Elliote Harold wrote an article about Fuzz testing which was published on IBM’s developerWorks web site. In it he gives some great reasons for choosing to use an XML format for your data files instead of using a text-based format (like CVS) or a binary format (like serialized objects).

The important point was that XML parser are designed to separate good input from bad. Why redo all that error-checking just to create your own format?

In Elliote’s own words:

The key characteristic that makes XML formats resistant to fuzz is that an XML parser assumes nothing about the input. This is precisely what you want in a robust file format. XML parsers are designed so that any input (well-formed or not, valid or not) is handled in a defined way. An XML parser can process any byte stream. If your data first passes through an XML parser, all you need to be ready for is whatever the parser can give you. For instance, you don't need to check whether the data contains null characters because an XML parser will never pass you a null. If the XML parser sees a null character in its input, it throws an exception and stops processing. You still need to handle this exception of course, but writing a catch block to handle a detected error is much simpler than writing code to detect all possible errors. For further security you can validate your document with a DTD and/or a schema. This checks not only that the XML is well-formed, but that it's at least close to what you expect. Validation will rarely tell you everything you need to know about a document, but it makes it easy to write a lot of simple checks. With XML, it is very straightforward to strictly limit the documents you accept to the formats you know how to handle.

Section 3.2 of the Linux From Scratch bookmentions 57(!) packages and 39 patches that need to be downloaded in order to create your own Linux. I’ve downloaded all of the packages and patches and combined them into two TAR files to save you time. The files are located at:

http://www.dbbits.com/LinuxFromScratchPackages.tar http://www.dbbits.com/LinuxFromScratchPatches.tar

You can grab these files with these commands:

wget http://www.dbbits.com/LinuxFromScratchPackages.tar
wget http://www.dbbits.com/LinuxFromScratchPatches.tar

And you can extract the files with this command:

tar -xvf LinuxFromScratchPackages.tar
tar -xvf LinuxFromScratchPatches.tar

Once the files are extracted, you will have a set of *.tar.gz and *.tar.bz2 files. The various package files can be uncompressed with these commands:

tar -zxvf packages/vim-7.0-lang.tar.gz
tar --bzip2 -xvf packages/binutils-2.16.1.tar.bz2