geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Jencks <david_jen...@yahoo.com>
Subject Re: SUN PROPRIETARY/CONFIDENTIAL code in myfaces
Date Sat, 04 Aug 2007 19:15:34 GMT

On Aug 4, 2007, at 4:23 AM, Sam Ruby wrote:

> On 8/4/07, David Jencks <david_jencks@yahoo.com> wrote:
>>
>> BTW, the theory under which we (geronimo) has been operating is that
>> the sun copyright and legal statements apply to the text
>> documentation in the schemas and that once that is removed the rest
>> forms a part of the javaee specifications that we have a license to
>> implement, so we can translate it by any means we want (such as
>> xmlbeans, jaxb, castor, etc etc) to produce source code or class
>> files or pretty much anything else.  I don't see how it's possible to
>> implement the specification without this: IMO without this
>> interpretation any javaee product must be cddl.
>
> I acknowledge that there was a time critical question in the portions
> that I snipped, but first I think that it is important that we come to
> a common understanding of what the problem is.  Given that there are
> lawyers on this list, I'm sure that somebody will point out the
> thousands of tiny mistakes that I'm about to make, but I'm confident
> that I have the broad brush strokes right, so here goes...
>
> In order for us to legally distribute some Work, we must comply with
> all the terms and conditions in the licenses that contribute to that
> work.  That's it.  End of sentence.
>
> Presuming that we do that, do we have the right to distribute code
> under the CDDL?  Yes, absolutely.  Are there any terms or conditions
> in CDDL that we would find overly burdensome to *us* (the ones
> releasing the software)?  Absolutely not.
>
> Furthermore, we even have the rights to distribute the version of XSDs
> that SUN PROPRIETARY/CONFIDENTIAL, even though Sun labeling it so
> brings into doubt what their true intentions were in licensing this
> materials, which makes our ability to demonstrate that we have
> complied with their intentions harder.  Note that I said harder, I
> didn't say impossible.  We have ample documentation to demonstrate
> that the ASF has the right to ship these XSDs, but who wants to have
> to go and explain all this time and time again, potentially to each
> and every new user of Geronimo?
>
> Back to CDDL.  I have no personal knowledge as to why Sun picked this
> particular license, but let's look at it in context.  Each of the XSDs
> in question represents a machine readable codification of a portion of
> a standard.  As a standard means that you and I do something the same
> way, any modification means that you and I are doing something
> different, so it isn't a standard.  So, effectively, we are taking
> these sources in and agreeing not to modify them, which makes them not
> open source.  If you think we have heartburn on CDDL, think about the
> idea of the ASF shipping code that contains portions that are not open
> source.
>
> But, we are not about to say that standards are not a good thing.  To
> the contrary.
>
> This is all absurd.  You can see the source.  You can change it, as
> long as you don't claim compatibility.  Now, with CDDL, that is
> explicit.  Yea!
>
> So, what's the problem here?
>
> The problem isn't with Sun.  The problem is with the ASF.  The ASF is
> about community (how we develop software) and license (what we permit
> users of our code to do).
>
> Our license is part of who we are.  Others may distribute things under
> different licenses, and that, in part, defines who they are.
>
> Our license intentionally allows users to modify, sublicense, and
> distribute our code.  All of it.  If people want to contribute back
> their changes to us, they can join our communities.  If people want to
> release their changes under their own license, they can do that too.
> If people want to retain their changes and only distribute binaries,
> that's OK too.
>
> Most of our code bases make it easy for our users.  Everything comes
> under one license.  A license that it relatively short, and well
> understood.
>
> Geronimo isn't one of those code bases.  It contains many parts from
> many sources.  In releasing Geronimo, we need to make sure that all of
> this is crystal clear.  The bulk is under the ASF license, and people
> are free to modify that bulk as they see fit.  Some portions are
> packaged with the distribution as a convenience (or in the case of
> these XSDs, as a necessity), but none of these subcomponents impose
> any additional restrictions on what you can do with the code that we
> produce, and all of it is clearly labeled.
>
> In particular, (and I may just be misreading your statement), it is
> NOT the case that "any javaee product must BE cddl", but rather "any
> javaee product must CONTAIN cddl" (actually, those files can be
> licensed under other licenses, but lets not digress).
>
> So... what is the ASF legal committee and the Geronimo PMC to do?
> Well, again, legally, Geronomo has the right to make releases as long
> as those releases comply with the appropriate licenses, so one could
> make the case that everything from that point on is up to the Geronimo
> PMC.  And, in fact, this stuff is complicated enough that how you make
> the determination as to what makes sense in any particular situation
> depends very much on the situation, so again, it is Geronimo's
> problem.
>
> On the other hand, given that this stuff is complicated, it makes
> sense for us to pool our knowledge.  Have a central place where
> projects can go to (and contribute back to) where general guidelines
> are captured and interesting special cases are referenced.
>
> Things like "yes, the license for foo.dtd requires people to provide
> source with any changes that they distribute, but project P only uses
> the dtd in a way that consumes the source itself directly at runtime,
> so that requirement doesn't apply in our situation", and "the license
> for bar.xsd requires that people provide source with any changes that
> they make, and we want to make this crystal clear to people.  Since
> bar.xsd doesn't change very often, we compile it into .class (or .jar)
> files, and check that into SVN, along with instructions on where to
> find the original sources".
>
> I've rambled for long enough now... so let me close with this: let's
> suppose somebody gets this wrong (it happens).  A bug report comes in.
>  Where do you think such a bug report would be routed?  To the board?
> To the legal committee?  To Geronimo?  If you guess the third, you
> would be right.
>
> How can I help?  Well, for starters, I don't want to spend any of my
> time answering any time critical hypotheticals.  Nor do I want
> somebody to back up a dump truck, and say "here's Geronimo, you figure
> it out".  But if there are specific questions that we can jointly work
> through, I am here to help.
>
> If this makes sense, we can go back to your specific question(s).  If
> not, let's see if we can come to a common understanding of the context
> before we proceed.
>
> Fair enough?
I hope so.
I think we've successfully derailed releasing any of the questionable  
geronimo artifacts, so except for the geronimo community wanting to  
get out our last years work the time critical element is gone.

As I see it there are two kinds of questions I'm asking:

1. Are the 6 questionable jars (4 I already mentioned plus a servlet  
spec jar with some retyped sun xsds and dtds) OK to release?   
Obviously the geronimo PMC thought so but this conversation has  
thrown that into doubt as far as I am concerned.  Is there some  
information you (or anyone else) would like in order to give an  
opinion?  I tried to explain the process used to generate these jars  
and the thinking behind the process already.  Note that none of these  
jars start from the cddl licensed sun schemas, they all start from or  
relate to the pre-cddl schemas.  I don't see these questions as being  
hypothetical, and I hope 6 jars isn't a dump truck.  The servlet spec  
jar under vote is at http://people.apache.org/~mcconne/geronimo- 
servlet_2.5_spec-1.1.tar.gz.  The vote passed but AFAICT it has not  
yet been called or the artifact actually released.

2. Hypothetically, starting from the cddl licensed schemas, what can  
we generate from them, what can we include in apache svn and  
releases, and what license is any of this under?  The geronimo pmc  
has previously thought that generated source was under asl.  Craig is  
claiming that generated source is cddl, however as I tried to explain  
this point of view seems to me to lead to the entire server being  
required to be cddl.  In other words I think either Craig is wrong or  
apache can't develop any javaee products.  In addition I think  
Craig's argument applied to the pre-cddl xsds would entirely prevent  
apache releasing any j2ee or javaee products whatsoever.

Following onto 2, sometimes there are mistakes in the sun schemas  
that, well, prevent using them directly in completely compliant  
implementations.  For instance the web-app-2.5.xsd had a incorrect  
regular expression for http-method.  Assuming we eventually do use  
the cddl licensed schemas, and these are in publicly accessible  
apache svn, can we fix these errors?

thanks
david jencks




Mime
View raw message