This file provides a quick tutorial which shows the basic functionality of src2pkg, and methods for making excellent packages -even if you know nothing about compiling software, packaging 'standards' or scripting.

src2pkg can be used in one of two ways -either as a command-line only tool, or to create and run 'build scripts' for each package you want to create. Since the use of src2pkg build scripts is a little more in-depth, we'll first discuss using src2pkg as a command-line-only tool.

For this Tutorial, it is convenient to use an example source archive -something which should compile on any system, that doesn't take long to compile, doesn't create many files in the package, and uses the most common method of configuration and compiling. This is summed up by the oft-repeated 'recipe' of unpacking sources and then running './configure', 'make' and 'make install' in the toplevel of the sources. Such an example can be found here:


The program 'di' is a small command-line utility for showing information about the disks and partition installed on your system and simply shows an output similar to that of the 'fdisk -l' command. If you already have some small source code you'd like to use as an example, use that instead and substitute any references below to the NAME 'di' with the name of your chosen program, and any references to the VERSION '3.11' with the version number of your chosen program.

Note: By default, src2pkg unpacks sources and creates a package tree as sub-directories under /tmp. There are good reasons for doing, so, but it makes it a little harder to keep an eye on what is happening in the sources and to inspect the contents of the finished package tree. For that reason, I suggest using the src2pkg option -CWD so that these dirs get created in the current working directory where you run src2pkg from. That way you can easily check the final package content, or look in the sources without having to navigate all over the system. If you want src2pkg to always create these directories in the current working directory, open the file /etc/src2pkg/src2pkg.conf with your favorite text editor, scroll down to lines ~203-207 and uncomment them -that is remove the '#' character at the beginning of each line. If you do this, then you don't need to use the -CWD option as shown below.

Command-line usage of src2pkg

Step 1

Create a new empty directory anywhere on your system where you have read and write permissions. This can be in your HOME directory or nearly anywhere else except you shouldn't do this inside regular 'system' directories. src2pkg will refuse to run if you are trying to use it from some place it considers 'wrong' or dangerous.

In the case of this example we'll work in our $HOME directory, so create the directory 'di' with such a command:

mkdir -p $HOME/di

Step 2

Download the di-3.11.tar.bz2 source archive from the link above and copy it into the directory $HOME/di. (Or use the source code archive of your choice)

cd into the subdirectory where the sources are located:

cd $HOME/di

or navigate to that location with your file manager if using a GUI inetrface. If using a GUI, open a terminal-emulator program there -usually this is done by right clicking on the filer-manager window and choosing an option named something like 'Open a terminal here' or 'xterm here'.

Step 3

Create your first package by running the command (from the terminal):

src2pkg -CWD di-3.11.tar.bz2

After a short time you should have a Slackware-type package created there, named di-3.11-ARCH-1.tgz, where ARCH is your machine's architecture. src2pkg will show the full path to the package when it finishes.

If you want to create some other type of package -for example a debian *.deb package, then tell src2pkg this by adding this to the command-line: -DEB

src2pkg -CWD -DEB di-3.11.tar.bz2

This will re-initiate the configuration and compilation of the sources and now create a package named: di_3.11-1_ARCH.deb

For other package types and the options needed to create them, see the 'Technical' file in the documents, or run the command:

src2pkg -hh

which shows the advanced command-line options for src2pkg.

Step 4

Use the SHIFT + PGUP/PGDN keys to scroll through the output from src2pkg. It is colorized by default, which should make it easier to read. Find the line which reads:

“Configuring sources using:”

followed by a line which shows the configuration options used to compile the program. Notice that the –prefix option has been automatically filled-in for you. By default, src2pkg will use –prefix=/usr as we believe that 'best practices' almost always install *packaged* software under /usr. But, you can change this by passing the -p=?? or –prefix option to src2pkg:

src2pkg -CWD -p='/usr/local' di-3.11.tar.bz2

When finished, again scroll through the output from src2pkg to see how the prefix option has changed.

By far, the most commonly used command-line option with src2pkg is the one used to pass extra options to the configure script. This is usually something like '–disable-something' or similar. Unfortunately, our example program 'di' does not need or accept any such options, but if it did, then you could pass them like this:

src2pkg -CWD -p='/usr/local' -e='–disable-cruft' di-3.11.tar.bz2

Notice that the option for the 'prefix' is kept separate from the other configuration options by src2pkg. This is done to simplify things -normally you shouldn't change the default prefix location of '/usr' except in special cases.

Step 5 (optional)

If you'd like to see how src2pkg can be used to directly download sources and compile them for you, follow this step.

Remove the example source archive we have been using. Make sure you are connected to the internet. Run src2pkg with the following command:

src2pkg -CWD http://distro.ibiblio.org/pub/linux/distributions/amigolinux/examples/di-3.11/di-3.11.tar.bz2>

This will download the same archive from the same location, or you can substitute the URL to any archive you wish to build. Once downloaded, the file remains unless you remove it -they are not downloaded each time -even if you run src2pkg using the full URL as above. It will look to see if the file is already present before trying to download it.

Using src2pkg build scripts

src2pkg started out as a script-driven API, and still many of the advanced features or options of src2pkg are only available by using a src2pkg build script. src2pkg can generate these build scripts for you, so you don't have to be bothered with knowing all the API syntax or with running a scripts many times just to debug them.

There are three basic command-line options specifically for use with src2pkg build scripts:

1. When run with the '-N' option, src2pkg will simply produce a basic build script with the name NAME.src2pkg.auto, where NAME is the name of the software package to be built. The command: src2pkg -CWD -N di-3.11.tar.bz2 will produce a src2pkg build script named 'di.src2pkg.auto' (see explanation of '*.src2pkg.auto' files below) It will also produce a generic description file for the package, name 'new.slack-desc' (see explanation of 'new.*' files below).

2. When run with the '-A' option, src2pkg tries to compile the sources and build the package first -and then writes the build script just before finishing the package. In some cases this will produce a more complete build script than using the '-N' option because src2pkg has a chance to find out more about the sources since it de-compresses and (tries to) build them. But, no script will be written if the build fails. Actually, you can use the '-A' option after starting with the '-N' option and src2pkg will update the script if needed. Using the -A (AUTO_SCRIPT) option also turns on various internal functions which create any extra files needed for the package, and then copies these files to the current directory -with the same name as the 'real' file inside the package, but with a 'new.' prefix.

<blockquote>* Note: Explanation of 'new.*' and '*.src2pkg.auto' files

If you repeatedly run src2pkg using the -N and/or -A options, files with the 'new.' prefix and/or src2pkg build scripts with the name: NAME.src2pkg.auto will be generated each time, overwriting existing ones. The 'new.*' files have the same name as the file which was included inside the package -except for the 'new.' prefix. The -A option causes these files to be copied to the current directory so you can edit them if needed or wanted. But, to make them be used and to be permanent you must change the name of the file to remove the 'new.' prefix. This tells src2pkg to use that file without changes and to not generate a new one. The main 'new.*' files which may be created by the -N option are the new.slack-desc and new.doinst.sh files (when creating Slackware-type packages).

Similarly, the file NAME.src2pkg.auto will be re-created each time until you change the name of it to drop the '.auto' suffix. Again, this tells src2pkg to not 'clobber' it any more and not change it. Be sure to change the name before you start editing it if needed or wanted. Still, remember that src2pkg will *edit* the file, if needed, if you first generated it with -N and then have src2pkg execute it with the -A option added.</blockquote>

3. Once you have a build script, you can easily execute it as many times as needed with the simple command: src2pkg -CWD -X

Using the '-X' option causes src2pkg to search the current directory and execute the first *.src2pkg or *.src2pkg.auto script it finds. This makes it handy to use without having to always type in the long name of the source archive, and you can still use most of the src2pkg command-line options in combination with '-X' -especially the options which use CAPITAL letters. For instance, the command: src2pkg -CWD -X -DEB -W tells src2pkg to run the first src2pkg script found in the current directory, to create a deban-type *.deb package, and to remove the temporary build files when finished( -W stands for Wipe Up).

You can also use src2pkg to create more complex build scripts which include other options. For instance, if you want to build a source archive which needs to have extra options passed to the configure script, you can pass the options on the command-line and src2pkg will translate them into its' own internal script syntax: You can demonstrate this with the following command:

src2pkg -N -e='–disable-cruft' di-3.11.tar.bz2

Afterward, open the file di.src2pkg.auto and you'll see this line near the top:


All the lowercase options available from the command-line can be easily written into your scripts this way. This makes it really easy to keep a record of what options were used to create the package. And the 'record' is also a brief 'recipe' for re-creating the package as many times as needed or wanted, simply by running src2pkg -CWD -X

Even though src2pkg provides a large number of command-line options, they are not nearly all the options available. Any of the other internal options can be easily used in your scripts. The syntax is similar to that of EXTRA_CONFIGS. That is, most internal options or features use CAPITAL letters which describe what the option does. There are many such options available -so many that I won't try to outline them in this short tutorial. You can find out about many of them by looking through the file /usr/libexec/src2pkg/DEFINES. This file is used to set up default behaviours when they are not already given elsewhere (as in your *.src2pkg script). It is worth mentioning, that all of these options can also be passed to src2pkg as 'environmental variables':

EXTRA_CONFIGS=”–disable-cruft” src2pkg -e='–disable-cruft' di-3.11.tar.bz2

will do the same as this:

src2pkg -e='–disable-cruft' di-3.11.tar.bz2

This means that it is easy to get src2pkg to 'play nice' with other scripts. For instance a very short script can be written which will build a whole 'tree' full of subdirectories with sources and build scripts with code like this:

for dir in $(ls -1) , do

cd $dir && src2pkg -X cd ..


Even with all the things src2pkg 'knows' how to do, there will always be cases where src2pkg doesn't know how to build certain sources, or you simply want it to do something different than what it 'knows' to do – maybe you want to add extra files to the package, remove files or even build extra sources and install them in the same package 'tree'. For these cases, src2pkg still provides an easy and reliable method for creating such build scripts. Nearly all the build 'instructions' which are part of the src2pkg API can be skipped by simply commenting them out in the script. The only one you should never skip is the first step 'pre_process'.

To get a better idea of what we are talking about, open the example script which you created above using src2pkg – or create one now with this command:

src2pkg -N di-3.11.tar.bz2

Any basic src2pkg such as the script di.src2pkg.auto has only about 50 lines and half of those are comments. The format is very clean and simple so that you can easily see, at a glance, any changes that you make to the script. At first, the only thing different between this example script and any other is in the lines which begin with SOURCE_NAME, NAME and VERSION. Lines which begin with the '#' character are comments and are ignored when the script is run. There are a couple of other lines near the top which are commented and contain defaults – like # BUILD=1 and # PRE_FIX=/usr. Scroll down further to see the real 'meat' of the script. At about line 20 there is an essential line like this:

. /usr/libexec/src2pkg/FUNCTIONS

This line 'sources' or 'reads in' the main code of src2pkg. In effect, it is just like copying and pasting all the src2pkg code into your script. But, since there are over 10,000 lines of that code, I think you'll agree that 'sourcing' it makes for a much more readable script! Just after that line you'll see this:

# Execute the named packaging steps:

What follows is the list of 16 instructions which are the basis of the src2pkg 'API'. Each instruction is actually a shell 'function'. A function is simply a block of code with a name. When the file /usr/libexec/src2pkg/FUNCTIONS gets 'sourced', all the functions are read into memory, but are only executed when they are called by name -as in the list inside the script.

Customizing a script is really simple. If one of the functions doesn't do what you need it to, you simply comment it out by placing a '#' character at the beginning of the line. And, if you need to carry out extra steps, like adding an extra file to the package, you simply insert a line or two of code between the regular functions at the correct spot. src2pkg makes this so easy that even people who don't know a thing about coding bash scripts cane a easily learn to customize the scripts. The syntax used is nearly exactly the same as what you'd use to perform the operation from the command-line – following what you already know how to do, or following the directions someone might give you on a forum, for instance.

Internally, src2pkg uses variables for all locations, so you don't even have to code (or know) where the sources have been unpacked or where the package 'tree' is. Let's clarify these terms a little. Normally, there are really only three locations you need to be aware of.

1. The first is the 'current working directory' – that is the location where you are runing src2pkg from -the same place where the NAME.src2pkg script is located. This is abbreviated in the scripts like this: “$CWD”

2. The second location to be aware of is the location where the sources have been unpacked before configuring and compiling them. This is called “$SRC_DIR” which means 'source directory'. The actual location where sources get unpacked is configurable by editing the file /etc/src2pkg/src2pkg.conf. By default, sources get unpacked under the /tmp directory, but $SRC_DIR includes the full path to the sources for *just these sources*. So, in the case of our example, if you write this line in your script (anywhere below the pre_process function): echo “Sources are in $SRC_DIR” when the script is run it will show this: “Sources are in /tmp/di-3.11-src-1” (or “Sources are in /path/to/SRC_DIR”). The name is composed of the main path /tmp (or CWD), plus the NAME and VERSION of the sources. The 'src' part is to distinguish the name from any other directories found there -including the package tree. The last part '1' is the BUILD (release) number for the package.

3. The third location to be aware of is the 'package tree' which has been mentioned several times. This is a temporary location where the various files which will be included in the final package get assembled. Once all the package content is in the 'package tree', the contents are compressed into the final installable package. Again, the package tree is represented by a variable, called $PKG_DIR. And again this location can be configured.

But from within your scripts you never have to refer to the absolute (real) location of these directories -you simply use the variable names and bash substitutes the real location -wherever that is at the time. This is handy because it means you don't have to edit your script in order to use it from anywhere on your system – even after changing the default locations where sources get unpacked or packages get assembled. It also means you can share your scripts with others and they will also be able to use them without having to edit all the paths.

Several example build scripts are included in the src2pkg documentation. Study them to get more ideas of how to accomplish any task needed for whatever package you want to build. Some of the examples are quite complex, but the README.examples file in the examples directory contains some notes about each example, making to easier to find any specific information you might be needing.

tutorial.txt · Last modified: 2012/03/31 10:20 (external edit)
Recent changes RSS feed Creative Commons License Donate Powered by PHP Valid XHTML 1.0 Valid CSS Driven by DokuWiki