For Approval: The netX Public Lisense (in plain text)

[Wilson, Andrew]

Bruce Perens wrote:

> OK, I guess I'm wrong about sublicensing being required. But in that
> case we must make sure that the license allows the usual distribution
> chain without sublicensing.
>
> Anyway, there are other reasons to reject this license.

Agreed.  Others have noted the special treatment of kernel interfaces
which applies discriminatory terms to a particular class of technology.
And as you note, all changes to netX "kernel interfaces" must be submitted
to the initial developer.  Many have proposed, but OSI has never approved,
such mandatory feedback licenses.

Another question I would like to raise is the netX license's limitation of liability.
A netX Licensor's limitation of liability is miniscule (to damages from "intention and
gross negligence"), but there is a difference between miniscule and zero.
All the most common open source licenses come with
language in SCREAMING ALL CAPS which state that they come with
NO WARRANTY OR INDEMNIFICATION FOR ANY PURPOSE, etc., etc.
My concern -- and this is a question for the qualified attorneys on
this list -- would be whether even miniscule indemnification renders
netX incompatible with <insert your favorite permissive license>.

In concrete terms, I question whether you can combine netX with
BSD, Apache, or MIT-licensed code, since you receive no indemnification
at all from BSD/Apache/MIT licensors.

netX is GPL-incompatible for a host of reasons.
If it is also incompatible with permissive licenses, OSI should not approve a new
license which creates a class of one, IMO, absent some truly compelling reason.

Andy Wilson
Intel open source technology center