> I rather admire Cygnus' FSB model: Cygnus provides a technology service
> for customers who need some useful thing BAR and are willing to pay for
> its timely implementation.  Yet once BAR has been implemented, it is
> published for the benefit of everyone.

This is true.  Our customers tend to agree to this because, as we
explain, they are getting the benefit from all the previous customers
who did the same thing.  And, if they continue to use the software as
it evolves, they also benefit from the work done for new customers.
This is a much better deal than an ordinary consulting contract to
implement a particular feature.

By the way, that's only half our model: the other half is straight
support (bug fixes and periodic upgrade releases).

> 					   [I could be completely wrong
> here; Cygnus might be making BAR's and withholding their publication,
> but I'm hoping they're not.  [I hope no one at Cygnus takes offense at
> this comment.]]

I'm not offended.

Cygnus has at various times made BAR's and withheld them until the
sponsoring organization was ready to release them.  This typically
involves a new piece of hardware that is still not publicly announced.
We write these contracts (when we get it right) such that we are free
to distribute the software when the secret is out (at hardware
announcement) or in N years, whichever comes first.  The sponsor
wins because good free software is available for their product when
it first ships to customers; free software wins for the same reason;
and we get paid to do the work and to support it afterward.

I believe that Richard Kenner did a similar deal to port GCC to the
Alpha architecture.  We were under non-disclosure on Solaris-2 while
we were doing the GCC/GDB port to it, also, though it was publicly
known that we were doing the port.  Our port shipped within days of
the release of Solaris-2, and beat the first commercial compiler
(SunPro's) to market by something like a month.

	John Gilmore
	Cygnus Support