Thijs (M.A.) van den Berg writes:

...

I suggest we should drop "Copyright Joe Coder 20xx" from all our

files. My

...

reasons:

You are asking authors to give up copyright.

No, I am not. If anything I suggest to attribute copyright properly.

There is no "we" that can decide that.

"We" are boost developers.

...

* I am never clear what this line really mean. Copyright Joe Coder 2010 - does this mean Joe has copyright up to 2010, starting 2010 or only in 2010? Should one keep updating these disclaimers every January?

Shouldn't you first learn what it means before forming an opinion to change it?

Does it matter? My point is that it is confusing in general for anyone looking at this line. Gennadiy

I think removing the copyright notice would not deprive anyone of their rights, and probably is an allowed modification under the Boost Software License (whereas removing the BSL notice is not).

And it looks like I'm wrong here - the BSL explicitly requires the copyright notices to be retained (as Jonathan correctly pointed out). I think this is pretty much it. Cheers, Filip

Thijs (M.A.) van den Berg writes:

On 15 Sep 2015, at 15:51, Filip Konvička

wrote:

...

...

I think removing the copyright notice would not deprive anyone of

their

...

...

rights, and probably is an allowed modification under the Boost Software License (whereas removing the BSL notice is not).

And it looks like I'm wrong here - the BSL explicitly requires the copyright notices to be retained (as Jonathan correctly pointed out).

I think this is pretty much it.

The copyright holder is the one that (can) release the source under various licenses, in this case the boost license.

Without a copyright statement you would not know who released the code under the boost license, and it that

And with this source code comment you would? What does it say about who released this code? I do not release the code - boost community does. Does that mean we all hold copyright to everything?

case the boost license might be invalid. That's the reason that the boost license requires a copyright statement, so that people can find the copyright holder and check if he indeed released the code under the boost license.

So according to you I can make a copy of boost, make a change to all source files claiming I authored them in 1995 and release it? And that would make me a copyright holder? Gennadiy

On 15.09.2015 17:06, Jonathan Wakely wrote:

On 15 September 2015 at 09:32, Filip Konvička wrote:

...

I think removing the copyright notice would not deprive anyone of their rights, and probably is an allowed modification under the Boost Software License (whereas removing the BSL notice is not).

I disagree, the BSL specifically mentions the copyright notices must be present:

"The copyright notices in the Software and this entire statement, including the above license grant, this restriction and the following disclaimer, must be included in all copies of the Software, in whole or in part, and all derivative works of the Software, unless such copies or derivative works are solely in the form of machine-executable object code generated by a source language processor."

It doesn't say they must be in the source code itself, but if you separated them how could you ensure the notices appear in derivative works, such as a subset of Boost headers copied into someone else's project?

AFAIU, retaining copyright notices is responsibility of the one who creates the derivative work. So technically it is enough to have a separate file(s) with the license and copyright holders bundled together with the source code.

Andrey Semashev writes:

On 15.09.2015 08:45, Gennadiy Rozental wrote:

...

Hi,

I suggest we should drop "Copyright Joe Coder 20xx" from all our files. My reasons:

* git/svn history provides much better copyright attribution. Each developer has a copyright on lines authored. As it is these copyright disclaimers are very frequently wrong.

There is no VCS history in the distributed packages. Also, I'm not a lawyer, but I'm not sure VCS history counts as a legal indication of authorship.

And arbitrary number I put in a source file body is? I'll put 1900 there and will claim that bubble sort implemented in my file is authored by me. How about that?

I'll add that personally I feel that history is something that can be lost relatively painlessly (e.g. due to transition to a different VCS or github being destroyed or whatever other reason). All relevant information should still be available in the source code.

That would require to comment on pretty much each line of code, claiming the copyright.

...

* For libraries which are maintained by multiple developers and/or which were moved from one maintainer to another maintainer - these disclaimers either wrong or pain to maintain.

I see no problem with that. Don't remove anyone from the copyright notice, add yourself only when your contribution is large enout (at your best judgement).

And? What do I get copyright to (Assuming one does not have access to VCS)?

Peter Dimov writes:

Gennadiy Rozental wrote:

...

* I am never clear what this line really mean. Copyright Joe Coder 2010 - does this mean Joe has copyright up to 2010, starting 2010 or only in 2010? Should one keep updating these disclaimers every January?

"Copyright 2010 Joe Coder" - Joe Coder has authored the file in 2010.

And? What If rewrote it completely in 2013? Do I still hold copyright to the new code starting with 2010?

"Copyright 2010, 2012 Joe Coder" - authored some parts in 2010, others in 2012.

Which part is 2010 and which is 2012?

"Copyright 2010-2013 Joe Coder" - has worked on the file from 2010 to 2013 without long interruptions.

So, does my copyright expired in 2013? If it expired in some number of years, wh do I need first number? (or second, depending on your point of view.

Paul A. Bristow writes:

Sorry but it ain't broke, so let's not fix it.

Sorry, but it is. The copyright information you type in your source code starting to lie practically immediately you commit the code. It does not hold the water in any format proceedings, lying to reader, and annoying to maintain to author. It does not says what specifically one has a copyright to and is wrong practically everywhere in boost. What does these lines indicate: // Copyright (c) Marshall Clow 2008-2012. or these // Copyright (c) 2002 Peter Dimov and Multi Media Ltd. // Copyright (c) 2008 Peter Dimov or these // Copyright Christopher Kormanyos 2002 - 2013. // Copyright 2011 -2013 John Maddock. or these // Copyright Beman Dawes, 2002-2005 // Copyright Vladimir Prus, 2002 These says nothing about who has the copyright to what, when it is expires, what happen after last year mentioned, does all source in the file was originated in first year mentioned and so on. Look at Google released HBase for example. And I can assure you Google has lots lots of lawyers. In other projects I see "Copyright xxxx <project> authors" disclaimer. Even this is unnecessary, IMO. It is fine probably to keep the information in one place somewhere in documentation about who worked on the library, but it not a source of copyright anyway. Gennadiy

On 2015-09-15 18:34, Gennadiy Rozental wrote:

Paul A. Bristow writes:

...

Sorry but it ain't broke, so let's not fix it.

Sorry, but it is. The copyright information you type in your source code starting to lie practically immediately you commit the code. It does not hold the water in any format proceedings, lying to reader, and annoying to maintain to author. It does not says what specifically one has a copyright to and is wrong practically everywhere in boost. What does these lines indicate:

// Copyright (c) Marshall Clow 2008-2012.

or these

// Copyright (c) 2002 Peter Dimov and Multi Media Ltd. // Copyright (c) 2008 Peter Dimov

or these

// Copyright Christopher Kormanyos 2002 - 2013. // Copyright 2011 -2013 John Maddock.

or these

// Copyright Beman Dawes, 2002-2005 // Copyright Vladimir Prus, 2002

These says nothing about who has the copyright to what, when it is expires, what happen after last year mentioned, does all source in the file was originated in first year mentioned and so on.

It does actually affect the expiry date. The date of first publication is important in some places, like the US, where the copyright laws were changed several times during the 1900's. If published after 1978, the copyright is valid for the author's lifetime + 70 years. Earlier it was a certain number of years. Therefore it is important to claim that the copyright period started after 1978. The exact year is not important (unless the laws are changed again. Who knows?) The names of the authors are just a service to you, when you have to check if they have all been dead for 70+ years. Unlikely for software right now, but important for music and books. Bo Persson

On 16/09/2015 04:34, Gennadiy Rozental wrote:

Sorry, but it is. The copyright information you type in your source code starting to lie practically immediately you commit the code. It does not hold the water in any format proceedings, lying to reader, and annoying to maintain to author. It does not says what specifically one has a copyright to and is wrong practically everywhere in boost.

No, it's not lying in the slightest. The important thing is who the *original author* of the work was (in this case, the individual source file). This is the person in the copyright notice and the year in which they created this file. (As others have noted, in most cases the specific year is not that interesting other than to indicate that it's after a certain date, since copyright tends to be tied to the author's lifetime rather than the origin date.) That original author has then chosen to license the work by the BSL, which in turn gives *other* authors the right to effectively "borrow" the original author's copyright and distribute the work (the file) and make derivative works based on it. The derivative works are still copyright by the original author and date unless both parties agree that the changes are substantive enough to warrant an additional or replacement notice. It is perfectly valid for the copyright notice to never be updated, even if the file is changed substantially. It is *not* valid for the original notice to be modified without specific agreement by the original person. It is *not* valid to remove it entirely, as then you do not know that the license grant is valid or who made it. It is *not* valid to specify a non-legal-entity as the copyright holder (so you can't say "lib X contributors").

. Unless

this changes, I am inclined to change disclaimer everywhere to "Copyright 2001 Boost.Test authors" and be done with that.

That would be illegal IMO but you would need to check your lawyer. The license and copyright statement impose rules about what you can legally do with files released under the boost license by authors.

Le 15/09/15 20:14, Andrey Semashev a écrit :

On 15.09.2015 19:46, Gennadiy Rozental wrote:

...

Rene Rivera writes:

...

I suggest you stop this thread as it's not going to help anyone. The only way to answer these questions to any satisfaction is to retain legal counsel. Which is what we did to get the license and copyright guidelines we have now. If you want to change the guidelines it might be more beneficial to bring up the issues with the steering committee and the SFC.

Well, I am bring it up. Steering committee is free to discuss, respond or ignore. As it stands my source code either lying almost all the time, or missing copyright or I need to keep updating 700+ file all the times. Unless this changes, I am inclined to change disclaimer everywhere to "Copyright 2001 Boost.Test authors" and be done with that.

I suppose, you could do that, if all current copyright holders agree to that in written form and the "Boost.Test authors" is either a list of persons documented somewhere and bundled with the library, or a legal body that owns the copyright. Otherwise I feel such a change would be illegal and not fair in a common sense.

Hi all, This is exactly what I did. I sent an email to Gennadiy (written form) that says that I give up all my copy rights and transfer them to boost.test development team (or something similar).

In any case, I wouldn't advise such changes without consulting a lawyer. Take note that authors live in different countries and Boost is an international project, so you'll have to abide the international law.

However, I find this thread far from being "unuseful", and since this is about open source and freedom, I would say that delegating the answer or the responsibility to a lawyer is not really an (affordable) option for most of us. Here are some resources I found on the Internet, that seem to emanate from reasonably good/serious sources or ppl. Some might argue that only a lawyer might state or commit on those matters. This is just not true: as pointed out already, those copyright matters should be covered in every country in the same way, which I doubt would ever happen. Also we are fully in a gray area, so I do not expect some lawyer would ever commit onto this. Now (short) the list: - http://softwarefreedom.org/resources/2008/foss-primer.html#x1-110002.3 from what I understand here, it is perfectly plausible to ask to contributors to give away their copyrights when they contribute to a project. - http://softwarefreedom.org/resources/2012/ManagingCopyrightInformation.html * DVCS is de facto an excellent tool for managing copyrights in case of competing claims, as the authorship in source file becomes difficult to track (authorship is copyright holder unless otherwise specified). * footnote 3/ copyright year, under the US law: "2) the year of the work’s first publication". - http://producingoss.com/en/index.html case studies on big projects. They are plenty resources on the Internet, sometimes giving contradictory information. It's also a pity that all those questions are not crystal clear for most of us. But it just tells me that no one nor organization is taking any commitment onto this, leaving the open source software developers with all the responsibilities. Best, Raffi

John Maddock writes:

...

Well, I am bring it up. Steering committee is free to discuss, respond

or

...

ignore. As it stands my source code either lying almost all the time, or missing copyright or I need to keep updating 700+ file all the times. Unless this changes, I am inclined to change disclaimer everywhere to "Copyright 2001 Boost.Test authors" and be done with that. That's explicitly against the terms of the BSL, if you did do that, expect ructions.

Why is that? Can you please point me specifically where does it say that? Line above at least is going to be "almost" true. Me (starting 2001-ish) and Raffi (starting 2013) are Boost.Test authors and we both agree to assign the copyright to the "team". We collectively started to work on this code in 2001. Code migrated over time from one file to another, so file creating time in irrelevant.

...

Each library has README file. It should be enough to just have Copyright in that one file IMO. And it won't be a real copyright attribution, but more like acknowledgement: these guys worked on this code.

Which part of "Boost took legal advice" when setting this up did you not understand?

I am not quite sure why you are taking this personally. As I mention this kind of copyright is used by Google (as well as complete absence of any Copyright disclaimers - only Apache licence for HBase project) and I would believe their lawyers more than anyone we can afford to consult. For me this is practical issue. I am not trying to make anyone to give up their copyright. Gennadiy

Peter Dimov writes:

Gennadiy Rozental wrote:

...

Unless this changes, I am inclined to change disclaimer everywhere to "Copyright 2001 Boost.Test authors" and be done with that.

You can in principle ADD

Copyright 2015 Boost.Test contributors

to each file and then update it to reflect the current year, so it

becomes

Copyright 2015-2017 Boost.Test contributors

Can you tell me why you yourself do not follow this advice? I do not see you updating copyright statements. Let's take any file with your copyright (do not take this personally - it is just an example). For example interlocked.hpp: You have Copyright 2005 Peter Dimov. This supposedly says that you started to work on this file in 2005 (which is indeed true) and you hold the copyright on content of this file ever since, right? Few questions: 1. Why is it not "Copyright 2005-2015 Peter Dimov." 2. I see quite a lot of changes made since 2005. Does your disclaimer automatically make you hold a copyright to these lines since 2005 even though changes were made let's say 2012? 3. I see few other people made significant changes to this file. Why don't we have a Copyright disclaimers for all of them?

in 2017, but you can't remove the existing lines.

Why? 1. Why I can't change my own copyright to this? 2. Why I can't change Raffi's copyright to this if he formally agrees to this? 3. Let's say someone had a copyright line introduced few years ago. And I removed lined which were authored by that person. Do I need to keep copyright? What is this Copyright to? Gennadiy

On Wednesday 16 September 2015 00:21:30 Peter Dimov wrote:

Gennadiy Rozental wrote:

...

Can you tell me why you yourself do not follow this advice? I do not see you updating copyright statements.

I do update them from time to time. :-)

...

Let's take any file with your copyright (do not take this personally - it is just an example). For example interlocked.hpp:

You have Copyright 2005 Peter Dimov.

This supposedly says that you started to work on this file in 2005 (which is indeed true) and you hold the copyright on content of this file ever since, right?

Few questions:

1. Why is it not "Copyright 2005-2015 Peter Dimov." 2. I see quite a lot of changes made since 2005. Does your disclaimer automatically make you hold a copyright to these lines since 2005 even though changes were made let's say 2012?

I automatically hold the copyright for everything I author, regardless of whether I include a proper copyright assignment at the top, but it depends on the jurisdiction and having the lines at the top is considered a good practice. That's my opinion though and I'm not a lawyer.

...

3. I see few other people made significant changes to this file. Why don't we have a Copyright disclaimers for all of them?

It's up to those people to add a line if they so decide. Copyright only covers expressions of creativity. If someone makes changes to a file but doesn't feel that these changes are original enough, he can decide to not claim copyright. Sometimes people err on the side of not claiming copyright out of modesty. Sometimes other people claim copyright for changes that deserve none.

...

...

in 2017, but you can't remove the existing lines.

Why?

Well you obviously can do whatever you like, it's just not considered good practice.

...

1. Why I can't change my own copyright to this?

You can in principle assign your copyright to some other person or a legal entity, in which case, you could legitimately go and change everything to match, but why would you do that?

...

2. Why I can't change Raffi's copyright to this if he formally agrees to this?

You could, but why?

...

3. Let's say someone had a copyright line introduced few years ago. And I removed lined which were authored by that person. Do I need to keep copyright?

Not in principle, but this is a gray area because one might argue that editions made later have been influenced by the creative ideas embodied in that person's lines and so constituting a derived work.

There is one other thing to consider. If one day, say in 2017, Boost.Test gets a new active contributor John Doe, how would you attribute work of his in the copyright notices? The current set of Boost.Test contributors is Gennadiy and Raffi, and by 2017 you'll have "Copyright 2015-2017 Boost.Test contributors" all over the code. Would you silently include John into this set and somehow attribute the prior work to him? And what if John refuses to be included? I think Peter made a very good point that these "Boost.Test contributors" has to be a legal entity, and every contributor has to explicitly give up his rights to this entity. Usually this is done by signing a CLA. To my knowledge, neither Boost nor Boost.Test is governed by such entity, so there's no body you could release your rights to. But frankly, even if there was, signing a CLA is starting to smell fishy to me. This reminds me. Recently I updated README.md in a few libraries I work on to include a notice that patch contributors implicitly agree to release their modifications under BSL [1][2][3]. Should these notes be added to all libraries or even on the main website or am I being paranoid? [1]: https://github.com/boostorg/core [2]: https://github.com/boostorg/log [3]: https://github.com/boostorg/atomic

On Wednesday 16 September 2015 01:35:00 Peter Dimov wrote:

Andrey Semashev wrote:

...

I think Peter made a very good point that these "Boost.Test contributors" has to be a legal entity, ...

No, I'm not saying that. It's reasonable to interpret "Boost.Test contributors" as a shorthand for a list of specific names who don't need to constitute a legal entity. It's basically a 'pointer' to the actual list of contributors that is presumably available on demand.

In that case I'm not sure how my John Doe example works. Does he receive the copyright over the code he did not submit by being included into the list of contributors?

Rene Rivera writes:

I suggest you stop this thread as it's not going to help anyone. The only way to answer these questions to any satisfaction is to retain legal counsel. Which is what we did to get the license and copyright guidelines we have now. If you want to change the guidelines it might be more beneficial to bring up the issues with the steering committee and the

SFC. Yesterday I contacted Daniel Berlin, who has a long track record on OS copyright issues and who is responsible for setting Google's policy on copyrighting Google's OS projects. Here is a very short summary of his reply: Copyright notices in source code are legally worthless. Here are excerpts from his reply in a form of my questions and his answers. Q: Do lawyers recommend putting copyright notices in source code? A: No, they don't. Or at least, not smart lawyers. I agree plenty of lawyers cargo cult this, but anyone with a brain realizes it does nothing. Q: Doe Joe has a copyright on the file if he adds "Copyright Joe Coder 2010" on top? A: No. It's just a notice. if others own stuff in the file, it just makes the notice wrong. Q: What about changes made since 2010? What about changes made by other people? A: Copyright always starts at creation time of the part written by whoever. So if multiple people contribute, each person owns copyright on the parts they contribute, and that copyright started when they wrote that part. It doesn't matter what they put in the notice. Q: If two people has copyright on top of the file who has copyright to what? Can I change notice introduced by someone else? A: Again, copyright is separate and independent from the notice. Changing the notice has no effect on legal ownership. The notice is mostly meaningless, and hearkens back to a time when the notice was required to maintain a copyright. That has not been true in 198 countries since 1988. Notices have not been legally required to maintain copyright since then, and in fact, have basically no legal effect. Q: In formal proceedings will anyone ever look at these? A: The only case these notices *ever* come into play is in the following situation: 1. You have no notice, or your notice is wrong 2. Someone buys a license from another person, *reasonably* relying on a wrong notice to believe that person had the right to license them the work. 3. You sue that person for copyright infringement #2 is called "innocent infringement". It is the only thing the notice prevents. The person committing innocent infringement must A. Actually acquire a license to use the software from somebody they B. *reasonably* believe has the right to license them the software because C. The notice is wrong. The likelihood of this ever happening is basically zero (and in fact, there are zero cases of innocent infringement i can find in the US in the past 10 years) Given that, and how much time people waste on copyright notices, we chose to get rid of them rather than waste time on them. Q: Why Google's OS software has "Copyright <project> authors 20xx" in source code? We only do this because most of the licenses refer to copyright notices, and when we don't have them, people get confused. We otherwise do not want them or care about them. So, here you go. Feel free to consider this when setting the policy. Gennadiy