Introduction to Building Open Source Software

Title	Introduction to building open source software
Overview	In the modern UNIX/Linux/BSD era, the most common way to build and install open source software is with the use of "configure" and "make" tools. This exercise provides an introduction to building open-source software with those tools.
Prerequisites	Students should be comfortable with: uncompressing source "tarballs", using the Linux commandline, using "su" (or "sudo") to gain root privileges, installing software packages with the native package manager.
Learning Objectives	After successfully completing this activity, the learner should be able to: Configure, build, and install software using the "configure" and "make" tools.
Process Skills Practiced

Background

• https://www.gnu.org/software/make/manual/html_node/Introduction.html
• http://airs.com/ian/configure/configure_1.html
• As an alternative to the below curl example, the htop project has one of the best build documentation files out there, and it is a relatively simple project.

Directions

First, let's get started by opening a terminal window. You should be logged in as your normal (non-root) user. If you are not in your home directory, go there by typing:

        cd ~
 

On Fedora, in GNOME, it will look something like this:

Preparing the system

Now, we need to make sure we have the software installed that will let us download, build, and install new software from source code. These tools are:

• wget - a command-line tool to download files from the internet
• libtool - a utility that helps to generate shared library files
• make - a utility that parses Makefiles and starts the build
• gcc - the GNU compiler collection
• tar - a program used to make and unpack archive files
• sed - a stream text editor utility that curl uses to configure
• grep - a pattern matching utility that curl uses to configure
• zlib - a compression library that curl detects and uses if present

Most (if not all) of these items should already be installed on your Linux system, but we'll make sure. In order to install software packages on Linux, you will need to use root privileges. You can either use "su" or "sudo" for this. Sudo is easier, but it requires that your user be configured in /etc/sudoers first. Ask your instructor if this is the case. Su works as long as you know the root password (it will ask you for it).

su method:

       su -c "yum install wget libtool gcc-c++ tar sed grep zlib-devel make -y"
 

sudo method:

        sudo yum install wget libtool gcc-c++ tar sed grep zlib-devel make -y
 

Now that we have these software tools installed, we will download the source code (curl) that we're going to be using today. At the command-line of your terminal, run:

        wget http://curl.haxx.se/download/curl-7.43.0.tar.bz2
 

This will download "curl-7.43.0.tar.bz2" into your home directory. Next, unpack this source archive (also called a "tarball") using the tar command:

        tar xvf curl-7.43.0.tar.bz2
 

The options to the tar command are:

• x - eXtract
• v - Verbose (show us what is happening)
• f - File (we're working with a file. tar started life a long time ago writing to tape drives for backups, so we have to tell it we're working on modern files instead)

You should see a lot of files scroll by, and end up with a "curl-7.43.0" directory. Go into that directory:

        cd curl-7.43.0
 

Answer these questions:

1. What did the `yum` command do?
2. Why did we need to use `su` or `sudo` to install packages?
3. What is a "tarball"? How can we tell if a file is a "tarball"?

Configuring and Building Source Code

Look at the files in the curl-7.43.0 directory with the `ls` command. This is the top-level of the source code tree for curl. There are a few files you should know about:

• configure - This is the script that you will run to configure the curl source code to be built on your Linux system. Sometimes, configure needs to be generated from configure.ac (using a tool called `autoconf`), but this curl source tree already has a valid `configure` script.
• configure.ac - This is the raw source file that is used to make the `configure` script. You will not need to touch it now, but you should open it in a text editor and compare it to the contents of `configure`.
• Makefile - This is the file that tells `make` how to build the code. It is generated from Makefile.in by the `configure` script. Curl comes with a preconfigured `Makefile`, but `configure` will replace it.
• Makefile.in - This is the source file that is used to make the `Makefile`. Open it in a text editor and compare it to the contents of `Makefile`.

TODO -finish

Answer these questions:

1. What utility generated `Makefile.in`?

Installing the code

TODO

Answer these questions:

1. TODO

Deliverables:

1. Answers and explanations for the questions in each section.
2. curl should successfully be installed in /opt/curl

Assessment:

Students can be graded on understanding the build & install process, as well as successful build and installation of curl from source code.

Criteria	Level 1 (fail)	Level 2 (pass)	Level 3 (good)	Level 4 (exceptional)
Understanding the build & install process	0-3 correct answers	3-6 correct answers	6-9 answers correct	All answers correct, with valid reasoning.
Successful build and installation of curl from source code	Code not built or installed	Student attempted to build, but could not install	Student built code, but did not install correctly into /opt/curl	Code is built and installed correctly into /opt/curl

Comments:

What should the instructor know before using this activity?

A basic familiarity with autotools generated configure and Makefiles is very helpful, though, not necessarily required.

Suggestions for the Open Source Project

What documentation do you provide on how to build your code? Is it current and accurate? Do you discuss the various configuration options and what effect they have on the final build?