Our file system application implements a simple hierarchical file system, similar to the file systems we find in Windows or Unix. To keep code examples to manageable size, we ignore many aspects of a real file system, such as ownership, permissions, symbolic links, and a number of other features. However, we build enough functionality to illustrate how you could implement a fully-featured file system, and we pay attention to things such as performance and scalability. In this way, we can create an application that presents us with real-world complexity without getting buried in large amounts of code.
Our file system consists of directories and files. Directories are containers that can contain either directories or files, meaning that the file system is hierarchical. A dedicated directory is at the root of the file system. Each directory and file has a name. Files and directories with a common parent directory must have different names (but files and directories with different parent directories can have the same name). In other words, directories form a naming scope, and entries with a single directory must have unique names. Directories allow you to list their contents.
For now, we do not have a concept of pathnames, or the creation and destruction of files and directories. Instead, the server provides a fixed number of directories and files. (We will address the creation and destruction of files and directories in
Chapter 34.)
Files can be read and written but, for now, reading and writing always replace the entire contents of a file; it is impossible to read or write only parts of a file.
