[linux-elitists] Open-sourcing a book

Derek Vadala derek@cynicism.com
Tue Mar 27 13:02:59 PST 2001

On Fri, 23 Mar 2001, Jonathan Corbet wrote:

> So, do we...
> - make a big errata/patch file for the third edition?

I suspect that when I'm done writing, I'll probably not want to look at
the book, or RAID for that matter, for at least a few months. And I
imagine that this would be the easiest solution during that
period. Fortunately taking this route requires a lot less effort than the
alternatives-- maintaining a current, patched version of the book
online. It seems that for the initial release period, an online errata
file and a diff for the online documents makes the most sense. Not only
because it gives the authors a break for a little bit, but also because it
helps keep the print version useful for at least a little while. I think
that this method could quickly grow out of control.

> - keep the online version current (probably rewriting the "patches" to
>   fit the rest of the work), quickly obsoleting the printed version?

This really the toughest question. I think it's essential to keep the
online version current, otherwise the point of releasing the material
under an open-source is mostly wasted. If the book splinters into a dozen
different version because of neglect by the author or the publisher, then
we're left with the same problem that usually exists for any technical
work: disparite and unmaintained versions scattered about the Net, causing
confusion among users. I've encountered this to no end when dealing with
RAID. There's a lot of material out there, but figuring out what's up to
date and what isn't can be extremely tiring. This doesn't only apply to
the Linux RAID aspects of the book but spills over into discussions of
things like SCSI and ATA as well. Disk protocols are a great example of
erroroneous documetation, since every manufacturer seems to have its own
take on their evolution, nomenclature and peformance.

When I started the book, I didn't really expect to be talking about things
like SCSI and ATA/IDE as much as will end up in the final version. In the
last week alone, the _architecture_ chapter has doubled in size, mostly
resulting from the discussion of these protocols. And, a large portion of
that discussion is spent debunking the hundreds of differeing discussions
of each of them.

I'm digressing. The NAG is a good example of a well-maintained book that
was released open-soure, and I hope that I can follow in their footsteps
when I'm done.

> - call the printed version perfect and forward the patches to /dev/null?
> The point, of course, is that we *would* kind of like it if some people
> bought the printed version, even though we want to make a substantial
> contribution to the free documentation base.

Ditto. I don't want to sound petty, but I hope to at least make some money
off of this. But, at the same time, if finance was my motivation, I never
would agreed to write this book in the first place. In fact, my book,
originally started as a primer on using Linux in high-performance
appliance situations: firewalls, IDS devices, storage systems,
etc. O'Reilly, and my editor, Andy Oram (also editor of the NAG) quickly
realized that a book on each of these topics would eventually be
worthwhile. I think that an ipchains/ipfilter book is in the works.

Simply ignoring the material after it is printed isn't an option either,
for reasons stated above: part of my motivation was trying to alleviate
the headache that I, and others, have had with sifting through outdated
information on this topic.

The beauty of this is that I don't need to be the maintainer of the
patched, online version either. 

As for licenses, thanks to all who have offered some suggestions thus
far. I think that the final license choice will probably be left up to
O'Reilly, especially since, for this project, it wasn't something we
discussed up front. But, I suspect that any of the licenses discussed
here, or used by O'Reilly in the past will work out for most readers.

Derek Vadala, derek@cynicism.com, http://www.cynicism.com/~derek

More information about the linux-elitists mailing list