good working condition. For those of you in the BYU-I Linux Users Group (LUG), I am willing to make copies of the discs if you provide the media or discuss some sort of compensation if you want me to provide the media.

BACKUP ALL IMPORTANT DATA BEFORE PROCEEDING

We all hate losing the projects that we've slaved over for weeks and months. Take the proper precautions to backup anything you wouldn't like to lose before installing any flavor of Linux. That's not to say that you will lose everything, but it's not unheard of to wipe out all data from your drive while attempting to install Linux. With that warning out of the way, let's get started!!!

Defragment Your Hard Drive

If you have a secondary drive which you plan to dedicate to Linux, this step is not necessary. However, if you plan to install Linux on the same drive as your Windows installation, I would suggest defragmenting your drive prior to repartitioning your drive. In order to defragment your hard drive in Windows XP, open your Start menu and open the Control Panel. Once here, descend into Administrative Tools and run Computer Management. This utility is quite handy. On the left side of the Computer Management window, you should see a tree of options. Under Storage you will see the Disk Defragmenter. Simply click the Defragment button, and the program will begin optimizing your files. Defragmenting your hard drive basically puts each of your files into one piece instead of scattered across the drive (as they tend to be written). Defragmentation is a good process to run on a regular basis. Once again, this step is not necessary to the installation of Linux, but it is a good practice.

Begin The Installation

Once you're done backing up your files and defragmenting your hard drive, insert the first SuSE Linux Enterprise Desktop 10 CD/DVD into your CD/DVD drive and reboot your computer. You should be presented with a fancy blue welcome screen, but if you see the Windows boot screen, you'll have to change the settings in your BIOS to enable you to boot from your CD/DVD drive. That process will vary from machine to machine, so I won't even try to explain how to do it. Once you see the welcome screen, you are presented with several boot options:

  • Boot from Hard Disk
  • Installation
  • Installation--ACPI Disabled
  • Installation--Local APIC Disabled
  • Installation--Safe Settings
  • Rescue System
  • Memory Test

We're going to go with the first Installation option. In my experience, starting with SuSE Linux 10.1, I haven't had any issues at all getting Linux installed. Most people shouldn't have problems booting into the installation program, but if you do, try the other Installation options. Once we begin the installation process, the bootloader will load the Linux kernel into memory and begin booting the SuSE Linux Enterprise Desktop installation utility. It will probe several devices to make the installation usable, after which you are presented with a language selection menu. Choose the language of your preference. Next you will be asked to accept the License Agreement, which I would recommend reading (with any product, not just SLED). If you accept, check the "Yes, I Agree to the License Agreement" option box and hit "Next." If you don't accept the agreement, this is the end of the road--your installation process will end.

Once we accept the License Agreement, a few more devices are probed and we're asked what kind of installation we'll be doing. With the assumptions I've made for this tutorial, we will proceed with a New Installation. If you have the "extras" CD, you can check the "Include Add-On Products from Separate Media" checkbox. You'll then be asked to insert the disc so that a catalog of available applications can be made. However, I'll assume that we all just have the required media. Click "Next" to create a catalog of the software available.

Now we're presented with a Clock and Time Zone screen. Choose the appropriate options for your situation and click next.

Installation Settings

Once we have set our clock and time zone, we are shown an overview of the current installation settings. I personally prefer to see all of the details, so I am going to click on the Expert tab. I don't like the predefined partitioning scheme, so I am going to change that.

Partition Your Drive

Partitioning a hard drive basically allows you to split up that brand new 500GB SATA-II drive you bought into smaller "virtual hard drives." I like to partition my hard drive because it allows me to manage my files easier, and I don't have to worry about losing ALL of my data if one of the partitions needs to be reformatted. Each partition can have a different file system on it, which allows us to run a Windows file system (NTFS) and a linux file system (ext2/3 and reiserfs are the two main ones, at least for workstations) on the same physical hard drive.

This is the first time I will entrust all of the data on my hard drive to a Linux installation partitioning utility in a very long time. We're talking about 7 years... However, for the sake of others, I am willing to put it to the test to see if SuSE will not wipe my drive when I try to resize the Windows partition. I usually use a utility such as Partition Magic to resize and create new partitions.

To change my partitioning scheme, I click on the "Partitioning" subtitle on the Expert tab in the Installation Settings section. I want to base my partition setup on the default proposal, so I select the second option "Base Partition Setup on This Proposal" and hit next. This part could be a bit hairy if you've never partitioned a drive before. I want to be able to share files between Windows and Linux, so I am going to create a small ~20GB partition which I will format to be FAT32, a format readable and writable in both Windows and Linux.

First, I must resize my Linux partitions. I don't need my home partition to be 22GB, so I'm going to resize that one to be 5GB. To do that, I select the partition with /home listed as it's Mount point and click the "Resize" button at the bottom of the screen. The window that appears shows a graphical representation of the changes we make. All I need do is enter "5" into the "Space Free (GB)" box or move the slider to the right spot and click "OK". Now I'm left with 17GB to share between Windows and Linux. To create this new partition, I click on the "Create" button at the bottom.

The new partition window asks me what type of format I wish to have on the new partition--I want to select FAT. I want to have this partition listed in a place that makes sense to me, so I'm going to change the "Mount Point" field to /windows/share and click OK. I think I'm now satisfied with the partitioning scheme, so I click "Finish" to return back to the Installation Settings screen.

Partitioning Pointers

Let me share some pointers for partitioning schemes. Traditional Linux installations would ask for a partition twice as big as the amount of RAM you have in your machine. This is for the swap, which is synonymous with virtual memory in Windows lingo. That means that if you have 512MB of RAM, your swap partition should be at least 1024MB (1GB). Likewise, if you have 1GB of RAM, your swap partition should be at least 2048MB (2GB). In my opinion, the average Linux desktop does not require more than 512MB for a swap partition. I may be mistaken, but I think the "double your RAM" rule became somewhat obsolete for desktop workstations with the advent of 2Ghz+ processors with 1GB+ of RAM. It could just be me, but I've never even filled 256MB of swap. Just something to consider while partitioning your drive.

If you plan on experimenting with several distros of Linux without wiping other installations of Linux, I would recommend a partition dedicated to your /home folder. This way, you are able to keep your personal settings across most if not all distros. I've found it useful on countless occasions.

Software Selection

One thing I really like about SLED is the ease of package selection. Their default package selection will suit most people just fine. However, I have developed my own tastes for how I like my Linux, so I am going to customize the package selection a bit. To do that, I click on the "Software" heading in the Expert tab of Installation Settings.

I personally prefer KDE to GNOME as my window manager. So I am going to deselect GNOME from the Desktops category, but not so the "Do Not Enter" symbol shows up where there once was a check. I want to click the checkbox until I see a white box (no checkmark). I'm not sure if this is required, but usually different environments will require libraries from other environments in order for certain programs to run. I suspect that the Do Not Enter sign means that nothing for GNOME will be installed, but this is not fact--it's simply a notion of mine. Now I want to put a check in the checkbox next to KDE. Being a nerd, I want to have my compilers around, so I will also select that option.

According to the disk usage graphs in the bottom right of the screen, I'm only going to be installing about 1.9GB of software. That's interesting because I downloaded a whole DVD... If anyone wants to see what other software is available, you can click the Details button below the software category list. This might scare a few off, but it's all quite simple. If you want to see more categories to choose from, select "Package Groups" from the Filter list in the top left. This is where you can explore all of your software options available on your installation media. I am going to leave that sort of customization until after I'm all installed and running.

Once you're done selecting the packages you wish to have installed, click Accept from the bottom right. You will probably encounter a few more license agreements at this point. These are for non-open-source applications (Adobe Acrobat Reader, Macromedia Flash plug-in, etc) included with SLED. I recommend reading and accepting each license agreement. Now I am presented with the same Expert tab in the Installation Settings stage. Now we're ready to proceed, so click Accept in the bottom right again. We're asked to confirm that we want to install Linux, with a warning that certain parts of your hard drive will be formatted, thus erasing any data that were there before. If you're ready, click Install, sit back, and enjoy.

Installing Everything

At this point your partitions will be resized/formatted and the appropriate files will be installed. This process can take anywhere from 20 minutes to 2 hours or more, depending on the packages you selected and the speed of your system. You might be interested in seeing what exactly is being installed on your system at this point. If so, you can click the Details tab and see each package being installed. Apparently I chose some other packages along the way or something, because now it says that I'm installing about 2.5GB of software and that this segment should take about 30 minutes.

Once all of the files are copied, the installation settings will be saved to the hard drive and your system will reboot for the first time in your brand new SLED. At this point, you shouldn't remove your installation media, as it is required in the following steps.

The Initial Boot And Final Settings

When we boot up our SLED for the first time after installation, we are asked to provide a hostname to identify our machine on the network. When you have set the hostname and domain name as you want them, click next. Now we're asked to set the root password. Make sure this password is one that you'll not forget, but at the same time make sure that it's not easy to guess. If someone gets root access to your machine, there's no end to what they can do.

Network Configuration

This is another section that is mostly correct, but a few settings are not the way I would like them to be. For example, my ssh port is listed as disabled under the Firewall heading. To enable it, just click the word "blocked"--it will change to "open". The rest of the settings look fine for now. If you have any customizations to be made, go ahead and make them. I'll wait.

When we're ready to move on, click next. At this point, our network configuration is saved. Next we're asked if we wish to try out our Internet connection. Do as you please. I usually skip this step, but for your sake, I will try out my connection. When we test, it tries to download the latest changelogs. If your connection works, you will see "Success" in the Result field. Click next.

User Authentication Method

Most home users won't have their own LDAP server or Windows Domain setup, so I won't go into how these are to be set up. Let's just go with Local authentication for now, the default option. Click next.

New Local User

This is when we create our very own user account. This set is essential. DO NOT EVER RUN EVERYDAY APPLICATIONS AS ROOT. There are serious security implications involved if you choose to login and perform your daily tasks as the root, or all-powerful administrator, user in Linux. It's much easier to just create an "unprivileged" user and do your regular business with that account. Only login as root when you need to perform system maintenance or install something. When you're done with those tasks, logout of the root environment immediately. Trust me.

Anyway, back to our installation. Go ahead and create your user. You may or may not want to check the checkbox to receive system mail. System mail includes certain security breaches on most distributions. If you don't wish to have to enter your password in order to use your computer, click the automatic login checkbox. If you wish to add more than one user at this time, you can click on the User Management button. The process is pretty much the same as it is to add the first unprivileged user. Click next when you're ready to proceed.

Now our system configuration is saved again (this seems to happen all the time in SuSE... it gets rather annoying in my opinion). After our settings are saved, we are presented with the release notes (which may have been more useful had they been displayed during the file copy process, but whatever). Read them if you wish. Click next when you're ready to proceed.

Hardware Configuration

Now this is one of the selling points for me with SuSE. I have a 17" widescreen (1680x1050 max resolution) for my laptop. There weren't many distros for a while that could handle the resolution out of the box. Fedora Core 5 was the first that I tried that handled it without any manual configuration, and SuSE was the second. I'm pretty happy with the configuration listed here, so click next when you are too. Once again, the settings will be saved (seems like saving settings in SuSE is as bad as rebooting in Windows...).

If you have several similar machines, you can save your installation configuration by checking the "Clone This System for Autoyast" checkbox. If you choose this, the system will determine what settings exactly were used for installation and create a file somewhere that you can use in later installations. When this is done, or when you click finish if you don't want to clone, a login screen will appear.

First Login

When you see this login screen, enter the username and password that you created for the unprivileged user. You'll see a fancy loading screen while your profile is being created for the first time.

Now, you may or may not have noticed, but I wrote this article as I installed SLED. I want to watch a movie now, so subsequent configuration (wireless, 3D acceleration, etc) will take place later. I hope this is good enough for the time being.

How To Create A Subversion Repository

In this tutorial, I assume that you have access to a Linux machine on which you plan to create the repository. I have never tried to create a repository on Windows or Mac, but perhaps in the future I will. I am doing this on a Slackware 10.2 machine with subversion 1.3.2. Version differences shouldn't have a great effect on the validity of the steps outlined here.

Install Subversion

If you haven't already installed subversion, you can download it from http://subversion.tigris.org/. You can download the svn package for a variety of platforms. If you download the source package, you should simply have to do the following as root:

  • unpack the archive and enter that folder
  • configure the application for your system by typing ./configure on the command line
  • compile the source code by typing make
  • install the package by typing make install

Note: these are the general steps for installing a package from the source; they may be slightly different with svn.

Choose A Home For The Repository

Next, choose a place for your server to keep track of all of the changes to your project. Try a place like /var/svn or /home/[your_username]/svn. I would recommend creating a folder specifically for the repository.

Create The Repository

Once you're in the directory in which you wish for your svn repository to reside, type the command svnadmin create repo or you can type the full path to be certain svnadmin create /var/svn/repo. This will create the repository.

Import Your Project

I do most of my development on my local area network, so I haven't created a repository that can be used over the Internet. I would imagine that it's pretty similar to what we do to create a repository for your LAN. Choose a project to import into your repository and navigate to the folder that the project is in. For example, if I have a project called Foo in /var/www/htdocs/foo, I would go to /var/www/htdocs. Next, use the command svn import to pull your project into the repository. For our Foo project, we would use the command svn import foo svn://localhost/var/svn/repo/foo -m "Initial Import". This will create a new folder in our repository at /var/svn/repo called foo (/var/svn/repo/foo). Our project files will put crammed into that directory for tracking.

Checkout Your Project

In order to be able to save your changes to our repository, we have to checkout the project again from the server. This is a pretty simple step, but it is critical. Even though we imported our project already, that folder remains untouched. SVN will not know how to handle the updates unless we checkout the project from the svn server. So, we proceed to checkout by removing or renaming /var/www/htdocs/foo and typing svn checkout svn://localhost/var/svn/repo/foo /var/www/htdocs/foo. This will place some hidden folders in each of the directories in your project. These hidden folders keep track of your local changes so that svn will know how to merge your changes with those of other developers.

Update The Repository

When you've made some changes that you want to save, you can send them out by typing svn commit /var/www/htdocs/foo. This command knows how to get to the server to save the changes, since the project was checked out from the repository.

Update Your Files

Sometimes you'll need to get changes that other developers have made, or revert back to a version of your project that didn't have as many problems. You can do this using the svn update command. If you're already in /var/www/htdocs/foo, you can type svn update and it will update files that have been changed in the repository since you last updated. If you're looking to get an older version of your project, you can type svn update -r PREV or a revision number in place of PREV.

Remember, these are just the basics of using SVN. You can do many amazing things with this utility. I actually did all of this as I was writing this article, so I'm sure that it works.