The first edition of this book was originally published in 1993 by John Wiley under the same name, Advanced Topics in UNIX. It was named an Alternate Main Selection of the Newbridge Book Club in the same year. That book stayed in print for thirteen years, an eternity in the computing literature. Due to changes in the publishing industry, this edition of the book is only available electronically.
Read alsoMany by Douglass and DuBois
This volume contains the complete text of several important works by Frederick Douglass and W. E. B. DuBois, two of the greatest minds of the American Negro community during the 19th century. (For quite some time the preferred term has been, of course, African-American.) Several of the most…
I was motivated to revise the book because of the increased popularity of several variants of UNIX and on what I learned from reviews of the previous published version of this book. Linux has become increasingly popular, due in no little part to it being so popular in the open source community and also because it is serving as the basis of the operating system for the Google Android phone. The Mach operating system, originally developed at Carnegie Mellon University, is the basis for the operating systems used to control Apple Macintosh computers. Solaris, originally developed by Sun Microsystems, is now considered by Oracle to be the top enterprise operating system and also claimed to have been especially built for cloud computing. It was clear to me that a revision of the book was necessary. I am currently running versions of both Linux (Ubuntu) and Solaris on my Windows PC. Of course, the operating system on my Macintosh is based on Mach.
My experience with multiple versions of UNIX-like operating systems showed me that end users, application programmers, system programmers, and system administrators often had difficulties in making programs and utilities work well across different UNIX variants, due to differences in file system organization, different locations of critical configuration files, and important, yet subtle, differences in how system calls operate. There are also issues with different utilities, many of which are either not available on all UNIX versions, or else require a substantial effort to even get them to install properly. One of the most interesting problems required detailed analysis of several Linux variants in order to get a single public domain application to work – the different Linux variants from Fedora (formerly Red Hat), SUSE, and Ubuntu were examined before the application would install and work properly.
I have chosen an approach guided by my own research and experience in the efficient development of large, high-quality, software systems in both UNIX and non-UNIX environments over much of the last twenty-five years. Much of my research in this area is based on the application of systematic approaches to software reuse as part of the software development process. In fact, I chose to produce a second edition of my book Software Reuse: Methods, Models, Costs before I began revising this book, Advanced Topics in UNIX.
What is the relevance of software reuse to a book on UNIX? A huge percentage of current software applications for UNIX and other operating systems are built using existing software components that are either partially or entirely reused. Older software components necessarily have been built using older technology. Hence, they are likely to use some of the older system calls as well as newly created calls. You clearly need to be able to understand the differences in different generations of system calls and their behavior if you are redeploying existing software components. That is, after all, the reason for learning system calls if you are an applications programmer. Of course, the same holds true for systems-level programming, especially kernel-level programming, since operating systems are rarely written from scratch.
Keep in mind that in the most extreme case of reusing software – the deployment of entire applications known as COTS – the internal structure of the application may not be known and all that is available to the typical software reuser is the published interface, which may include older system calls. You simply need to understand how these older system calls work, in order to be able to reuse the software and have it work with other, existing software. (The acronym COTS stands for Commercial, Off-The-Shelf Software; there are variants for government-produced software and military-produced software, with the obvious meanings.)
As a result of the importance of reusing software, especially in a UNIX environment, we will discuss both newer and older – what Marc Rochkind calls obsolete – system calls. The idea is for you to understand enough about system calls and what types of interactions can occur between them, to be able to, at the very least, write the kind of functions known as wrappers, glueware, or bridgeware. Here the terms wrappers, glueware, or bridgeware refer to functions coded with the sole purpose of allowing different program modules and system components to communicate, passing all required data as necessary.