www-legal-discuss mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jeffrey Thompson <jt...@us.ibm.com>
Subject Re: IBM and WS-Security
Date Thu, 23 Jun 2005 22:59:41 GMT
Davanum Srinivas <davanum@gmail.com> wrote on 06/23/2005 11:22:45 AM:

> Jeff,
> 
> Let's start fresh. Is IBM willing to allow Apache to write a
> WS-Security Implementation?

Certainly.  That was why we contributed it to OASIS and why we made an RF 
patent commitment.  As am aside, I'm not aware that we've identified any 
necessary patents for WS-Security. 

> 
> We already have code, and have participated in interops, but have not
> made a release yet (http://ws.apache.org/ws-fx/wss4j/). Current terms
> published by IBM (http://www.ibm.com/ibm/licensing/977Q/2112.shtml)
> are incompatible with Apache License/Process. 

I think we have a basic disagreement here.  IBM's patent terms are not 
incompatible with Apache's license or Apache's open source model.

>                                                I have detailed feedback
> almost clause-by-clause for the IBM License (and the Microsoft
> License), but that defeats the purpose...It's better for us to specify
> what we'd like to see in a modified license from IBM (and Microsoft).

Actually, I haven't seen that (I'm not always able to read all of the list 
postings).  No need to bore everyone else with repetition, but if you want 
to send it to me off list, I'd appreciate it.

> BenL and me were discussing this morning on what to request IBM and
> MSFT w.r.t WS-Security license. Here is the list:
> 
> 1. Users should be able to download our code and run it without further
> action on their part (obviously they should be aware of the licence and
> comply with it, but should not need to do anything beyond the normal
> requirements of the Apache Licence 2.0).

Obviously.  At that's what happens under the current licenses.  Under the 
doctrine of patent exhaustion, if a maker of a product (here, Apache) is 
licensed to practice a patent when it makes the product, users of that 
product are also covered.

> 
> 2. The licence should not require implementations to be compliant (as we
> agreed, this is an incomprehensible requirement anyway).

I'm not sure why you are singling out WS-Security on this point.  Every 
Java specification license includes compliance requirements and I expect 
that all of the patent licenses pursuant to the W3C Patent Policy are 
limited to "implementations of the recommendation".  Is your point that it 
is not clear what "compliant" means in the absence of a test suite, or 
that the concept of compliance is itself incomprehensible.  If its the 
former, take the normal meaning -- that the implementation actually 
implements the spec correctly.  If its the latter, I'm not sure how to 
respond.

> 
> 3. There should be no restrictions beyond those imposed by Apache
> Licence 2.0.

Again, why are we singling out WS-Security.  The Java spec licenses 
include additional restrictions as do pretty much all existing patent 
licenses for standards (W3C or otherwise).  The question is whether the 
additional restrictions impose an inappropriate burden on Apache or its 
licensees.  There was a germ of a discussion a few months ago (February 
05) on this list about what Standards are compatible with Apache's 
approach to life.  At that time, I wrote in response to a post by Larry 
Rosen:

--In any event, Apache guidelines would address, in my mind, at least 
three basic questions:
--
--1.  Can Apache get the standard?  If we can't get it, we can't implement 
it.
--
--2.  Can Apache publish its implementation under Apache's license?  This 
is 
--the most critical.  Any standards agreement that prevents open source 
--implementations shouldn't be embraced by Apache.
--
--3.  Is Apache opening itself or its customers to royalties for necessary 

--patents?  This is the hardest to answer.  Your definition of open 
standards 
--spent a lot of time talking about the details of the patent licenses, 
but in 
--the end, the question is whether the open source project and its 
customers 
--qualify for the free license. 

The third item is the most relevant to the current conversation.  I think 
that this list is still accurate, and as far as I can tell WS-Security 
meets those requirements. 

> 
> 4. Another instance of conflict with AL 2.0 is the requirement for
> compliance with U.S. Export laws - this needs to go.

I don't see an Export law term in the IBM license.

> 
> 5. Note that the Apache Licence
> (http://www.apache.org/licenses/LICENSE-2.0.html) has a clause relating
> to patents which may well work in the way you want already - clause 3.

The patent grant in the AL2 performs a completely different function than 
a patent grant that applies to implementations of a Spec and necessarily 
focuses on different issues.  When a company contributes CODE to Apache, 
knowing that the code will be licensed liberally to the world, it is 
important to know that that company isn't going to go around and sue the 
licensees of that code for patent infringement just for using that code. 
There are few restrictions on that patent license, but it is tied to the 
code.

For specification related patent licenses, there is no code, or at least 
not yet.  The license is necessarily focused on implementations of the 
Spec.  Anyone who writes code that implements that Spec is covered, unlike 
the license in AL2 which only covers licensees of the Apache code.  So, in 
some ways, the AL2 license is too broad (it covers the code, whatever it 
is used for), and in others, its too narrow (it doesn't cover non-Apache 
licensed code).

> 
> Thanks,
> Davanum Srinivas
> Vice President, Web services, Apache.
> 
> Thanks,
> dims

I think that this is an important issue for Apache, because it seems to me 
that if Apache applies the rules that it seems you are applying here, most 
(if not all) of the current projects will have problems.  In some sense, 
patent licenses that are tied to specifications are orthogonal to the AL2 
license.  I don't think you can force them to be parallel, and if you 
filter out all licenses which are not, you will likely end up with a null 
set of specifications to implement.

FYI, I'll be non-connected for most of tomorrow, but will try to respond 
to any comments when I get e-mail access.


Jeff

Staff Counsel, IBM Corporation  (914)766-1757  (tie)8-826  (fax) -8160
(notes) jthom@ibmus  (internet) jthom@us.ibm.com (home) jeff@beff.net
(web) http://www.beff.net/ 


Mime
View raw message