Delayed release of GPLed code 'ok'?

[Olliver Schinagl]

Hey list,

I was talking to a friend the other day and it was discussed what the
legal ramifications where on delayed release.

Ignoring the use case specifically, if I would write a program, based on
a GPLed program, I have to release source or accompany it with an offer
to do so. Before releasing this code however, it will have to pass
through the legal department for example where they sanction it all. So
that introduces a delay. It could also be very much that a company would
use that as an excuse, to delay the release of the code.

How long of a delay would be acceptable? This is not a case where one
would try to circumvent the release or anything. The intention is to
release everything as expected, it is not used as some con to not
release things as needed. It may merely be used as a way to maximize
profits (its hard enough to earn money on OSS and OSHW, but lets not get
into that argument).

Thanks for your thoughts,

Olliver

[Bruce Perens]

If the legal department has not already sanctioned the source code
release, you should certainly not even think of distributing a binary.
All of the mechanism to provide source must be in place before the
binary goes out.

Under the terms of the GPL, the acceptable delay is zero.

There are many ways to make money involving the GPL. This just isn't one
of them.

Thanks

Bruce

Post by Olliver Schinagl
Hey list,
I was talking to a friend the other day and it was discussed what the
legal ramifications where on delayed release.
Ignoring the use case specifically, if I would write a program, based
on a GPLed program, I have to release source or accompany it with an
offer to do so. Before releasing this code however, it will have to
pass through the legal department for example where they sanction it
all. So that introduces a delay. It could also be very much that a
company would use that as an excuse, to delay the release of the code.
How long of a delay would be acceptable? This is not a case where one
would try to circumvent the release or anything. The intention is to
release everything as expected, it is not used as some con to not
release things as needed. It may merely be used as a way to maximize
profits (its hard enough to earn money on OSS and OSHW, but lets not
get into that argument).
Thanks for your thoughts,
Olliver

[Eric Appleman]

Tell that to Samsung.

It takes them days after a Verizon Galaxy S3 over-the-air update AND an
explicit GPL request using their OSS form on their website to get them to
release sources.

- Eric

Post by Bruce Perens
If the legal department has not already sanctioned the source code
release, you should certainly not even think of distributing a binary. All
of the mechanism to provide source must be in place before the binary goes
out.
Under the terms of the GPL, the acceptable delay is zero.
There are many ways to make money involving the GPL. This just isn't one
of them.
Thanks
Bruce

Post by Olliver Schinagl
Hey list,
I was talking to a friend the other day and it was discussed what the
legal ramifications where on delayed release.
Ignoring the use case specifically, if I would write a program, based on
a GPLed program, I have to release source or accompany it with an offer to
do so. Before releasing this code however, it will have to pass through the
legal department for example where they sanction it all. So that introduces
a delay. It could also be very much that a company would use that as an
excuse, to delay the release of the code.
How long of a delay would be acceptable? This is not a case where one
would try to circumvent the release or anything. The intention is to
release everything as expected, it is not used as some con to not release
things as needed. It may merely be used as a way to maximize profits (its
hard enough to earn money on OSS and OSHW, but lets not get into that
argument).
Thanks for your thoughts,
Olliver

[Florian Weimer]

Post by Olliver Schinagl
How long of a delay would be acceptable? This is not a case where one
would try to circumvent the release or anything. The intention is to
release everything as expected, it is not used as some con to not
release things as needed.

Of course it is. Why have a legal review if it never can block a
source release?

As Bruce said, you have to to prepare the source code release as part
of the release process for the binaries.

There have been cases where the FSF has approved a delay in publishing
source code. The FSFE has entered a strange licensing deal with Bacula
Systems. But the gist is that you need an additional agreement, there
is an expectation that you can't do it with the plain GPL (although
counsel to various companies in the Android sphere thinks it's okay).

[Kern Sibbald]

Hello,

Please see below ...

Post by Florian Weimer

Post by Olliver Schinagl
How long of a delay would be acceptable? This is not a case where one
would try to circumvent the release or anything. The intention is to
release everything as expected, it is not used as some con to not
release things as needed.

Of course it is. Why have a legal review if it never can block a
source release?
As Bruce said, you have to to prepare the source code release as part
of the release process for the binaries.
There have been cases where the FSF has approved a delay in publishing
source code. The FSFE has entered a strange licensing deal with Bacula
Systems.

The licensing deal with Bacula Systems has nothing directly to do with
delay in publishing source code. It is perhaps unusual (or strange as
you say), because Bacula Systems holds a dual license to the source code
basically because I am the original author and have FLAs from all the
contributors. Though there are a number of points, for me, the main
point of the license is that Bacula Systems, because I believe in open
source, agrees to give all code it writes that is not encumbered by 3rd
party proprietary licenses (example Oracle plugin) to the community
within a five year period. So Bacula Systems though it is an "open
core" company has made a very strong commitment to support the Bacula
community version. Disclaimer: the FSFE does not encourage the open
core business model. End Disclaimer. I personally believe that other
open core companies might be willing to make similar agreements, so they
could be encouraged to have a business model that is somewhere between
open source and open core, which could significantly increase the amount
of open source code.

Best regards,
Kern Sibbald
Bacula Project Manager,
CTO Bacula Systems

Post by Florian Weimer
But the gist is that you need an additional agreement, there
is an expectation that you can't do it with the plain GPL (although
counsel to various companies in the Android sphere thinks it's okay).

[Bradley M. Kuhn]

As Bruce said, you have to to prepare the source code release as part of
the release process for the binaries.

Post by Bruce Perens
All of the mechanism to provide source must be in place before the
binary goes out.
Under the terms of the GPL, the acceptable delay is zero.

It indeed concerns me greatly that companies have a tendency to do what I
call "offer-fail" violations, where they have an offer for source out there
with the binaries, but when you request the source, they take months to get
you source code and it clearly doesn't build at all when you get it. That
shows pretty clearly that they were just not in compliance from the start and
made the offer as a cover-up of their violation.

There have been cases where the FSF has approved a delay in publishing
source code.

I wonder if you're conflating here issues of general licensing policy with GPL
enforcement policy.

For example, a classic example of FSF's stance on general licensing policy
about "publishing delay of source" is the case of Aladdin Ghostscript
(proprietary) delay-released release as GNU Ghostscript under GPL.
Situations like that are a question of good policy generally for Free
Software, but not related to license violation. Since Aladdin was the 100%
copyright holder of Ghostscript at the time, there was no GPL violation by
Aladdin's proprietary version.

So, that has nothing to do with GPL enforcement policy, since Aladdin didn't
violate GPL.


-- bkuhn

[Florian Weimer]

Post by Bradley M. Kuhn

Post by Florian Weimer
There have been cases where the FSF has approved a delay in publishing
source code.

I wonder if you're conflating here issues of general licensing policy with GPL
enforcement policy.

No, I don't think so. The example I had on my mind was the Itanium
architecture bootstrap. But that doesn't match the usual GPL
violation because the in that case, the NDA also prevented the
distribution of binaries.

[Neil Brown]

Post by Olliver Schinagl
How long of a delay would be acceptable?

As others have said, there is no position within GNU GPL 2.0 or 3.0 which permits a delay in the provision of the source code. However, perhaps, if your question were phrased slightly differently, it might elucidate some further interesting and useful comments? Perhaps:

A company develops, manufacturers and retails a product containing GNU GPL 2.0 covered code, and distribution of the binary is to be accompanied by a written offer conforming to section 3(b).

Someone chooses to take up the written offer, and asks the company for a copy of the source code.

How long need the delay be between someone asking for the code, and the company providing it, for that company to be considered non-compliant?

(Hopefully formulating the question in this way does not make me one of the horrible lawyers intent on circumventing the GPL but, if it does, at least take comfort from the fact that I am not *your* horrible lawyer :))

Neil

__________

Neil Brown
***@neilzone.co.uk | http://neilzone.co.uk

[Olliver Schinagl]

Post by Neil Brown

Post by Olliver Schinagl
How long of a delay would be acceptable?

As others have said, there is no position within GNU GPL 2.0 or 3.0
which permits a delay in the provision of the source code. However,
perhaps, if your question were phrased slightly differently, it might
A company develops, manufacturers and retails a product containing GNU
GPL 2.0 covered code, and distribution of the binary is to be
accompanied by a written offer conforming to section 3(b).
Someone chooses to take up the written offer, and asks the company for a
copy of the source code.
How long need the delay be between someone asking for the code, and the
company providing it, for that company to be considered non-compliant?
(Hopefully formulating the question in this way does not make me one of
the horrible lawyers intent on circumventing the GPL but, if it does, at
least take comfort from the fact that I am not *your* horrible lawyer :))

That also is an interesting question, to which I hope some will offer a
sensible answer.

Another way of rephrasing it would be.

What if I am the sole copyright holder and I understand and feel the
importance of the GPL (I stress this as I said before, there is no
intention whatsoever to circumvent the GPL) and release a product (hard
+ software) of which I'm the sole copyright holder with a promise of
releasing the code after 6 months. Technically I haven't released any
code yet, so there's no license to adhere to yet, but that's less
important I guess.

Maybe that makes it a little bit different. Now if I where taking a
existing GNU/Linux system, had modifications done to it, kept those
patches back for 6 months, I do think that is far 'worse', but going
with Neil's rephrasing above makes it an interesting point.

Post by Neil Brown
Neil
__________
Neil Brown

[Bruce Perens]

If you are the sole copyright holder, you have the right to do what you
want with the software. If you choose to issue that software with
statements that the GPL applies to it, and /then/ you don't honor the
GPL terms, it's not an issue of copyright infringement. It may be some
sort of fraud, or violation of the contract which you offered to others
(not that FSF says it /is/ a contract).

If it's not egregiously deceptive it would probably not be a high
priority for anyone to enforce against it.

Thanks

Bruce

Post by Olliver Schinagl
What if I am the sole copyright holder and I understand and feel the
importance of the GPL (I stress this as I said before, there is no
intention whatsoever to circumvent the GPL) and release a product
(hard + software) of which I'm the sole copyright holder with a
promise of releasing the code after 6 months. Technically I haven't
released any code yet, so there's no license to adhere to yet, but
that's less important I guess.

[Thomas Charron]

Post by Olliver Schinagl
How long of a delay would be acceptable?
As others have said, there is no position within GNU GPL 2.0 or 3.0 which
permits a delay in the provision of the source code. However, perhaps, if
your question were phrased slightly differently, it might elucidate some
A company develops, manufacturers and retails a product containing GNU GPL
2.0 covered code, and distribution of the binary is to be accompanied by a
written offer conforming to section 3(b).
Someone chooses to take up the written offer, and asks the company for a
copy of the source code.
How long need the delay be between someone asking for the code, and the
company providing it, for that company to be considered non-compliant?
(Hopefully formulating the question in this way does not make me one of
the horrible lawyers intent on circumventing the GPL but, if it does, at
least take comfort from the fact that I am not *your* horrible lawyer :))

IANAL, but, 'reasonable' amount of time, which has no firm definition.

Bad example, but, let's say someone requests something that I did 2 years
ago, and I no longer support it. In order to get the source, I have to get
it from off site storage, restore it, and then get it to you. If that
happens to take me a few weeks, I wouldn't think it was unreasonable. Some
zealots might, but in the end, there is no firm definition. So long as the
effort required to provide the code is reasonable, then there is no issues.

I suspect the question in a court of law would be, was the intent of the
delays to impede or discourage requests for the source code.

[Bruce Perens]

Please take the name-calling elsewhere.

Imagine a film company waiting patiently in court until you could find
the right files to cure an infringement of their property.

In this sort of case, the copyright holder can persuade the judge that
there is probability of an award of some size, and that your financial
resources should be attached lest you fail to find the right files and
then have to pay damages.

We're used to the Free Software folks nicely just asking for compliance.
But they have all of the tools that the law makes available to more
aggressive copyright holders.

Thanks

Bruce

Post by Thomas Charron
Bad example, but, let's say someone requests something that I did 2
years ago, and I no longer support it. In order to get the source, I
have to get it from off site storage, restore it, and then get it to
you. If that happens to take me a few weeks, I wouldn't think it was
unreasonable. Some zealots might, but in the end, there is no firm
definition. So long as the effort required to provide the code is
reasonable, then there is no issues.
I suspect the question in a court of law would be, was the intent of
the delays to impede or discourage requests for the source code.
--
-- Thomas

[n***@neilzone.co.uk]

Post by Bruce Perens

Post by Thomas Charron
Bad example, but, let's say someone requests something that I did 2
years ago, and I no longer support it. In order to get the source,
I have to get it from off site storage, restore it, and then get it
to you. If that happens to take me a few weeks, I wouldn't think
it was unreasonable.

Imagine a film company waiting patiently in court until you could
find the right files to cure an infringement of their property.

The key difference, to my mind, in these two examples, is that, in
Bruce's example, there is an already-established "infringement of
their property", whereas, in Thomas' there is not yet an established
infringement. That is the very question at the heart of the
discussion, as I understand it: if I make a written offer, how long do
I have to supply the source code before my failure to do so would be
interpreted as a failure to comply with the conditions of GNU GPL 2.0,
such that, by failing to comply with the licence terms, my
distribution infringes copyright?

I am not sure that the two situations above are analagous.

However (and I am musing now), this discussion presupposes that
failure to provide the source code *does* lead to copyright
infringement.

The requirement of GNU GPL 2.0, s3(b), is that Company A, which is
distributing a product containing covered code, must, if using the
"written offer" route, accompany distribution of the binary with a
written offer.

Let us assume that Company A has included a written offer with their
product, saying "Company A will give the complete corresponding source
code for $product to any third party seeking it, for $1. Contact
$address".

Company A has satisfied its obligation under the licence: it has made
a written offer.

With what section of the licence does Company A fail to comply if it
does not subsequently send the source code to someone asking for it?

Or would the requesting party need to establish some sort of direct
relationship by virtue of Company A having made an offer, such as
breach of contract?



Neil

[Bruce Perens]

If you are able to follow the GPL terms, your distribution of binaries of GPL software is non-infringing. But as soon as you can't follow all of your obligations under the GPL terms, you are back to "All Rights Reserved" and none of the GPL terms apply to you any longer. At that moment, you are infringing.

If you do not have the complete and corresponding source code on hand for the three years during which you are obligated to fulfill a request, and have to recover them in some way that takes a significant amount of time, that is what the court would see as a failure of due diligence and you are infringing at that time. If there is any "reasonable time", it is only the time necessary to respond with a file that you already have on hand because you have pursued your due diligence in maintaining it.

Due diligence is something you are expected to take care of before being sued. To the extent that you take care of it after being sued, that is a remedy. You can still have damages for the period of infringement before your remedy.

So, if you fail to provide source code within just a few days, a lawsuit is filed, and you then manage to come up with a bit-perfect binary match (and good luck with _that_), you potentially have some infringing time for which damages can be assessed by the court.

In practice you have that nice Bradley Kuhn who is more lenient than any court (I am speaking from experience). But he doesn't have to be.

Thanks

Bruce

[Bruce Perens]

The courts don't ask if the plaintiff allowed the defendant a month to
respond. They act first to mitigate the damage and give the defendant a
strong incentive to remedy it through injunction , enjoinment, etc. They
do this because they are persuaded that an infringement of any duration
causes irreparable damage to the copyright holder.

/T//hen/ the defendant gets time to remedy it, with the assumption that
the duration before that remedy will be taken into account in levying
damages.

I work for infringers why are trying to clean up their act, and their
attorneys, every day. Most of them haven't been noticed by you guys yet,
but their attorneys have counseled them regarding what can go wrong.
None of them are proceeding under the glib assumption of a court that
makes sure everything will be reasonable.

Thanks

Bruce

[Bradley M. Kuhn]

I agree with others on this thread that the question of what
partially-Free/delayed-release licenses that a sole copyright might
grant are not really relevant to discussion about GPL compliance.

I further agree with those in this thread who point out that the GPL
doesn't *explicitly* state an amount of time a binary distributor using
a written offer has to respond with the CCS once the written offer is
exercised. That much is plainly seen by reading GPL, after all.

I even further agree that the Courts (if they ever consider the question
at all -- see below) will likely use a "reasonableness" standard on this
issue. Likely, I and my fellow copyleft advocates will disagree with
what the violator says is reasonable. (Already, violators typically
claim to me that three months (or even more) is "reasonable". I
disagree vehemently with them on this.)

However, there are two broader issues (both of which are hinted at in
Bruce's email that) IMO are worth noting in more detail:

(0) Remember that in almost every jurisdiction -- and the USA in
particular -- the GPL is what the *Defendant* has to argue BACK as
their license that gave them permission in the first place. i.e.:

- Plaintiff says: "You infringed our copyrights."

- Defendant says: "No, I didn't, I had a license."

- Plaintiff says: "What license? Only license we gave you was GPL."

- Defendant says: "Yeah, that's the one, the GPL!!!"

- Plaintiff says: "Well, we told you already before the case
started: you failed to comply with GPL, which caused those
permissions to terminate. Now you have no permission. We say
again: You're infringed our copyrights".

- Rinse, repeat.


Believe it or not, the Judge *doesn't* actually care that much about
GPL. The Judge *only* pays attention to the GPL to the extent to
which the *Defendant* argues "a record of perfect GPL compliance"
(because, without a record of flawless compliance, the rights under
GPL terminated, and the GPL is moot for that case.)

(1) Focus now on this fact: for the Defendant to convince the Judge that
their compliance was perfect, the Defendant must show compliance
with GPL *fully* for the Plaintiff's copyrights. That's for *ever*
*single* *clause* of GPL -- going back to the Defendant's very first
act of copying, modifying and/or distribution of Plaintiff's
copyrights. Any *single* mistake anywhere in there? Well, then,
the rights terminated, and it's a pure copyright infringement
question, and the GPL is moot.

Consider this non-hypothetical example, particularly relevant to the
current discussion (this happens to me all the time, BTW):

- An offer for source is tested.

- The violator takes six months to send some sources. [This itself
is a violation, using the aforementioned "reasonableness"
standard. Maybe the violator argues it isn't. They're wrong,
but for the sake of argument, let's keep going one more step:]

- The CCS they sent is grossly inadequate, missing "scripts used
to control compilation and installation of the executable",
missing the source for various parts of combined works, etc.
The CCS further has incorrect licensing notices, including some
files marked as proprietary that are clearly GPL'd.

- So, now they violate on *many* clauses of the GPL. I stop
looking and send them a report about how bad their CCS is.

Now, years later, maybe we end up in Court over this. Consider how
arguments are constructed in Court. In Court, your lawyer often
stacks arguments by "arguing in the alternative". Like this:

We understand that the Defendant claims that it was reasonable to
take six months to get us CCS. We think that was an unreasonable
amount of time and is a violation of the license. Even if Your
Honor finds, despite our objections to the contrary, that taking
six months to get us source code is compliant with our license,
THEN we refer your honor to the fact that the source provided
after six months lacked 'scripts used to control compilation and
installation of the executable', which is also a violation of our
license. If Your Honor finds, despite our objection, that those
scripts were actually included, THEN we refer Your Honor to [ yet
another violation] ...."

In summary:

No violator in my decade and a half of enforcement work has had a
violation that hinged *ONLY* on one little tiny clause in GPL: it's
always lots of different clauses. Meanwhile, the GPL is *their*
defense, not our affirmative statement. Our affirmative statement is
copyright infringement, pure and simple. The *Defendant* has the
burden of "perfect compliance proof" to the Judge.

====

All that said (and truly hoping it's useful), I have to add this again:

As always, I'm frustrated by this thread: I really get the impression
that very few of you opining on these threads have done very much (if
any) GPL enforcement.

When I read these threads, I find myself thinking: "the issues everyone
on these threads want to obsess over aren't the important ones, and
meanwhile the stuff that no one mentions and/or everyone glosses over
are the really hard issues".

As the most prolific GPL enforcer in the world, I admittedly have a lot
of background info, knowledge, and experience that makes it easy for me
to see that. In the last year, I've been participating more on this
list because, as a GPL enforcement expert, I feel it's my duty to share
my knowledge with all of you. But, I hope you will all indulge me
briefly when I shout: "YOU ALL ARE DRIVING ME CRAZY WITH THIS STUFF." :)
It's really hard to care about how many angels dance on the head of any
particular pin that folks on this list think they've seen, when the
world is filled with hundreds of unresolved no-source-nor-offer
violations. And with that, I'm going back to work!

Also as always: volunteers who want to help me actually enforce the GPL
are welcome and learn how this works in the Real World are welcome.
Contact me off list for that.

[Neil Brown]

Post by Bradley M. Kuhn
I hope you will all indulge me
briefly when I shout: "YOU ALL ARE DRIVING ME CRAZY WITH THIS STUFF." :)

Indulgence granted :)

It is always a little dangerous to speak on behalf of someone else, but, assuming that risk, I would have thought that, for Andrew and me, it is more a matter of identifying potential ambiguity and discussing it - perhaps primarily out of pure intellectual curiosity - than attempting to irk you (or Bruce), or to create a cottage industry of "here's how to avoid the GPL" advisory services.

(As for not having years of experience doing enforcement work: absolutely. But almost ten years, for me, at least, in my employed capacity, trying to *prevent* the need for enforcement work, by promoting and advising on good development practice, and supply chain management, to drive compliance :))

(If I genuinely have done something to irk you, you've got my contact details, and do feel free to pick up off-list!)

Best wishes


Neil


Neil Brown
***@neilzone.co.uk | http://neilzone.co.uk

[Bradley M. Kuhn]

Neil,

Post by Neil Brown

Post by Bradley M. Kuhn
I hope you will all indulge me
briefly when I shout: "YOU ALL ARE DRIVING ME CRAZY WITH THIS STUFF." :)

Indulgence granted :)

Thanks. :)

Post by Neil Brown
for Andrew and me, it is more a matter of identifying potential
ambiguity and discussing it - perhaps primarily out of pure
intellectual curiosity - than attempting to irk you (or Bruce), or to
create a cottage industry of "here's how to avoid the GPL" advisory
services.

Well, if it were just about irking me, I wouldn't care. I would agree
if someone said "you deserve to be irked once in a while". The issue is
rather that I am in the (perhaps unique) position of having lived
through situations where emails on this list (and/or others like it)
have been studied by GPL violator defense counsel and *have been*
introduced into evidence in GPL enforcement cases. In fact, I've been
diposed more than once in GPL enforcement cases and presented with (and
asked questions about) documents from mailing lists like this.


To explain it another way: the situation reminds me of why I dropped out
of my CS PhD program: I found that most people in academic computer
science didn't care in the least about the sociopolitical ramifications
of their work. They just loved the "problem", whatever problem they
specialized in. Similarly, I feel that most people who discuss GPL have
a tendency to look at the GPL as a document of nonpolitical academic
interest, not as a key policy cornerstone that people have to go out
there and use to defend software freedom.

Admittedly, people give me (IMO an unduly) hard time about "failing to
discuss dispassionately about these topics". I'm not dispassionate; by
contrast, I'm emotionally invested, because I believe proprietary software
harms people, and Free Software is the solution. In turn, GPL is the shield
that defends people from that harm. So, damn right, I'm passionate. I
even argue that those who take a dispassionate, "what if"-"devil's
advocate"-kind-of attitude to GPL discussion (at best) are neutral on
whether or not proprietary software harms society. That not only irks me,
it creates real adversarial danger (or, at best, becomes useful raw
material for such adversarial danger) for those of us fighting for
software freedom and the GPL.

Post by Neil Brown
(If I genuinely have done something to irk you, you've got my contact
details, and do feel free to pick up off-list!)

As I said, I don't mind if you irk me; feel free to do so any time you
like. :) I just mind if, as an expert in licensing, you say something
"devil's advocate"-ish today on this list that ends up in a culmination
of you being called as an opposing witness to defend some GPL violator
from whom I'm trying to compel their required GPL source code release.

And, *that* is not a hypothetical scenario. I've *more than once* faced
the defense naming as their expert witness participants in this list and
others like it.

Yes, I'm a admittedly a cynic, but given that fact above, I have to
assume that anyone making odd statements about GPL compliance is going
to get hired to fight against enforcement efforts later: it's happened
before, it will happen again. GPL violators pay a lot more than my job
does, pure and simple, so if one doesn't believe that proprietary
software is harmful to society, one is likely to take such a job because
it's, simply put, going to pay better than being on the "right side".

Hopefully that explains why I get so upset when folks make such
blatantly false claims like "making the offer for source ends your
compliance obligations", even if done with a devil's advocate point of
view. (As I've already said on the record, that claim has no basis.)

And, to be a bit less cynical, I truly believe that nearly everyone who
posts on this list is well-intentioned and wants software to be Free.
My primary point here is there's a slippery slope of policy positions
that starts at "devil's advocate" and ends with helping violators avoid
their obligations under GPL.

[Thomas Charron]

Post by Bruce Perens
If you are able to follow the GPL terms, your distribution of binaries of
GPL software is non-infringing. But as soon as you can't follow all of your
obligations under the GPL terms, you are back to "All Rights Reserved" and
none of the GPL terms apply to you any longer. At that moment, you are
infringing.
If you do not have the complete and corresponding source code on hand for
the three years during which you are obligated to fulfill a request, and
have to recover them in some way that takes a significant amount of time,
that is what the court would see as a failure of due diligence and you are
infringing at that time. If there is any "reasonable time", it is only the
time necessary to respond with a file that you already have on hand because
you have pursued your due diligence in maintaining it.

I never said it was smart, I never even said it was legally a great
idea. But let's say it takes a month to get it to you. Is that too long?
How about 2 weeks? Where is the number? Since it is not specified, it
would be left to the court to decide, based on what would be considered
'reasonable'. I'd say a month is reasonable, personally. But then again,
that's just me. I also know that a company such as Motorola Mobility in
many cases would wait until a request was made for a given kernel, until
preparing and sanitizing it for public release, but would not generally do
so prior to the request. And they where, in these cases, compliant.

Now, common sense would say what Bruce is saying. If you don't HAVE it,
how do you even KNOW you can produce it? And there you have the risk an
entity takes when they decide to NOT have what Bruce assumes should be the
norm. They might not be able to comply, and open themselves up for the
liability which would result in non compliance.

Post by Bruce Perens
Due diligence is something you are expected to take care of before being
sued. To the extent that you take care of it after being sued, that is a
remedy. You can still have damages for the period of infringement before
your remedy.

The fact of the matter is, most entities who do not end up complying, or
wait for the initial request to produce something that ends up not even
being identical to what they shipped, run the same risk as being hit by
lightning. It happens, yes, but the likelihood? Not very.

Post by Bruce Perens
So, if you fail to provide source code within just a few days, a lawsuit
is filed, and you then manage to come up with a bit-perfect binary match
(and good luck with _that_), you potentially have some infringing time for
which damages can be assessed by the court.

Your now talking about a hypothetical that has never, to my knowledge,
happened. Fail to produce the code in a few days? By all means quote a
single example of this being considered a reasonable time.

Post by Bruce Perens
In practice you have that nice Bradley Kuhn who is more lenient than any
court (I am speaking from experience). But he doesn't have to be.

We're having academic arguments, which in my opinion do turn into hit by
lightning comparisons. We're all honest here. How many instances of GPL
infringements exist in the work in the 20+ years of the GPL? How many
cases have led to the liabilities your talking about?

Bruce, you're right there, and I *KNOW* we agree that knowingly violating
the terms of the GPL should lead to much bigger and harsher judgements then
have been passed down. You yourself had your own work infringed on, much
more then I could even dream anything of mine being used and abused. But
taking the stance that 1 millisecond after the request has been made your
not in compliance really isn't fair to anyone.

Thomas

[Bradley M. Kuhn]

Company A has satisfied its obligation under the licence: it has made a
written offer.

The GPL clearly says the offer has to be valid. The word valid clearly
appears in this section in both GPLv2 and GPLv3.

Company A has no argument: if someone exercises the offer and
it fails to work, Company A hasn't complied. Simple as that.

With what section of the licence does Company A fail to comply if it does
not subsequently send the source code to someone asking for it?

This question seems purposely naive to me, almost as if it seeks to elicit
GPL-violator-friendly answers. My answer above settles this question fully,
IMO. Valid means valid.

-- bkuhn

[Andrew Katz]

Bradley

Post by Bradley M. Kuhn

Company A has satisfied its obligation under the licence: it has made a
written offer.

The GPL clearly says the offer has to be valid. The word valid clearly
appears in this section in both GPLv2 and GPLv3.

A valid offer is one that is legally enforceable. Valid, as you say, means valid. If the offeror chooses not to comply, then the offeree, if the offer is valid, has a claim. If, on the other hand, your interpretation is correct, then if the offeror fails to comply, the offer is somehow not valid, and the court will therefore be unable enforce it in favour of the offeree. So you're saying that a court can't enforce a validly formed contract against a party because that party has breached the contract? That makes no sense at all to me. In what sense does failing to honour a contractual commitment render the underlying offer invalid? f that's correct no contract could ever be enforced.

If the offeree succeeds in enforcing the offer, then it is, by definition, valid. The offeree has the source code, and everyone is happy. If the offeree does not succeed in enforcing the offer, then the offer is invalid, and there has been a breach of the GPL, and a violation of the licensor's copyright.

Post by Bradley M. Kuhn
Company A has no argument: if someone exercises the offer and
it fails to work, Company A hasn't complied. Simple as that.

I tried very hard to convince myself that this interpretation is correct, but I can't. (And I agree, I think the word 'valid' is the best place to find the argument we're all looking for).

Post by Bradley M. Kuhn

With what section of the licence does Company A fail to comply if it does
not subsequently send the source code to someone asking for it?

This question seems purposely naive to me, almost as if it seeks to elicit
GPL-violator-friendly answers. My answer above settles this question fully,
IMO. Valid means valid.

As I've made it clear throughout this thread, I'm really, really trying to convince myself that this analysis is wrong. But to stretch the word 'valid' so that it no longer means 'legally enforceable' but 'legally enforceable and actually carried out' seems a step too far, and results in the logical inconsistency I describe above.

Also, note that this is not necessarily GPL-violator friendly. The licensor is also a person who can accept the offer and require the source code to be delivered up, and is therefore in a position to re-realease the (presumably modified) source. This is achieved though contract, and not copyright, but the enforcement mechanism is still there. In fact, in some ways, it's stronger, because it provides a path to compulsion of release of the source through specific performance, something that is unlikely to exist (at least in common law jurisdictions), if no offer is made.

If the drafters wanted to say 'You can distribute the binary so long as you provide the complete corresponding source to anyone who asks within three years of distributing the binary', why didn't they say that? I can only think it's because you can't make the lawfulness of a present act conditional on a future event, hence the (I think, rather neat), wording, but it does cause this issue.

FWIW I'm spending time doing this because I think GPL2 is a wonderful thing and because I'm trying to understand it, and the best way to understand something is often to pull it apart and see how it works (admittedly this doesn't work so well for kittens). I don't, however, think it's a religious text, and that, accordingly, certain questions are taboo.



Best


Andrew

[Bradley M. Kuhn]

Post by Andrew Katz
A valid offer is one that is legally enforceable. Valid, as you say,
means valid. If the offeror chooses not to comply, then the offeree,
if the offer is valid, has a claim. If, on the other hand, your
interpretation is correct, then if the offeror fails to comply, the
offer is somehow not valid, and the court will therefore be unable
enforce it in favour of the offeree.

The above reads like sophistry to me. It's simpler than that:

An offer being invalid means that the offer failed to yield the
materials that were offered. The offeror, who is engaged in distribution
of copyrighted works, has therefore failed to meet the obligations under
GPL, and has violated GPL and has thus lost the only license that the
offerer had to copy, modify, and distribute those copyrighted works.
So, any further distribution on their part is now copyright
infringement. That's all there is to it!

Post by Andrew Katz

Post by Bradley M. Kuhn
Company A has no argument: if someone exercises the offer and
it fails to work, Company A hasn't complied. Simple as that.

I tried very hard to convince myself that this interpretation is
correct, but I can't.

The fact that you're not convinced doesn't make it false. :)

Post by Andrew Katz
I think the word 'valid' is the best place to find the argument we're
all looking for).

I think you and I are looking for very different arguments.

Post by Andrew Katz
This is achieved though contract, and not copyright, but the
enforcement mechanism is still there.

I've been supportive of the methods to enforce GPL via contract law,
such as the successful work in Australia to use Australian fraud law as
a cause of action for a GPL violation victim.

However, I don't support GPL violators' pathetic attempts to eviscerate
some copyright claim, even if it happens to help bolster a contract
claim under GPL. I don't argue that GPL *can't* be used as a contract,
too, but it's primarily a copyright license.

Post by Andrew Katz
I don't, however, think ... that ... certain questions are taboo.

Upon rereading this thread, I verify that *I* didn't say any questions
were taboo. Rather, to restate my objection using litigation
terminology for all you lawyers: the question was stated like it was a
leading question for a hostile witness.

[Andrew Katz]

Hi Bradley
Re: Delayed release of GPLed code 'ok'?

Post by Andrew Katz
A valid offer is one that is legally enforceable. Valid, as you say,
means valid. If the offeror chooses not to comply, then the offeree,
if the offer is valid, has a claim. If, on the other hand, your
interpretation is correct, then if the offeror fails to comply, the
offer is somehow not valid, and the court will therefore be unable
enforce it in favour of the offeree.

An offer being invalid means that the offer failed to yield the materials that
were offered. The offeror, who is engaged in distribution of copyrighted works,
has therefore failed to meet the obligations under GPL, and has violated GPL
and has thus lost the only license that the offerer had to copy, modify, and
distribute those copyrighted works.
So, any further distribution on their part is now copyright infringement. That's
all there is to it!

I think we're going to have to disagree on this. My analysis still seems like the simplest, and also explains the slightly convoluted language (i.e. the condition in 3b is a requirement to make the offer, not to make the source available on request, which would be a lot simpler). Feel free to try to convince me otherwise, but any convincing explanation will have, at the very least to (1) explain that convoluted language; and (2) explain how an act restricted by copyright (distribution) can become retroactively infringing based upon an omission occurring at a later time.

Post by Andrew Katz

Company A has no argument: if someone exercises the offer and it
fails to work, Company A hasn't complied. Simple as that.

I tried very hard to convince myself that this interpretation is
correct, but I can't.

The fact that you're not convinced doesn't make it false. :)

No, but I can be pretty persuasive when I'm trying to convince myself ;-)

Post by Andrew Katz
I think the word 'valid' is the best place to find the argument we're
all looking for).

I think you and I are looking for very different arguments.

Well, I'm assuming that we're trying to find an interpretation which means that failure to provide complete corresponding source is a violation of the GPL and therefore a breach of the underlying copyright to the code. You seem to be satisfied with your interpretation, but I'm still very uncomfortable with it.

Post by Andrew Katz
This is achieved though contract, and not copyright, but the
enforcement mechanism is still there.

I've been supportive of the methods to enforce GPL via contract law, such as
the successful work in Australia to use Australian fraud law as a cause of action
for a GPL violation victim.

Yes, this is an interesting aspect. I think consumer protection law, contract and possibly tort are all potentially fruitful places to look.

However, I don't support GPL violators' pathetic attempts to eviscerate some
copyright claim, even if it happens to help bolster a contract claim under GPL. I
don't argue that GPL *can't* be used as a contract, too, but it's primarily a
copyright license.

If they successfully eviscerate the copyright claim, they are not a GPL violator. Whether the attempt is pathetic or not is dependent on whether it works (although any successful attempt is clearly not pathetic, any unsuccessful attempts might be). They may, of course, be antipathetic to free software, which is a different question entirely.

My view is that that, under English law at least, the GPL is a bare licence but it's fairly finely balanced. (I'm told that in Civil Law jurisdictions there is no option but to regard it as a contract). However, the offer issued under 3b (and the subsequent contract formed by its acceptance) has nothing to do with whether the GPL itself is a licence or a contract.

Post by Andrew Katz
I don't, however, think ... that ... certain questions are taboo.

Upon rereading this thread, I verify that *I* didn't say any questions were
taboo. Rather, to restate my objection using litigation terminology for all you
lawyers: the question was stated like it was a leading question for a hostile
witness.

Indeed. So long as we're all agreed that this is a good place to explore these questions, that is good.

All the best


Andrew


Andrew Katz 
Moorcrofts LLP
Corporate Law | Technology Law | Commercial Law | Employment Law 
Employee Incentivisation & Share Schemes | Intellectual Property Law 
Commercial Property Law | Secured Lending
James House, Mere Park, Dedmere Road, Marlow, Bucks SL7 1FJ
[phone] (phone) |  [phone] (mobile) 
Skype: andrewjskatz | Twitter: andrewjskatz

mailto:***@moorcrofts.com 
www.moorcrofts.com
Partners - Adrian Phillips, Andrew Katz, Theresa Hunter, Craig Havard
Registered in England & Wales OC 311818 
Regulated and authorised by the Solicitors Regulation Authority 
"Partner" means a member of Moorcrofts LLP
THIS EMAIL IS CONFIDENTIAL. IF YOU ARE NOT THE INTENDED RECIPIENT, PLEASE LET US KNOW. 
We store email addresses and the names of addressees to assist with future correspondence.

[Bradley M. Kuhn]

Post by Andrew Katz
I think we're going to have to disagree on this.

Probably. But fortunately I'm used to disagreeing with lawyers
who make up bizarre arguments to get GPL violators off the hook. :)

The upside for you: I'm sure you'll have GPL violators lining
up to hire you now, to make this argument on their behalf.

I look forward to seeing you in front of a judge trying to convince the
judge that GPL never intended for the offer to be valid and that the
fact that the violator made an offer but failed to provide CCS somehow
magically met their obligations under all the terms of the license.

Post by Andrew Katz
Feel free to try to convince me otherwise,

I gave up a long time ago trying to convince lawyers to take a
reasonable and correct position, rather than a sophistic one that benefits
their clients. :) It seems I broke that rule in engaging in this
conversation with you. ;)

Post by Andrew Katz
If they successfully eviscerate the copyright claim, they are not a
GPL violator.

Just because a lawyer finds some baseless argument to claim their client
isn't violating, it doesn't magically make them not a violator.

You do realize that I hear claims like this at least once a week from
some violator's defense lawyer? The details change, but
the general tenor is the same. It's like talking to kids who
stole a cookie from the jar and come up with more and more weird
tales of how it couldn't have been them that took the cookie.

In fact, one lawyer recently sent me one of these werid arguments,
and I was able to point out in email response that the lawyer
had argued *against* that point in the past in public talks. The
lawyer called me up to say "You know I just have zeaously defend my
client, right?" Basically, the lawyer was admitting in a non-discoverable
way that he didn't believe the argument either, but it was the
best one for his client, so he made it.

Better lawyers do "look ahead" and make sure their public record
has arguments made that might be useful to future clients. :)

Post by Andrew Katz
Indeed. So long as we're all agreed that this is a good place to
explore these questions, that is good.

No, we're not agreed on that. I don't think it's a good use of this
forum to explore how to make purposely naïve legal arguments in an
effort to help violators fight users' rights to CCS.

Post by Andrew Katz
THIS EMAIL IS CONFIDENTIAL.

This is an absolutely bizarre thing to say in an email on a public
mailing list. Why would you would say something like this in an email
to a public list, when you didn't even set X-No-Archive in
your headers. Your claim that the email is confidential is thus pretty
baseless too. :)

[Andrew Katz]

Bradley

Post by Andrew Katz
I think we're going to have to disagree on this.

Probably. But fortunately I'm used to disagreeing with lawyers who make up
bizarre arguments to get GPL violators off the hook. :)

I have no interest in getting GPL violators off the hook...

The upside for you: I'm sure you'll have GPL violators lining up to hire you now,
to make this argument on their behalf.

Maybe it's a shame for both me and them that I won't accept this sort of work. Thankfully, I'm no longer a barrister, or my professional conduct rules would prevent me from turning the work away.

I look forward to seeing you in front of a judge trying to convince the judge that
GPL never intended for the offer to be valid and that the fact that the violator
made an offer but failed to provide CCS somehow magically met their
obligations under all the terms of the license.

Well, that's not going to happen, but it doesn't mean that wouldn't like to know the answer.

Post by Andrew Katz
Feel free to try to convince me otherwise,

I gave up a long time ago trying to convince lawyers to take a reasonable and
correct position, rather than a sophistic one that benefits their clients. :) It
seems I broke that rule in engaging in this conversation with you. ;)

This position doesn't benefit my clients, many of whom are rightsholders who want to the see the GPL enforced on their behalf. Does that mean you'll carry on talking to me?

Post by Andrew Katz
If they successfully eviscerate the copyright claim, they are not a
GPL violator.

Just because a lawyer finds some baseless argument to claim their client isn't
violating, it doesn't magically make them not a violator.

The point is that if the claim is *successfully* eviscerated (i.e. a judgment is obtained that favours the defendant), then, of course, they are not, at law, a violator.

You do realize that I hear claims like this at least once a week from some
violator's defense lawyer? The details change, but the general tenor is the
same. It's like talking to kids who stole a cookie from the jar and come up with
more and more weird tales of how it couldn't have been them that took the
cookie.

<...!...>

As a matter of interest, if a violator is not complying with the written offer, would you think it is a good idea (irrespective of any copyright claim) to try to enforce the offer, (possibly seeking an order for specific performance for delivery up of the source)?

In fact, one lawyer recently sent me one of these werid arguments, and I was
able to point out in email response that the lawyer had argued *against* that
point in the past in public talks. The lawyer called me up to say "You know I just
have zeaously defend my client, right?" Basically, the lawyer was admitting in
a non-discoverable way that he didn't believe the argument either, but it was
the best one for his client, so he made it.

This may be a feature of mature legal systems that you don't like, but I'm in favour of a rule that says that clients are entitled to best representation. Thankfully, as I say, I am no longer subject to a rule which says that I have to accept work from clients I don't like, but I'm not going to suggest that those people should not be able to get best representation.

Better lawyers do "look ahead" and make sure their public record has
arguments made that might be useful to future clients. :)

Indeed.

Post by Andrew Katz
Indeed. So long as we're all agreed that this is a good place to
explore these questions, that is good.

No, we're not agreed on that. I don't think it's a good use of this forum to
explore how to make purposely naïve legal arguments in an effort to help
violators fight users' rights to CCS.

I'm a tad stumped. If I'm not welcome to stress test the GPL here, then so be it.

Post by Andrew Katz
THIS EMAIL IS CONFIDENTIAL.

This is an absolutely bizarre thing to say in an email on a public mailing list.
Why would you would say something like this in an email to a public list, when
you didn't even set X-No-Archive in your headers. Your claim that the email is
confidential is thus pretty baseless too. :)

Agreed. That shouldn't have been there. Our professional conduct rules do require us to include the rest of that gumph though, so I apologise for not deleting that bit.

Best


Andrew

[Neil Brown]

Post by Andrew Katz
If I'm not welcome to stress test the GPL here, then so be it.

If Harald and Armijn do not interject, I would see that this is a list to discuss legal issues relating to GPL and potential violations of the GPL. This seems bang on topic to me, but it's Harald's list :)

(So far, no regulars have objected either!)


Neil


Neil Brown
***@neilzone.co.uk | http://neilzone.co.uk

[Bradley M. Kuhn]

Post by Neil Brown
If Harald and Armijn do not interject, I would see that this is a list to
discuss legal issues relating to GPL and potential violations of the
GPL. This seems bang on topic to me, but it's Harald's list :)

I agreed that Harald is our host, and his rules are the ones that go. To be
clear: my opinions on the topic are just that -- my opinions -- but I hope my
emails have explained that there's reasoning behind those opinions.

I'd ask everyone to consider the fact that any email you post to this list
*will* be read by a lawyer who is trying to help their client get away with
violating the GPL.

If you think such an email in no way makes that lawyer's job easier, then by
all means post it. If you think there's a chance your email might help a
lawyer defending a violator get a violator off the hook, I'd urge you not to
post it. While I uphold your Free Speech rights to say what you want, I
would also argue that you're not working to help copyleft if you post
the latter kinds of emails.

Post by Neil Brown
(So far, no regulars have objected either!)

Bruce Perens did, FWIW.

-- bkuhn

[Neil Brown]

On 21 Mar 2014, at 00:12, Bradley M. Kuhn <***@ebb.org> wrote:

Morning, Bradley

Post by Bradley M. Kuhn
I hope my
emails have explained that there's reasoning behind those opinions.

Definitely — I might not agree with some of the conclusions, but the exchange so far has certainly helped me understand better why you hold the opinions that you do so. That kind of broadening of appreciation / understanding is one of the things about open academic debate which strongly appeals to me.

Post by Bradley M. Kuhn
I'd ask everyone to consider the fact that any email you post to this list
*will* be read by a lawyer who is trying to help their client get away with
violating the GPL.

Leaving aside perhaps the sub-text of the current debate, as to whether someone is, or is not, a GPL “violator” in any given situation (views are not all aligned here :)), this did resonate with me, and made me think about whether another venue for the discussion might be more appropriate. Whilst there is a reasonably obvious candidate venue, it is not a public one, and would exclude many of those who contribute regularly and thoughtfully (in my view!) to this particular list, and who may have useful insights to offer.

That, perhaps, is the trade-off. I can see why those who feel the entire debate is unnecessary, even harmful, might opt for a “if you really have to, at least do it somewhere private” approach, but, as of yet, I’m not sure where I sit on this.


Best wishes

Neil

__________

Neil Brown
***@neilzone.co.uk | http://neilzone.co.uk

[Bradley M. Kuhn]

Post by Andrew Katz
Well, that's not going to happen, but it doesn't mean that wouldn't like to
know the answer.

I gave you the answer; you just don't like and/or agree with it. The burden
of proof here is on the violator to prove that I'm wrong in a Court. I told
you why they won't succeed. There's nothing more to be said on the topic,
IMO.

Post by Andrew Katz
This position doesn't benefit my clients, many of whom are rightsholders
who want to the see the GPL enforced on their behalf. Does that mean you'll
carry on talking to me?

I'll carry on talking to you in any event; I believe in the principle that
the only way to get smarter is to talk with smart people one disagrees
with. :)

Post by Andrew Katz
The point is that if the claim is *successfully* eviscerated (i.e. a
judgment is obtained that favours the defendant), then, of course, they are
not, at law, a violator.

That statement is so broad that it's trivially true and not helpful. The
point is simply that "if a violator proves the GPL can't be enforced, then
they can't violate it because the license is then unenforcable". But you
never showed here any shred of case law, nor text of the GPL that indicates
the CCS provisions of any version of the GPL won't be enforceable. You
theorize that maybe it's true, and ask me to prove that it's not.

In turn, I ask you to prove that it is, and you also refuse. So, we're at
stand-off.


And, the idea that making an invalid offer is still GPL-compliant is -- to
paraphrase The Dude in the *Big Lebowski* -- "just, like, your opinion, man".

Post by Andrew Katz
As a matter of interest, if a violator is not complying with the written
offer, would you think it is a good idea (irrespective of any copyright
claim) to try to enforce the offer, (possibly seeking an order for specific
performance for delivery up of the source)?

I don't know; I've never thought about it, because, so far, I've never tried
(nor needed to try) to enforce the GPL without the involvement of an
interested copyright holder who wanted me to help them simply do copyright
enforcement.

The GPL *may* be enforceable by a non-copyright-holder under some different
legal mechanism -- to use an example I already gave, Australian fraud law.
I'm told by Brendan Scott he had some success with that.

For my part, I think *I* would only chase such remedies if for some reason
copyright-based enforcement failed to work, or if it were a useful "argument
in the alternative" to use in some specific situation.

But, generally speaking, copyright enforcement always works, so far. The only
problem I've ever had is funding that work. But that's not the fault the
copyright, it's because I'm -- frankly -- too nice and don't chase to the
bitter end the statutory maximums for copyright infringement. I often
restore violators copyright permissions for a financial amount much *less*
than my (and/or my org's) financial cost of doing GPL enforcement because the
violator came into full compliance. I've always felt taking less money in
exchange for good GPL compliance was the morally right thing to do, even if
it means funding more GPL enforcement remains difficult.

Having said that, I encourage people to donate to GPL Compliance Project for
Linux Developers ( http://sfconservancy.org/linux-compliance/ ). :)

Post by Andrew Katz
This may be a feature of mature legal systems that you don't like, but I'm
in favour of a rule that says that clients are entitled to best
representation.

I certainly agree with that regarding criminal issues. I'm dubious on civil
issues, but I haven't thought of a better system, so I tend to agree with it
in civil systems as well, by default.

I also note that *generally speaking*, when I do GPL enforcement, it's much
better if there's a lawyer with some clue on the other side. The problem I
find (which is probably USA-centric) is that litigators' goal is to increase
their billable hours rather than get the best deal for their client, and as
such they have a tendency to give subtly bad advice (not enough to be
malpractice, but enough to edge things every-so-slightly in their favor).

E.g., many legal representatives who know better will tell their clients that
a GPL enforcer like Conservancy or FSF is "just like a patent troll". We're
a world of difference from that in a dozen directions, including that the
penalties for copyright infringement are much less and because we don't want
money primarily but we want compliance with the GPL and/or an injunction.
But, to the clueless violator, it's hard to tell the difference, and I've
seen many a defense attorney bank on that fact by merely "failing to point
out the distinction".

Post by Andrew Katz
If I'm not welcome to stress test the GPL here, then so be it.

I'd point out that "what if" academic discussion about hypothetical scenarios
isn't a "stress test". In fact, I'd argue that more than anyone in the
world, *I* stress test the GPL for (part of) my living because I actually try
to enforce it for its intended purpose, and thus I'm constantly having to
make sure it works. The results after 15 years of my work, plus a bunch of
that of gpl-violations (when they used to actively do enforcement) is that
GPL passes all known stress tests. :)

BTW, I should give Harald credit (as I've done publicly in my conference
talks) that *he* was more ambitious stress tester than I was, because I took
bad legal advice for years that litigation should be avoided at all costs,
even if it meant perpetual non-compliance by some entities. I'm thankful for
Harald in proving that litigation -- while it shouldn't be used lightly -- is
a tool that must necessarily be used semi-regularly in GPL enforcement.

And, Andrew, if you want to help me enforce the GPL, I'm always looking for
volunteers. ;)

Finally, the only other way I can imagine doing a stress test is from the
other side: trying to get away with various types of violations and seeing
what happens. There are probably people trying that right now, but I'm sure
you're not welcome on this list to talk about trying to do *that*.

Post by Andrew Katz
Agreed. That shouldn't have been there. Our professional conduct rules do
require us to include the rest of that gumph though, so I apologise for not
deleting that bit.

Apology accepted! I admit that I pity you that you are in a profession that
makes all your communications confidential by default, and I'm exuberant that
I have exactly the opposite.... working for a Free Software non-profit, I
typically have to make the case for why something *should* be kept
confidential. :)

-- bkuhn

[Andrew Katz]

Hi Bradley

<Bradley and Andrew said stuff>

I really appreciate your time responding on this - it's been a very worthwhile learning experience for me, for which, thank you.

I also heed your warning about being careful not to be too generous with my speculations on GPL enforcement (even if the tools may turn out to be of dubious utility).

FWIW, my argument works less well with GPL3, where the preamble to clause 6 makes it clear that you have to convey the source.

Looking forward to more constructive conversations in the future.

All the best


Andrew

(Oh, and Neil, thanks for being the ever-present voice of reason, even though you ARE ON HOLIDAY)

[Thomas Charron]

Post by Bruce Perens
Please take the name-calling elsewhere.
Imagine a film company waiting patiently in court until you could find the
right files to cure an infringement of their property.

The comparison doesn't apply. There is no infringement, as the GPL does
not stipulate a time frame anywhere. There are requirements to make an
offer for the source code. That offer must be fulfilled. However, our
practical stance on reasonable time to provide it is based on our own
belief, and minus any specified requirement, the law would revert to the
question of, 'Is it reasonable'.

Post by Bruce Perens
In this sort of case, the copyright holder can persuade the judge that
there is probability of an award of some size, and that your financial
resources should be attached lest you fail to find the right files and then
have to pay damages.

Again, not comparable. There is no time. If you ordered a product, and
it did not specify a delivery date, there is no legal requirement that the
thing you ordered even have been produced. Again, it falls back to what is
reasonable.

Post by Bruce Perens
We're used to the Free Software folks nicely just asking for compliance.
But they have all of the tools that the law makes available to more
aggressive copyright holders.

Bruce, I have a lot of respect to you, but just think about this for a
second. It is a HORRIBLE idea for a company to not have everything ready.
They put themselves in a situation where compliance could possible be
impossible. But strictly speaking, bad practice != copyright
infringement. If I'm mistaken, by all means, please point to where in the
GPL it states or even infers what state your software must be in prior to
distribution.

[Andrew Katz]

HI Neil

On 10 Mar 2014, at 21:09, Neil Brown <***@neilzone.co.uk<mailto:***@neilzone.co.uk>> wrote:


On 9 Mar 2014, at 19:06, Olliver Schinagl <oliver+***@schinagl.nl<mailto:oliver+***@schinagl.nl>> wrote:

How long of a delay would be acceptable?

As others have said, there is no position within GNU GPL 2.0 or 3.0 which permits a delay in the provision of the source code. However, perhaps, if your question were phrased slightly differently, it might elucidate some further interesting and useful comments? Perhaps:

A company develops, manufacturers and retails a product containing GNU GPL 2.0 covered code, and distribution of the binary is to be accompanied by a written offer conforming to section 3(b).

Someone chooses to take up the written offer, and asks the company for a copy of the source code.

How long need the delay be between someone asking for the code, and the company providing it, for that company to be considered non-compliant?

This is exactly the right way to analyse this. It's worth considering why 3(b) doesn't say "provide any third party with [the complete corresponding source on request made within 3 years of distribution...]"

I think the answer to this is quite subtle: the GPL is a distribution licence, and, logically, it must be possible to determine whether the condition is complied with at the time of distribution, otherwise the condition is in a Schroedinger superposition of states, and it will never be clear whether the condition was fulfilled or not at the time of distribution, until a period of three years has elapsed (or a request for the source code has been refused).

I suspect that RMS and Eben realised this, and drafted the condition to avoid it. The condition is fulfilled by making the written offer, and not by making the source available.

This has some interesting consequences. There is no guidance as to when the source has to be provided. It is arguable...



(Hopefully formulating the question in this way does not make me one of the horrible lawyers intent on circumventing the GPL but, if it does, at least take comfort from the fact that I am not *your* horrible lawyer :))

...(which puts me in danger of being one of Neil's 'horrible lawyers') that an offer which says 'we will provide you with the source in 10 years' time' is compliant, although clearly completely against the spirit of the GPL. However, assuming that the offer is 'We will provide you with the complete corresponding source on request', and providing that this is legally binding offer (lawyers will realise that there is, in itself, a problem here), then courts would be likely to interpret the lack of a time period as a 'reasonable time' (what else are they to do, if the offer is binding).

The other consequences of this are that, although under the GPL itself, assuming it is interpreted as a bare licence, there is no mechanism to compel the release of the source (although failure to release under 3(a) clearly leads to a breach of copyright), if the written offer is accepted and forms a binding contract, than it may be susceptible to specific performance, so release of the source can be compelled under that secondary contract.

It is also the case that failure to comply with the offer is not a breach of the GPL (so enforceable by the licensor), but is a contractual breach, enforceable by the recipient of the offer. This does suggest a route for a 'horrible lawyer' to circumvent the GPL.

The licensor would have to demonstrate that the 'offer' was not a 'written offer, valid for at least three years...' that there would a valid claim that the GPL had been breached. That's not necessarily impossible, but it could be a very difficult argument to maintain.

All the best


Andrew

[Bruce Perens]

I think this is a subclass of the larger problem "Does the law require surety in the face of an uncertain future". Often it does.

Andy,
I think you are confusing the commission of an act and the revelation
of evidence to others. If you consider it that way. there is no reason
to treat it as a quantum physics experiment.
If you fail to distribute source code on request, at what time does it
become infringement?
The answer is "at the moment that you distributed the binary".
At the time you distributed the binary, you had a due diligence
requirement to have the complete and corresponding source code on hand
and ready for distribution, to have a structure in place to handle that
distribution in a timely manner, and to retain that structure in place
for three years. If you are unable to assure that at the time of
distribution, it's an act of infringement. The only question is when
you are caught.
An embezzler does not become one when it is found that the books have
been cooked. He is an embezzler the moment he diverts the funds. And of
course many embezzlers tell themselves that they will put the money
back, and then they fail to do so. He may have a change of heart,
replace the funds, and never get caught. He may lose the money in Las
Vegas and eventually be revealed. But the commission of the act is the
original diversion rather than the later loss or replacement.
Thanks
Bruce
On March 13, 2014 2:41:06 AM PDT, Andrew Katz

Post by Andrew Katz
HI Neil
On 10 Mar 2014, at 21:09, Neil Brown
On 9 Mar 2014, at 19:06, Olliver Schinagl
How long of a delay would be acceptable?
As others have said, there is no position within GNU GPL 2.0 or 3.0
which permits a delay in the provision of the source code. However,
perhaps, if your question were phrased slightly differently, it might
A company develops, manufacturers and retails a product containing GNU
GPL 2.0 covered code, and distribution of the binary is to be
accompanied by a written offer conforming to section 3(b).
Someone chooses to take up the written offer, and asks the company for
a copy of the source code.
How long need the delay be between someone asking for the code, and

the

Post by Andrew Katz
company providing it, for that company to be considered non-compliant?
This is exactly the right way to analyse this. It's worth considering
why 3(b) doesn't say "provide any third party with [the complete
corresponding source on request made within 3 years of
distribution...]"
I think the answer to this is quite subtle: the GPL is a distribution
licence, and, logically, it must be possible to determine whether the
condition is complied with at the time of distribution, otherwise the
condition is in a Schroedinger superposition of states, and it will
never be clear whether the condition was fulfilled or not at the time
of distribution, until a period of three years has elapsed (or a
request for the source code has been refused).
I suspect that RMS and Eben realised this, and drafted the condition

to

Post by Andrew Katz
avoid it. The condition is fulfilled by making the written offer, and
not by making the source available.
This has some interesting consequences. There is no guidance as to

when

Post by Andrew Katz
the source has to be provided. It is arguable...
(Hopefully formulating the question in this way does not make me one

of

Post by Andrew Katz
the horrible lawyers intent on circumventing the GPL but, if it does,
at least take comfort from the fact that I am not *your* horrible
lawyer :))
...(which puts me in danger of being one of Neil's 'horrible lawyers')
that an offer which says 'we will provide you with the source in 10
years' time' is compliant, although clearly completely against the
spirit of the GPL. However, assuming that the offer is 'We will

provide

Post by Andrew Katz
you with the complete corresponding source on request', and providing
that this is legally binding offer (lawyers will realise that there

is,

Post by Andrew Katz
in itself, a problem here), then courts would be likely to interpret
the lack of a time period as a 'reasonable time' (what else are they

to

Post by Andrew Katz
do, if the offer is binding).
The other consequences of this are that, although under the GPL

itself,

Post by Andrew Katz
assuming it is interpreted as a bare licence, there is no mechanism to
compel the release of the source (although failure to release under
3(a) clearly leads to a breach of copyright), if the written offer is
accepted and forms a binding contract, than it may be susceptible to
specific performance, so release of the source can be compelled under
that secondary contract.
It is also the case that failure to comply with the offer is not a
breach of the GPL (so enforceable by the licensor), but is a
contractual breach, enforceable by the recipient of the offer. This
does suggest a route for a 'horrible lawyer' to circumvent the GPL.
The licensor would have to demonstrate that the 'offer' was not a
'written offer, valid for at least three years...' that there would a
valid claim that the GPL had been breached. That's not necessarily
impossible, but it could be a very difficult argument to maintain.
All the best
Andrew

--
Sent from my Android phone with K-9 Mail. Please excuse my brevity.

[Neil Brown]

Post by Bruce Perens
I think this is a subclass of the larger problem "Does the law require surety in the face of an uncertain future"

And *that* sounds like a debate to be had over a beer...


Neil


Neil Brown
***@neilzone.co.uk | http://neilzone.co.uk

[Bruce Perens]

Andy,

I think you are confusing the commission of an act and the revelation of evidence to others. If you consider it that way. there is no reason to treat it as a quantum physics experiment.

If you fail to distribute source code on request, at what time does it become infringement?

The answer is "at the moment that you distributed the binary".

At the time you distributed the binary, you had a due diligence requirement to have the complete and corresponding source code on hand and ready for distribution, to have a structure in place to handle that distribution in a timely manner, and to retain that structure in place for three years. If you are unable to assure that at the time of distribution, it's an act of infringement. The only question is when you are caught.

An embezzler does not become one when it is found that the books have been cooked. He is an embezzler the moment he diverts the funds. And of course many embezzlers tell themselves that they will put the money back, and then they fail to do so. He may have a change of heart, replace the funds, and never get caught. He may lose the money in Las Vegas and eventually be revealed. But the commission of the act is the original diversion rather than the later loss or replacement.

Thanks

Bruce

Post by Andrew Katz
HI Neil
On 10 Mar 2014, at 21:09, Neil Brown
On 9 Mar 2014, at 19:06, Olliver Schinagl
How long of a delay would be acceptable?
As others have said, there is no position within GNU GPL 2.0 or 3.0
which permits a delay in the provision of the source code. However,
perhaps, if your question were phrased slightly differently, it might
A company develops, manufacturers and retails a product containing GNU
GPL 2.0 covered code, and distribution of the binary is to be
accompanied by a written offer conforming to section 3(b).
Someone chooses to take up the written offer, and asks the company for
a copy of the source code.
How long need the delay be between someone asking for the code, and the
company providing it, for that company to be considered non-compliant?
This is exactly the right way to analyse this. It's worth considering
why 3(b) doesn't say "provide any third party with [the complete
corresponding source on request made within 3 years of
distribution...]"
I think the answer to this is quite subtle: the GPL is a distribution
licence, and, logically, it must be possible to determine whether the
condition is complied with at the time of distribution, otherwise the
condition is in a Schroedinger superposition of states, and it will
never be clear whether the condition was fulfilled or not at the time
of distribution, until a period of three years has elapsed (or a
request for the source code has been refused).
I suspect that RMS and Eben realised this, and drafted the condition to
avoid it. The condition is fulfilled by making the written offer, and
not by making the source available.
This has some interesting consequences. There is no guidance as to when
the source has to be provided. It is arguable...
(Hopefully formulating the question in this way does not make me one of
the horrible lawyers intent on circumventing the GPL but, if it does,
at least take comfort from the fact that I am not *your* horrible
lawyer :))
...(which puts me in danger of being one of Neil's 'horrible lawyers')
that an offer which says 'we will provide you with the source in 10
years' time' is compliant, although clearly completely against the
spirit of the GPL. However, assuming that the offer is 'We will provide
you with the complete corresponding source on request', and providing
that this is legally binding offer (lawyers will realise that there is,
in itself, a problem here), then courts would be likely to interpret
the lack of a time period as a 'reasonable time' (what else are they to
do, if the offer is binding).
The other consequences of this are that, although under the GPL itself,
assuming it is interpreted as a bare licence, there is no mechanism to
compel the release of the source (although failure to release under
3(a) clearly leads to a breach of copyright), if the written offer is
accepted and forms a binding contract, than it may be susceptible to
specific performance, so release of the source can be compelled under
that secondary contract.
It is also the case that failure to comply with the offer is not a
breach of the GPL (so enforceable by the licensor), but is a
contractual breach, enforceable by the recipient of the offer. This
does suggest a route for a 'horrible lawyer' to circumvent the GPL.
The licensor would have to demonstrate that the 'offer' was not a
'written offer, valid for at least three years...' that there would a
valid claim that the GPL had been breached. That's not necessarily
impossible, but it could be a very difficult argument to maintain.
All the best
Andrew

[Andrew Katz]

Hi Bruce

On 13 Mar 2014, at 18:33, Bruce Perens <***@perens.com<mailto:***@perens.com>> wrote:

Andy,

I think you are confusing the commission of an act and the revelation of evidence to others. If you consider it that way. there is no reason to treat it as a quantum physics experiment.

I don't believe an act can become retrospectively infringing. (In the spirit of full disclosure, I have come across a case in the UK where the courts disagree with me, but I still have problems with the logic). And I think that must be by RMS and Eben drafted the clause as they did. Certainly, I occasionally draft contracts where I say that 'A agrees to enter into an agreement with third Party B to do X' very deliberately, instead of saying 'A will procure that B does X'. I believe that legally they are very different things. In this case, because of the drafting, we don't have to consider the Schroedinger issue in any case, as it's possible to determine whether the condition has been fulfilled at the time the copying takes place.


If you fail to distribute source code on request, at what time does it become infringement?

Unfortunately, from the perspective of the GPL, never (this is only the case if you employ 3b, not 3a).


The answer is "at the moment that you distributed the binary".

Only if you didn't make the offer at the same time.



At the time you distributed the binary, you had a due diligence requirement to have the complete and corresponding source code on hand and ready for distribution, to have a structure in place to handle that distribution in a timely manner, and to retain that structure in place for three years. If you are unable to assure that at the time of distribution, it's an act of infringement. The only question is when you are caught.

The GPL doesn't explicitly require this, and the courts will only imply the minimum term to give business efficacy to the contract (I'm assuming they'd apply the same rule to the interpretation of a bare licence, but I'm not aware that this point has ever been considered in the English courts, at least). This goes way beyond that requirement - as no term need be implied. That the licensee has given a valid written offer is good enough. If the offer is accepted, there is a remedy - it belongs to the recipient of the code, not the licensor.


An embezzler does not become one when it is found that the books have been cooked. He is an embezzler the moment he diverts the funds. And of course many embezzlers tell themselves that they will put the money back, and then they fail to do so. He may have a change of heart, replace the funds, and never get caught. He may lose the money in Las Vegas and eventually be revealed. But the commission of the act is the original diversion rather than the later loss or replacement.

This is another point - and a very interesting one. If the offeror makes the offer with the intention that he is never going to fulfil it, he may be making a fraudulent representation to the offeree, and he may be committing some sort of offence, but I'm still struggling to see how the licensor has any claim. (Except in the sense that the licensor is also a person who has a right to the code under the offer, so can take up the licensee's offer, and then, if the licensee doesn't respond, seek an order for specific performance, and then release the modified code him/herself). So in a roundabout way, the licensee does have a remedy, but not for copyright infringement.

I'm going back to the original wording of GPL2 - the condition is to accompany the code with the offer, not to make the code available. Once the offer has been made, the condition is fulfilled. I'm even struggling to find extreme cases where court might deem the offer not to be validly made: it was have to be a case where the offeror cannot be bound. Even if the offeror makes the offer fraudulently, the offeree can void the subsequent agreement, but not the offeror. if the offeror is minor, or otherwise doesn't have legal capacity, then the offer wouldn't be valid (and, interestingly, if it's an offer to make the code available for free it also wouldn't be legally binding so arguably fails to fulfil the condition, although an offer made under seal or an offer to provide the source for a penny would be).

It's unfortunate, but I'm pretty adamant that under a strict legal interpretation, provided the offer is validly made, the GPL cannot be infringed for failure to provide the source. I'm trying hard to reach the conclusion we all want to reach, but I can't in any consistent way - not under English law (and I suspect US copyright law) at least.

Best



Andrew






Thanks

Bruce


On March 13, 2014 2:41:06 AM PDT, Andrew Katz <***@moorcrofts.com<mailto:***@moorcrofts.com>> wrote:

HI Neil

On 10 Mar 2014, at 21:09, Neil Brown <***@neilzone.co.uk<mailto:***@neilzone.co.uk><mailto:***@neilzone.co.uk>> wrote:


On 9 Mar 2014, at 19:06, Olliver Schinagl <oliver+***@schinagl.nl<mailto:oliver+***@schinagl.nl><mailto:oliver+***@schinagl.nl>> wrote:

How long of a delay would be acceptable?

As others have said, there is no position within GNU GPL 2.0 or 3.0 which permits a delay in the provision of the source code. However, perhaps, if your question were phrased slightly differently, it might elucidate some further interesting and useful comments? Perhaps:

A company develops, manufacturers and retails a product containing GNU GPL 2.0 covered code, and distribution of the binary is to be accompanied by a written offer conforming to section 3(b).

Someone chooses to take up the written offer, and asks the company for a copy of the source code.

How long need the delay be between someone asking fo
r the
code, and the company providing it, for that company to be considered non-compliant?

This is exactly the right way to analyse this. It's worth considering why 3(b) doesn't say "provide any third party with [the complete corresponding source on request made within 3 years of distribution...]"

I think the answer to this is quite subtle: the GPL is a distribution licence, and, logically, it must be possible to determine whether the condition is complied with at the time of distribution, otherwise the condition is in a Schroedinger superposition of states, and it will never be clear whether the condition was fulfilled or not at the time of distribution, until a period of three years has elapsed (or a request for the source code has been refused).

I suspect that RMS and Eben realised this, and drafted the condition to avoid it. The condition is fulfilled by making the written offer, and not by making the source available.

This has some
interesting consequences. There is no guidance as to when the source has to be provided. It is arguable...



(Hopefully formulating the question in this way does not make me one of the horrible lawyers intent on circumventing the GPL but, if it does, at least take comfort from the fact that I am not *your* horrible lawyer :))

...(which puts me in danger of being one of Neil's 'horrible lawyers') that an offer which says 'we will provide you with the source in 10 years' time' is compliant, although clearly completely against the spirit of the GPL. However, assuming that the offer is 'We will provide you with the complete corresponding source on request', and providing that this is legally binding offer (lawyers will realise that there is, in itself, a problem here), then courts would be likely to interpret the lack of a time period as a 'reasonable time' (what else are they to do, if the offer is binding).

The other consequences of th
is are
that, although under the GPL itself, assuming it is interpreted as a bare licence, there is no mechanism to compel the release of the source (although failure to release under 3(a) clearly leads to a breach of copyright), if the written offer is accepted and forms a binding contract, than it may be susceptible to specific performance, so release of the source can be compelled under that secondary contract.

It is also the case that failure to comply with the offer is not a breach of the GPL (so enforceable by the licensor), but is a contractual breach, enforceable by the recipient of the offer. This does suggest a route for a 'horrible lawyer' to circumvent the GPL.

The licensor would have to demonstrate that the 'offer' was not a 'written offer, valid for at least three years...' that there would a valid claim that the GPL had been breached. That's not necessarily impossible, but it could be a very difficult argument to maintain.

All the best<
br
/>

Andrew




--
Sent from my Android phone with K-9 Mail. Please excuse my brevity.

[Richard Fontana]

Post by Andrew Katz
I don't believe an act can become retrospectively infringing. (In the spirit of full disclosure, I have come across a case in the UK where the courts disagree with me, but I still have problems with the logic). And I think that must be by RMS and Eben drafted the clause as they did.

I haven't been following this discussion too closely but I believe it
concerns GPLv2. Just to correct the historical record, RMS and Eben
did not meet until several years after GPLv2 was drafted.

As FLOSS history buffs know, GPLv2 was drafted with the help of a
lawyer named Jerry Cohen, who still I believe does some legal work for
the FSF.

- RF

[Andrew Katz]

Richard

Post by Richard Fontana

Post by Andrew Katz
I don't believe an act can become retrospectively infringing. (In the spirit of full disclosure, I have come across a case in the UK where the courts disagree with me, but I still have problems with the logic). And I think that must be by RMS and Eben drafted the clause as they did.

I haven't been following this discussion too closely but I believe it
concerns GPLv2. Just to correct the historical record, RMS and Eben
did not meet until several years after GPLv2 was drafted.

Ah, thank you. I stand corrected.

[Bradley M. Kuhn]

Post by Andrew Katz
And I think that must be by RMS and Eben drafted the clause as they
did.

Just to correct the historical record, RMS and Eben did not meet until
several years after GPLv2 was drafted. ... GPLv2 was drafted with the
help of a lawyer named Jerry Cohen[.]

FWIW, just to add to this with another slight correction/addition to the
historical record on this point, I've seen and heard RMS publicly claim
in his post-GPLv3 speeches that he is *the* author of all versions of
the GPL, and that others involved were advisors, not co-authors.

RMS has also told me personally that during the GPLv3 process, he
considered lots of the proposals being offered to him, but vetted and/or
rewrote them himself before accepting them.

I have no reason to doubt RMS' claim of authorship and having exchanged
thousands of emails with RMS over the years, I attest that (IMO) GPLv2
and GPLv3 are both unmistakably written in RMS' writing style. I thus
think we should give RMS credit for being both the inventor of copyleft
and author of GPLv1-3 (as well as the Emacs Public License, etc. ;).

[Richard Fontana]

Post by Bradley M. Kuhn
FWIW, just to add to this with another slight correction/addition to the
historical record on this point, I've seen and heard RMS publicly claim
in his post-GPLv3 speeches that he is *the* author of all versions of
the GPL, and that others involved were advisors, not co-authors.
RMS has also told me personally that during the GPLv3 process, he
considered lots of the proposals being offered to him, but vetted and/or
rewrote them himself before accepting them.

I should respond to this since I've claimed to be a "co-author" of
GPLv3. I think both RMS's assertions about being "the author" is not
inconsistent with this; we're using different definitions of
"authorship".

It is absolutely the case that RMS had final authorial control over
the text in the GPLv3 license family, and that RMS (and the FSF)
conceived of RMS as having final authorial control. In that sense, RMS
was indeed *the* author of the GPLv3 license family. I have no reason
to think the same wasn't true for predecessor versions of the GNU
copyleft licenses -- probably much more so (leaving aside the evidence
of Cygnus influence over LGPLv2.0 -- and BTW Bradley wouldn't you say
that some of LGPLv2.0 doesn't really look like RMS's unmistakeable
writing style? :).

I have sometimes specifically said that I was the co-author of GPLv3
along with RMS and Eben. The reason I think of it that way is that the
substantial portion of the work that went into drafting GPLv3 from
December 2005 through June 2007 was done by the three of us (though
there were of course earlier nonpublic drafts that predated my joining
SFLC). I don't think of what I did, over a eighteen-month period, as
mere 'advisory' work. Indeed would describe the final period of GPLv3
drafting, which began essentially with a near-total rewrite (DD3
[released March 2007] to the final release]) as largely a drafting
collaboration between me and RMS. But no question, RMS had final say
on both text and policy. I want to also mention two people who played
a significant role in the drafting of the GPLv3 license family who are
often overlooked: David (Novalis) Turner and Brett Smith. I can point
to specific provisions in the license texts that reflect their ideas
and indeed wording. And of course Bradley you played an instrumental
role in the drafting of AGPLv3 section 13. :)

Post by Bradley M. Kuhn
I have no reason to doubt RMS' claim of authorship and having exchanged
thousands of emails with RMS over the years, I attest that (IMO) GPLv2
and GPLv3 are both unmistakably written in RMS' writing style.

I think if you do a really careful analysis of the GPLv3 text you'll
see some stylistic variations across sections or even parts of
sections. But RMS certainly in general made it be 'his voice', and I
believe that in the work I did I was trying to emulate that voice
(perhaps giving it a bit of lawyerly flair).

Since you bring the issue up, Bradley, I will note that I've found
over the years some lawyers claiming to have been involved in the
drafting of GPLv3 who I can confirm never were there... odd
phenomenon. :)

- RF

[Bradley M. Kuhn]

Post by Richard Fontana
Since you bring the issue up, Bradley,

Actually, you brought up the issue first. ;)

Post by Richard Fontana
I will note that I've found over the years some lawyers claiming to have
been involved in the drafting of GPLv3 who I can confirm never were
there... odd phenomenon. :)

I find it odd as well.

For my part, I've publicly and clearly said that I (much to my chagrin)
was *not* involved in the drafting of GPLv3 (except for some influence
in discussions with RMS when it was still called GPLv2.2 back circa
2003). Because of a certain comedy of professional errors, my main role
in the GPLv3 process ended up being the primarily technical task of
making sure the public commenting system worked.

Admittedly, I did have a seat on both Committee A and Committee D (two
of the four "discussion committees"), which I suppose was unprecedented
for anyone who wasn't involved in drafting.

I did have some influence over the drafting of Affero GPLv3, in part because
I had originally been the person to invent the idea behind the Affero clause,
and because the main thing I tried (unsuccessfully) to influence during GPLv3
drafting was the original plan to include the Affero clause in GPLv3 itself
wasn't abandoned. When that battle was lost, I and Henry Poole both urged
the FSF to at least publish an Affero variant of GPLv3, and FSF agreed.

But, even there: the 10 words or so in AGPLv3§13 that I helped write
were mostly rewritten by RMS. :) I can probably take credit for the word
"opportunity" appearing there, which ISTR suggesting. :)

Anyway, I say again that the GPL family of licenses is in my view primarily
RMS' work. As I said, he deserves the credit for this. But, success has a
thousand parents. Only failure is an orphan. :)

-- bkuhn

[Richard Fontana]

Post by Bradley M. Kuhn
For my part, I've publicly and clearly said that I (much to my chagrin)
was *not* involved in the drafting of GPLv3 (except for some influence
in discussions with RMS when it was still called GPLv2.2 back circa
2003).

Right.

Oh, on the issue of RMS's authorial control, I just remembered that
the official (head) draft of GPLv3 during the SFLC era, apart from the
public releases, was always conceptually 'the one stored in RMS's
laptop', which I think is telling as to who was 'the author'. We made
a halfhearted attempt to use subversion at SFLC (and as I recall you
really pushed for this) but for the most part we really had no system
of version control apart from emailing RMS. :)

Post by Bradley M. Kuhn
I did have some influence over the drafting of Affero GPLv3, in part because
I had originally been the person to invent the idea behind the Affero clause,
and because the main thing I tried (unsuccessfully) to influence during GPLv3
drafting was the original plan to include the Affero clause in GPLv3 itself
wasn't abandoned. When that battle was lost, I and Henry Poole both urged
the FSF to at least publish an Affero variant of GPLv3, and FSF agreed.

Though I take the earlier credit for the idea of an Affero variant as
a replacement for the notion of an open set of GPLv3-compatible
Affero-like terms, which was being criticized from left, center and
right. I've heard you say it was never what you expected -- you
anticipated that GPLv3 would incorporate something like AGPLv1 2d. The
legal form in which the GNU AGPLv3 then took -- once RMS agreed to
this idea -- was something you indeed had some influence on.

Post by Bradley M. Kuhn
But, even there: the 10 words or so in AGPLv3§13 that I helped write
were mostly rewritten by RMS. :) I can probably take credit for the word
"opportunity" appearing there, which ISTR suggesting. :)

That section was also one of the ones that Brett Smith had a certain
amount of textual influence over, as to its final version.

- RF

[Robinson Tryon]

Post by Bradley M. Kuhn

Post by Andrew Katz
And I think that must be by RMS and Eben drafted the clause as they
did.

Just to correct the historical record, RMS and Eben did not meet until
several years after GPLv2 was drafted. ... GPLv2 was drafted with the
help of a lawyer named Jerry Cohen[.]

FWIW, just to add to this with another slight correction/addition to the
historical record on this point, I've seen and heard RMS publicly claim
in his post-GPLv3 speeches that he is *the* author of all versions of
the GPL, and that others involved were advisors, not co-authors.

The impression I got at the time was that there were multiple people
working on drafting (i.e. writing parts of) the license:

http://gplv3.fsf.org/original-process-definition
---
The GPL revision comment process is a matter of information sharing.
Below, in sections 3 and 4.1, we lay out ways that the community as a
whole will tell version 3 drafters of issues with the current
license...[the process is] designed so that...the drafters at FSF can
directly address the community and present the reasoning behind
changes.
---

Post by Bradley M. Kuhn
RMS has also told me personally that during the GPLv3 process, he
considered lots of the proposals being offered to him, but vetted and/or
rewrote them himself before accepting them.

Vetting and rewriting proffered content sounds more like the role of
an editor than an author.

Post by Bradley M. Kuhn
I have no reason to doubt RMS' claim of authorship and having exchanged
thousands of emails with RMS over the years, I attest that (IMO) GPLv2
and GPLv3 are both unmistakably written in RMS' writing style. I thus
think we should give RMS credit for being both the inventor of copyleft
and author of GPLv1-3 (as well as the Emacs Public License, etc. ;).

Based on my (albeit quite limited) understanding of the situation, I
would be generally inclined to refer to RMS as the 'primary author' or
'chief architect' of GPLv3. Not that any such title should diminish
his contributions to the process -- indeed, working collaboratively in
the open can be much more challenging than laboring in private and
alone! :-)

Cheers,
--R

[Richard Fontana]

Post by Robinson Tryon
The impression I got at the time was that there were multiple people
http://gplv3.fsf.org/original-process-definition

This document was released a couple of months before the public phase
of the GPLv3 drafting process started, so it has a certain
aspirational quality to it.

Post by Robinson Tryon
---
The GPL revision comment process is a matter of information sharing.
Below, in sections 3 and 4.1, we lay out ways that the community as a
whole will tell version 3 drafters of issues with the current
license...[the process is] designed so that...the drafters at FSF can
directly address the community and present the reasoning behind
changes.
---

Based on remarks by Eben Moglen given at a panel discussion at OSBC
East 2005 which I attended (shortly before I was hired and joined
SFLC), I believe the FSF at this time thought that the GPLv3 drafts
would elicit tens (hundreds?) of thousands of community commentary
about the license. I'm afraid that is not what happened, though the
degree of public commentary received was unprecedented for a software
license.

There were more like a thousand or so comments on the draft, and the
FSF and its counsel made a good effort to respond to them, and some of
them had a material influence over the draft. The FSF also formed a
set of four (or, according to some more dubious versions of the
history, five) committees (loosely representative of free software
interest groups) to nominally aid in the drafting process. These
committees, two of them in particular which were dominated by vendor
and commercial end-user lawyers, did have a notable degree of
influence over a few significant parts of the draft at the level of
legal critique and, to a lesser degree, policy.[1] However, for the
most part they were not responsible for *authoring* anything, the way
I would understand 'author'. I can point to a few sentences here and
there that originated textually with some members of these committees,
but this was not typical.

In any case, this is not inconsistent with Bradley's statement. GPLv3
was never envisioned as a license that would be collectively authored
by hundreds, let alone tens of thousands, of people, and it
wasn't. Indeed I sometimes hear people say that GPLv3 suffers from
having the appearance of having been 'drafted by committee' (as though
it were like certain standards specs) but that isn't accurate either.

Post by Robinson Tryon

Post by Bradley M. Kuhn
RMS has also told me personally that during the GPLv3 process, he
considered lots of the proposals being offered to him, but vetted and/or
rewrote them himself before accepting them.

Vetting and rewriting proffered content sounds more like the role of
an editor than an author.

RMS's role was certainly more than that of a mere editor, I can assure
you. As I noted, he certainly saw himself as 'final author'. But I
don't quite agree with the impression Bradley's remarks give. RMS did
delegate much of the actual drafting -- after late 2005 that is -- to
SFLC, specifically Eben and me. But he was ultimately responsible for
all basic matters of policy reflected in the current text, that is for
certain. Nothing we wrote ended up in the license without his
approval, and this involved discussion and debate and rewording by RMS
in many cases.

Post by Robinson Tryon
Based on my (albeit quite limited) understanding of the situation, I
would be generally inclined to refer to RMS as the 'primary author' or
'chief architect' of GPLv3. Not that any such title should diminish
his contributions to the process -- indeed, working collaboratively in
the open can be much more challenging than laboring in private and
alone! :-)

I'm sure it is historically accurate to say that GPLv3 was much more
of a collective effort than earlier versions of the GPL (though I
believe the FSF attempted to gather and incorporate community feedback
into GPLv1 and GPLv2). I have to also say that it is a myth that GPLv3
was developed 'collaboratively in the open' -- though it may have
approximated this more than any predecessor free software license. The
most productive period of drafting involved the period between July
2007 and March 2007 when the FSF was largely silent with respect to
the general public regarding the status of the draft.

- Richard

[1] It is interesting to look back on the reasons those lawyers got
involved at all, because it may seem so bizarre today. Bradley put it
well in an episode of Free as in Freedom: paraphrasing, lots of
commercial lawyers were under a naive 'mass delusion' (I think that's
the phrase Bradley used) that the entire free software ecosystem would
be migrating to GPLv3, making the commercial stakes extraordinarily
high -- right around the zenith of what some now call the 'open source
bubble' of the mid-2000s.

[Neil Brown]

If you fail to distribute source code

To me, this is the nub of the issue: if someone asks for the code, how long does the binary distributor have to provide the source before he is considered non-compliant? At what point does someone "fail to distribute" the source code? A day after being asked? A week? A month?

(If, indeed, failure to live up to one's written offer is indeed non-compliance; on the wording of the licence, I'm not convinced it is, although clearly it would be based based on the intention.)


Best wishes


Neil


Neil Brown
***@neilzone.co.uk | http://neilzone.co.uk

[Thomas Charron]

Post by Bruce Perens
Andy,
I think you are confusing the commission of an act and the revelation of
evidence to others. If you consider it that way. there is no reason to
treat it as a quantum physics experiment.
If you fail to distribute source code on request, at what time does it become infringement?
The answer is "at the moment that you distributed the binary".

At what moment do you 'fail' to distribute it?

Post by Bruce Perens
At the time you distributed the binary, you had a due diligence
requirement to have the complete and corresponding source code on hand and
ready for distribution, to have a structure in place to handle that
distribution in a timely manner, and to retain that structure in place for
three years. If you are unable to assure that at the time of distribution,
it's an act of infringement. The only question is when you are caught.

Actually, you do NOT have a due diligence requirement. You are inserting
requirements which are NOT present. You are at a degree of risk if you do
not, certainly in some cases possibly rather large risk, but you have not
broken license if you fully intend on fulfilling the terms at the time a
request is made to take you up on the offer.

Post by Bruce Perens
An embezzler does not become one when it is found that the books have been
cooked. He is an embezzler the moment he diverts the funds. And of course
many embezzlers tell themselves that they will put the money back, and then
they fail to do so. He may have a change of heart, replace the funds, and
never get caught. He may lose the money in Las Vegas and eventually be
revealed. But the commission of the act is the original diversion rather
than the later loss or replacement.

An embezzler is actively doing something. Your saying guilty until
proven innocent, and I go back to my original examples. Motorola Mobility
has provided sources after sanitizing and verifying source trees only
*after* a request for the given sources was made. It is your opinion that
since they failed to produce IMMEDIATELY on request, but spent several
weeks producing the code after request, they they have stolen the Linux
kernel and various GPL components, and have done a crime on par with
embezzling funds for the Columbian Drug Cartel.