That said, there have been boners, and not just in the recent past.
"RCS and SCCS" is 2/3 of a decent book. The coverage of _existing_
tools is good, but where the authors describe their own half-developed
TCCS, things waver, in large part because...it's vapor. Lesson: don't
let authors plug their own software until it actually exists in a usable
state.

I've seen and passed by more than one "Definitive Guide" book. There
have been several, and the topic and content have been so unmemorable
that I really can't bring them to mind, which describe themselves as
"the only reference you'll ever need for foo...", but *never tell me
what foo is*. I do much of my shopping in a bookstore, so I'll pick up
a title, scan the contents, scan the first chapter, check the index, and
check out some of the intervening material. If I don't know what the
book is about, I've got absolutely no interest in it, and it cheapens my
opinion of O'Reilly. At the hight of the boom it seemed that there were
more than a few such titles offered, a bad trand.

The "Building Linux Clusters" book was a disaster, as ORA has recognized
(the book was withdrawn from publication). I don't know what internal
training and development O'Reilly has, but this should be a flagship
case of what goes wrong: the book was poorly developed, poorly edited,
and ultimately withdrawn, a major embarassment. A hot topic, known
author, lots of diagrams, and CDROM are neither sufficient nor necessary
for a classic.

The "Learning Debian GNU/Linux" book was, while not anywhere near the
disaster of the Clusters book, not up to snuff. I'd compare it to the
complement of "UNIX Power Tools", "X User Tools". Maybe it's just
harder to write solid material about GUI apps, but the volume never
seemed to have the same heft as UPT. Similarly, McCarty's book focussed
too much on fluff -- applications, X Windows, Yet Another Bash Tutorial,
and the like. While there was an extensive installation guide (a
rapidly dated topic, addressed adequately in the Debian Projects own
documentation), there was insufficient coverage of principals and tools
behind the package management system, Debian's true strength and
distinguishing characteristic. I still recommend the equivalent Sams
title over O'Reilly for this niche (a very rare occurance). Lessons:
pick relevant, fundamental principles which help insure a long shelf
life. Defer to external references (including a projects own
documentation) where appropriate, or even your own. I strongly suspect
O'Reilly may have covered bash once or twice elsewhere.

I've picked up several of the Microsoft OS "In a Nutshell" books. I'm
less impressed by these than the equivalent GNU/Linux or Unix imprints.
Quite possibly due to my innate hostility to the subject. Also due in
no small part to the fact that legacy MS Windows *has* no unifying
philosophy as 'Nix does, making the task of ordering and presenting a
reference that much more difficult. But an effort to uncover
fundamentals and principles (even if they are fundamental flaws rather
than features) would help greatly.

Overall points? O'Reilly's greatest strengths are hard-hitting, dense
(but often entertaining) publications which get to the meat of a topic.
They focus on principles more than transitory aspects. They avoid
page-count inflating fluff (needless code listings, screenshots,
diagrams, bullet lists). Though the books inform, they also serve a
bootstrap purpose: by illuminating principles, and additional sources
of information, they far outlive the typical technical publication. The
contents provide a clear roadmap of the work, and the index is
comprehensive and intelligent. The physical production is good: paper
doesn't crumble or yellow, bindings last, printing is crisp. There have
been some lesser titles over the years, but the ratio to all books
produced hasn't markedly increased. More imprints in recent years has
brought more duds, but this isn't unexpected.

And while there have been duds, overall quality remains high. Remember
that it's a process, not a product. O'Reilly's forrays onto the web and
into conference space have been both well executed and in line with the
company's vision and profile. Overall: no sharks in these waters, but
keep a watch out for them.

Karsten M. Self <kmself@ix.netcom.com>