incubator-ooo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dennis E. Hamilton" <>
Subject OT: Generated File IP
Date Thu, 29 Sep 2011 13:58:28 GMT
I WENT TOO FAR.  (I am changing the thread because a lot of this is not relevant to IP review.
I think the conclusion C is all that matters with regard to copyright and copyright-based

My comment about binaries is completely off the mark.  Ever since software was made copyrightable
subject matter by statute (i.e., the 1976 Copyright Act), there have been special provisions
concerning software with regard to the exclusive rights that authors enjoy. First, copyright
does extend to executables and binaries, even though they are not meant to be readable by
humans, and some of the provisions limit the exclusive rights.  (E.g., backups are permitted,
and there is business about the ephemeral copy that exists in order to execute software.)
 I don't know that this applies to any software for which there is no source code, but that
need not bother us here.

In the exchange with Norbert, there remains the question of license conditions that originate
exclusively through the exclusive rights of copyright holders and the license conditions that
may arise in some other manner.

A.  Exercise and Licensing of Exclusive Rights under Copyright
 1. Let us assume that there is some construction by which a source program, which is exclusively
my original authorship and bearing my copyright and license notice is processed by something
that introduces material that is not subject to my copyright but is subject to the copyright
of some other.  Let us also assume, that, somehow, the processor that spits out this derivative
affixes notices on behalf of *its* authors.  (This would be like a preprocessor adding notices
and license statements.)  So the intermediate work and any further conversion of that into
a delivered software program (e.g., directly or in binary form) has this mixed licensing situation,
and presumably the terms of all of those licenses have to be satisfied.  Let's just accept

 2. The next question, in the case of my work being open-source, is whether or not the various
licenses are all qualified and satisfiable in accord with the Open Source Definition.  

 3. The next question for us is whether or not the various licenses are compatible with the
ALv2 (and for the latest releases of so far, LGPL3).  
    Note that for input to Bison, the license is whatever the author of the input says it
    For output from Bison, the potential of additional license conditions and possible conflict
is determined by notices that are included in the output from Bison, the parser implementation
One has to know that by examining the file for notices and "special exception" information.
    I assume that Berkeley YACC doesn't have these problems, and as long as no special features
Bison are involved, Berkeley YACC can be substituted.

B. Another wrinkle
I don't think this is applicable to any situation in hand, but one problem with (A.1) is that
it can't have the effect of achieving a patent by use of copyright.  (There are other US-specific
pre-emptions too, but I doubt that there is a free-speech issue at hand.)

C. What to do if the copyright and license conditions injected into the intermediate product
unacceptable?  Don't worry about it being a valid declaration.  Get rid of the dependency.

 - Dennis

-----Original Message-----
From: Rob Weir [] 
Sent: Thursday, September 29, 2011 05:00
Subject: Re: A systematic approach to IP review?

On Thu, Sep 29, 2011 at 1:53 AM, Dennis E. Hamilton
<> wrote:
> Let me recall the bidding a little here.  What I said was
> " It is unlikely that machine-generated files of any kind are copyrightable subject matter."
> You point out that computer-generated files might incorporate copyrightable subject matter.
 I hadn't considered a hybrid case where copyrightable subject matter would subsist in such
a work, and I have no idea how and to what extend the output qualifies as a work of authorship,
but it is certainly a case to be reckoned with.
> Then there is the issue of macro expansion, template parameter substitution, etc., and
the cases becomes blurrier and blurrier.  For example, if I wrote a program and then put it
through the C Language pre-processor, in how much of the expanded result does the copyright
declared on the original subsist?  (I am willing to concede, for purposes of argument, that
the second is a derivative work of the former, even though the derivation occurred dynamically.)
> I fancy this example because it is commonplace that the pre-processor incorporated files
that have their own copyright and license notices too.  Also, the original might include macro
calls, with
> parameters using macros defined in one or more of those incorporated files.

Under US law:  "Copyright protection subsists, in accordance with this
title, in original works of authorship fixed in any tangible medium of
expression, now known or later developed, from which they can be
perceived, reproduced, or otherwise communicated, either directly or
with the aid of a machine or device"

IANAL, but I believe Dennis is correct that a machine cannot be an
author, in terms of copyright.  But the author of that program might.
It comes down to who exactly put the work into a "fixed in any
tangible medium of expression".

When I used a n ordinary code editor, the machine acts as a tool that
I use to create an original work. It is a tool, like a paintbrush.  In
other cases, a tool can be used to transform a work.

If there is an original work in fixed form that I transform, then I
may have copyright interest in the transformed work. That is how
copyright law protects software binaries as well as source code.

As for the GNU Bison example, if I created the BNF, then I have
copyright interest in the generated code.  That does not mean that I
have exclusive ownership of all the generated code.  It might be a
mashup of original template code from the Bison authors, along with
code that is a transformation of my original grammar definition.  It
isn't an either/or situation.  A work can have mixed authorship.


> I concede that copyrightable matter can survive into a machine-generated file.  And I
maintain that there can be other conditions on the use of such a file other than by virtue
of it containing portions in which copyright subsists.  For example, I don't think the Copyright
office is going to accept registration of compiled binaries any time soon, even though there
may be conditions on the license of the source code that carries over onto those binaries.
> And, yes, it is murky all the way down.
>  - Dennis
> -----Original Message-----
> From: Dennis E. Hamilton []
> Sent: Wednesday, September 28, 2011 22:32
> To: ''
> Subject: RE: A systematic approach to IP review?
> Not to put too fine a point on this, but it sounds like you are talking about boilerplate
(and authored) template code that Bison incorporates in its output.  It is also tricky because
the Bison output is computer source code.  That is an interesting case.
> In the US, original work of authorship is pretty specific in the case of literary works,
which is where software copyright falls the last time I checked (too long ago, though).  I
suspect that a license (in the contractual sense) can deal with more than copyright.  And,
if Bison spits out copyright notices, they still only apply to that part of the output, if
any, that qualifies as copyrightable subject matter.
> Has the Bison claim ever been tested in court?  Has anyone been pursued or challenged
for infringement? I'm just curious.
>  - Dennis
> -----Original Message-----
> From: Norbert Thiebaud []
> Sent: Wednesday, September 28, 2011 22:11
> To:;
> Subject: Re: A systematic approach to IP review?
> On Wed, Sep 28, 2011 at 7:55 PM, Dennis E. Hamilton
> <> wrote:
>> I'll stand by my original statement.
>> I'm not going to get into the Pixar case since it doesn't apply here.
> I did not say it applied to the Visual studio generated cruft... I
> merely commented on the blanket assertion that 'computer generated =>
> no copyright'
>> The Bison manual may have license conditions on what can be done with the generated
artifact, but I suggest that is not about copyrightable subject matter in the artifact.
> Actually it is. The only claim they could legally have _is_ on the
> generated bit that are substantial piece of code copied from template
> they provide, namely in the case of a bison generated parser the whole
> parser skeleton needed to exploit the generated state-graph. the whole
> paragraph is about the copyright disposition of these bits. and in the
> case of bison they explicitly grant you a license to use these bits in
> the 'normal' use case... my point being that the existence of that
> paragraph also disprove the assertion that 'computer  generated => no
> copyright'
> You could write a program that print itself... the mere fact that it
> print itself does not mean you lose the copyright on your program...
> That being said, I do think you are on the clear with the Visual
> Studio generated cruft... but not merely because there is 'computer
> generation' involved.
> Norbert

View raw message