Books seem to have enjoyed a quiet resurgence lately. No, really. After paper-based communication had been pronounced dead, more books were published than ever before. Ironically, books on programming, security and web design, despite the major slump that followed the post-2000 downturn, seem to be more successful than ever.

On-demand printing, essay collections derived from electronic sources, even blogs in print make up a potent brew that seems to intoxicate the internet-weary. Electronic books, pronounced dead a few times, are slowly establishing themselves in the canonical way in which technical and scientific texts are distributed.


OS Books
Books about operating systems are a fairly recent phenomenon. The days when technical documentation consisted of Xeroxed piles of badly bound A4 paper are not so long ago. Indeed, operating systems as a separate software category did not enter public consciousness before the 1980s. The importance of operating systems became self-evident, of course, when mass-produced mini-computers and PCs appeared; they brought simple applications and programming tools within reach of middle class consumers.

Technically, most features of modern operating systems were tried, if not fully tested, after the s/360 mainframe architecture made an appearance in 1964. Naturally, not many operating system books even included a chapter on the s/360, since there were never more than a few thousand installations all over the globe.

BSD books
BSD Unix came into existence in the late 70s - but only after PCs became acceptable as Unix platforms in the mid 90s, did the stream of popular publications on operating systems begin. BSD Unix ran on PDPs as well as early VAX machines and therefore its relevance for the nascent PC market was not immediately apparent.

BSD Unix and the early BSD trinity versions (Net, Free and later Open) had not been documented in newer publications at all. Earlier BSD Unix releases had been written up fairly well, particularly since Marshall Kirk McCusick and Samuel Leffler had done a good Unix design book on BSD Unix 4.3 in 1991. There had been an earlier command reference for BSD Unix 4.2 and 4.3 in 1987, but naturally it was not much of a seller. O'Reilly took a stab at BSD in 1994, but the success of its Nutshell series, covering Unix in both System V and BSD variants eclipsed the BSD books, mainly because the BSD books' origins in technical documentation were too obvious.

For a long time, the reading audience was presented with Unix books that did not reveal too much of the rather substantial differences between the different kernels and shell implementations. Commercial Unices were accompanied by swathes of electronic documentation, which - and I tested this extensively in 1997 - contained many areas of documentation overlap between various Unix versions, although Unix command line interfaces were by no means assured of working in an identical fashion. It was a case of caveat emptor, to the frustration of system administrators.

A Little Earthquake
If we skip forward from 1994 to 2004, we find a much-changed landscape. Since 2001, a small cluster of BSD books (in English) has appeared. FreeBSD has been covered from quite a few angles; network programming on BSD Unix was already covered in Stevens' classic on Unix network programming, but FreeBSD got extensive coverage from a user's and a network administrator's perspective. OpenBSD attracted quite a bit of attention, and so far, 4 titles have appeared under several imprints. NetBSD is in the process of gaining some attention as well, and several titles are due to be published during the next year.

What has changed? First of all, one would have to explain the glacial slowness with which the phenomenon has occurred. All BSDs have excellent electronic documentation: man pages and the FreeBSD handbook with Greg Lehey at the helm have been justifiably famous for years. The latter has been available in print from the BSD Mall since the mid-1990s, but its very quality and universal availability on the web and in print deterred publishers from joining the fray. Of course O'Reilly now publishes the FreeBSD guide, which goes to show that further corporate attempts at taking over any of the BSDs are futile.

The difference between Linux and BSD publishing and documentation could not be greater: the Linux documentation project encompassed everything from Kernel documentation and basic introduction to networking to traffic shaping and bash shell prompt configuration. There was little or no coordination. Linux books tended to rely on the previous Unix or Nutshell template, an approach that didn't serve newbies well.

The landscape of Unix publishing had changed irrevocably after the dotcom boom collapsed. Proprietary Unices could not count on deep corporate pockets. Linux slowly gained acceptance in the boardroom. BSD-derivates had no corporate backers, except, rather surprisingly, Apple. The visionaries from Cupertino had a fairly strong interest in FreeBSD and now incorporate some of its codebase in OS X. In essence, what had appeared was a slow retreat from operating systems as a major battlefield between corporations: they became more configurable, and, thanks to an increasing number of books on free operating systems, more intelligible. Usability made great strides as well, with the appearance of desktops and a large number of window managers. Corporate lock-in was extremely unlikely and publishers realized that the fairly corporate-friendly license notwithstanding, the BSDs were here to stay.

One Unix, too Many
It also meant that one-size-fits-all publishing or the Unix in a Nutshell school of documentation died an ignominious death: Linux cannot be considered as an operating system exclusively - and regardless of early FUD and later Microsoft rearguard actions, its flavours became more and distinct. Although a layer of POSIX compliant and Unix-tested utilities were rewritten under the FSF imprimatur, larger tool chains, such as the installation and package administration, differed wildly from distro to distro. The BSDs represented full operating systems, and the common codebase had been maintained for decades. Although package management and installation widely differed as well, only OpenBSD acquired substantially different conventions, particularly with regard to file systems and networking utilities that required special treatment.
Thus FreeBSD and NetBSD had been covered in generic Unix books, but these books did not find a ready audience anymore. The printed and electronic/web publishing market presented a very different image: Unix no longer provided the common ground, but each and every Linux distribution and BSD flavour became a separate market for writers and editors. There was an implicit recognition that Unix and POSIX standards were just not sufficient as a starting point for the day-to-day user of KDE/GNOME/Windowmaker or database and office users. The common excuse for xBSD users to refer to standard Unix titles stopped working. Therefore the stable of BSD books published heretofore began looking anaemic, and a more realistic approach to publishing BSD literature was taken across the publishing industry.

BSD Admiration Codified
It is interesting to see that, as opposed to the Linux publishing mania that swept the technical bookshelves around 2000, books about the open source BSDs were written not so much to invite readers to use the operating system, but to deepen their already existing appreciation. Technically, xBSD internals are quite advanced these days, but applications and runtime characteristics are aspects about which even fairly experienced users can always learn something new. Of course, tuning an OS is something that presupposes a modicum of experience, and using applications that are unique to the BSD world takes more than just the ability to find the open-file dialogue in Abiword.

The obvious avenue was security: BSD is not a major target for virus writers and script kiddies, which enables security specialists to use BSD hosts as bastions. It appeared that relative obscurity translated to relative security.
For writers and publishers this meant that although security books require more investment than a book about Excel scripting, once the title had appeared and been written up on specialist websites, a small, but steady, income stream was assured.

In addition, the rather more conservative Unix system administrators' crowd was glad to read that underneath the hood, few radical changes had been made, (an impression that is actually quite false!), and administering a Unix system was still something that could be read about in books assuming a modicum of Unix expertise.

We Are Unix Brains
One has to realize that in the years before the Internet boom, access to information pertaining to information technology was presumed to be the province of scientists and programmers. Not that this judgment carries the force of 10 years of social science research, but books on Free Software were mostly confined to what amounted to rather amateurish or computer science-style publications. They did not appeal to a mass market by any means. Books about the BSDs carry a certain elitist cache to this day, a clear sign that someone has done more than run a few Linux installations and play around with Openoffice.

More seriously, though, the slow appearance of BSD books and BSD news carries within itself the realization that operating systems are not easily interchangeable, regardless of the standards they might adhere to. It does not really matter whether all command line utilities adhere to POSIX, or whether localisation is handled according to the latest i18n standards. Different operating systems are tuned to do different tasks. Many Unix ideas might have been preserved in the BSDs, but an OpenBSD book describes a very different set of scenarios from a FreeBSD title or - shock, horror - a Mac OS X book for Unix geeks. Books that work across BSD flavours, are unlikely to succeed, since hardly anyone uses several flavours at the same time.

So, are we going to see more books on the BSDs? I believe that we might see more BSD publishing ventures that emphasize its file management capabilities: books on filesystems, clustering, tuning for web farms, and system level programming. Technical documentation that is available right now might be made available in book form more often, but is likely to be a fairly limited exercise in opportunism.

In fact, unified "Unix" publishing is no more.

Frank Pohlmann

Refernces

Daemon news
www.daemonnews.org

FreeBSD
www.freebsd.org

OpenBSD
www.openbsd.org

OSNews
www.osnews.com