taskSavannah Administration - Tasks: task #14528, Submission of relax

 
 

task #14528: Submission of relax

Submitted by:  Edward d'Auvergne <bugman>
Submitted on:  Mon 29 May 2017 09:28:33 AM UTC  
 
Should Start On:  Sun 28 May 2017 12:00:00 AM UTC Should be Finished on:  Wed 01 May 2019 12:00:00 AM UTC
Category:  Project Approval Priority:  5 - Normal
Status:  In Progress Privacy:  Public
Percent Complete:  0% Assigned to:  Ineiev <ineiev>
Open/Closed:  Open Effort:  0.00

Add a New Comment(Rich Markup)
   

You are not logged in

Please log in, so followups can be emailed to you.

 

( Jump to the original submission)

Sun 04 Nov 2018 11:14:33 AM UTC, comment #42:

> Take for example a "user" who deposits a 3D structure but does not have the authority to make the work public domain - i.e. they are not the copyright holder. Then the "user" and not the Protein Data Bank is legally responsible for the copyright violation. Again this is a legal shield for the PDB.


If this means that the real copyright holder of those data may sue users of your package for unauthorized distribution of their data, then I believe this is an issue with your package. You should avoid the possibility of such outcome.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Tue 30 Oct 2018 12:27:54 PM UTC, comment #41:

The statement says:

"Data files contained in the PDB archive (ftp://ftp.wwpdb.org) are free of all copyright restrictions..."

There are no restrictions or clauses on this. The text "free of all copyright restrictions" means there is no copyright.

Note that you cannot patent a 3D structure by itself (at least in the US or Europe). You can only have patent claims on its application for specific uses. So the absence of copyright and the non-patentability of 3D structures by themselves is equivalent to all contents of the PDB being public domain.

As for the RCSB PDB IP advisory notice, this is an obvious legal shield for the RCSB PDB organisation that covers patents, copyright, or any other IP claims:

Patents: Certain proteins, small molecules (e.g. medicines), DNA/RNA, and other macromolecules are patented for specific uses in some countries. The advisory notice shields the PDB against patent trolls, in that the "user" who submitted the 3D structure is legally responsible for defending against any patent claims. "Troll" is the appropriate term here as you cannot claim the 3D structure itself in a patent.

Copyright: Take for example a "user" who deposits a 3D structure but does not have the authority to make the work public domain - i.e. they are not the copyright holder. Then the "user" and not the Protein Data Bank is legally responsible for the copyright violation. Again this is a legal shield for the PDB. An example of this situation might be a PhD student determining the 3D structure, and the student's professor submitting the structure after the student has graduated, without the student agreeing to their 3D structure file being made public domain.

I do not see how this IP advisory notice invalidates the public domain nature of the entire Protein Data Bank. I would assume that is why the RCSB who maintain the PDB have the following Full Privacy Statement:

"RCSB PDB operates an open-access portal for public domain information about the 3D shapes of proteins, nucleic acids, and complex assemblies that helps researchers, educators, and students understand fundamental biology, biomedicine, and bioenergy. As a member of the Worldwide Protein Data Bank partnership (wwPDB; wwpdb.org), RCSB PDB validates and biocurates data deposited into the global PDB archive."

The emphasis is mine to highlight that the contents of this database is officially labelled as being "public domain" by the RCSB. Looking at more recent wwPDB parent organisation, the mission statement on the main webpage states:

"Ensure universal open access to public domain structural biology data with no limitations on usage."

This is repeated on their FAQ. And from the PDB newsletter number 2 from April 1999:

"The contents of PDB are in the public domain, but it is expected that the authors of an entry as well as the PDB be properly cited whenever their work is referred to. Structures used from the PDB should be cited with the PDB id and the JRNL reference."

Again the emphasis is mine. "Expected" here is not a legal requirement that invalidates the "public domain" nature of the 3D structure files, but is common courtesy for scientists standing on the shoulders of others. Similar statements are repeated in a number of the PDB newsletters (e.g. number 30 with the text "The contents of the RCSB PDB are in the public domain.", or number 33 with the same text).

I am not a legal expert. But the contents of the PDB database really are, and always have been, public domain.

Edward d'Auvergne <bugman>
Mon 29 Oct 2018 03:55:56 PM UTC, comment #40:

Yes, I think that was an improvement (I'd rewrapped the text to something like 70 characters per line or less, but that isn't crucial).

Now, https://www.rcsb.org/pdb/static.do?p=general_information/about_pdb/policies_references.html doesn't say the data are exactly public domain, it says that the original authors should be attributed. Further, the Advisory Notice says something controversial:

"The user assumes all responsibility for insuring that intellectual property claims associated with any data set deposited in the PDB archive are honored. It should be understood that the PDB data files do not contain any information on intellectual property claims with the exception in some cases of a reference for a patent involving the structure."

I'm not sure how this matches the statement that "Data files contained in the PDB archive (ftp://ftp.wwpdb.org) are free of all copyright restrictions and made fully and freely available for both non-commercial and commercial use."

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Mon 22 Oct 2018 10:10:19 AM UTC, comment #39:

I was wondering if the improved documentation for the 3D structure files is sufficient?

Edward d'Auvergne <bugman>
Sun 05 Aug 2018 08:19:26 PM UTC, comment #38:

No problems. I have pushed the improved documentation:

https://sourceforge.net/p/nmr-relax/code/ci/92fa48c65f970a9f81daa53be16347166a5aa573/

Edward d'Auvergne <bugman>
Fri 27 Jul 2018 04:15:56 PM UTC, comment #37:

Ok, then you could add to README these explanations about why it's public domain and why the notices are written in README rather than in the files themselves.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Sun 22 Jul 2018 12:41:53 PM UTC, comment #36:

>> ...the REMARK section numbering and format can be important... in many cases intricate, automated, software-specific formatting is used that simple hand edits could easily break
>
>It's up to you to make sure that the license notices don't break it for your package; since it's only about distribution within relax, I think you don't need to care about any other software.


Actually, relax is always used as part of a toolchain of different softwares. Interoperability with other software is very important. The relax UI includes interoperability functions for 13 different softwares, and support for reading input or writing output text files for over 15 other unique softwares. Here breaking the PDB format will, for example, hinder our support for Xplor-NIH. That is a very undesirable outcome. Hence why placing the copyright or public domain notice in an external README file, rather than breaking the strict PDB format, is really how these PDB files should be dealt with.

>> we absolutely cannot identify the original copyright holder.
>
>
>Then I think you should write down the reasons why this file is effectively public domain. We know there are cultures that neglect copyright, but the GNU Project holds the position that using copyrightable material is forbidden unless explicitly allowed.
>
>However, please feel free to discuss this with the FSF and RMS; they are in the position to make decisions about such corner cases.


Sorry for not making it clearer. The Protein Data Bank contents are public domain. All authors agree to waiver their copyright upon submission of their data. From the RCSB PDB Policies & References:


Usage Policies

Data files contained in the PDB archive (ftp://ftp.wwpdb.org) are free of all copyright restrictions and made fully and freely available for both non-commercial and commercial use. Users of the data should attribute the original authors of that structural data. By using the materials available in the PDB archive, the user agrees to abide by the conditions described in the PDB Advisory Notice.


Or from within in the Full Privacy Statement:


RCSB PDB operates an open-access portal for public domain information about the 3D shapes of proteins, nucleic acids, and complex assemblies that helps researchers, educators, and students understand fundamental biology, biomedicine, and bioenergy.


Or from the Worldwide PDB website:


Mission

Ensure open access to public domain experimentally determined structural biology data.


This is clearly in the public domain. The PDB files themselves always identify all authors of the corresponding peer-reviewed scientific publication. This publication is a strict requirement for deposited data, and almost all scientific journals conversely require the data to be deposited as public domain to the PDB when publishing a paper about a 3D structure. Hence we do not believe that we need to extract from this already present author list the pre-waiver copyright holders (which is not possible anyway) and credit them.

Edward d'Auvergne <bugman>
Sat 21 Jul 2018 10:36:03 AM UTC, comment #35:

> ...the REMARK section numbering and format can be important... in many cases intricate, automated, software-specific formatting is used that simple hand edits could easily break


It's up to you to make sure that the license notices don't break it for your package; since it's only about distribution within relax, I think you don't need to care about any other software.

> we absolutely cannot identify the original copyright holder.


Then I think you should write down the reasons why this file is effectively public domain. We know there are cultures that neglect copyright, but the GNU Project holds the position that using copyrightable material is forbidden unless explicitly allowed.

However, please feel free to discuss this with the FSF and RMS; they are in the position to make decisions about such corner cases.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Tue 17 Jul 2018 07:52:49 AM UTC, comment #34:

Sorry for the delay, I have just returned from holidays.

> These are text files (as well as AUTHORS), they should include the notices. README is only a fallback for binary files.


The PDB text file format is very rigid, it is a strict fixed-columnar formatted text file. For example this structure distributed with relax:

- 1OSA Protein Data Bank entry.
- 1OSA PDB file.

The original copyright holders prior to the content being made public domain could theoretically be added to the PDB REMARK sections. However the REMARK section numbering and format can be important - although they can be in free form, in many cases intricate, automated, software-specific formatting is used that simple hand edits could easily break. Also, as discussed previously, we absolutely cannot identify the original copyright holder. This task is impossible with the design of the public domain Protein Data Bank, as neither the PDB nor the original scientific publication identify the original copyright holder. This can sometimes be in the AUTHOR record, but again this could simply be the name of the supervisor, the lead scientist (as in the 1OSA example), or all the authors rather than the original copyright holder. In the almost 50 year history of the PDB, recording the true original copyright holder(s) has not be considered important, as that person(s) has waived their copyrights. In science, it is all the authors of the original paper that matter - and all those authors are credited in the PDB database files in the JRNL AUTH record.

In the case of our software relax, most PDB files are used in the test suite fall into two categories:

1) Those that are deliberately unmodified from the originals found in the Protein Data Bank database. This is so that we can forever test that we can correctly read the format. Being able to read modified PDB files does not guarantee that we can correctly read the originals.

2) Those PDB files that are generated either by relax or by other software. Again we need to test that the unmodified original can be read correctly. For a few tests, we read the PDB file generated by relax into internal data structures, output a new PDB file, and check that the diff is zero.

We believe that having public domain statements (not identifying the pre-copyright-waiver owners), as well as copyright notices, in an accompanying README file is the correct way to handle PDB text files:

- test_suite/shared_data/structures/README

> WRT the GFDL, I'm not sure. Perhaps you'll have to speak to licensing@fsf.org.


I will try corresponding with RMS and the FSF licensing department.

Edward d'Auvergne <bugman>
Sat 23 Jun 2018 06:58:56 PM UTC, comment #33:

>> ...but anyway, when the user looks at these PDB files...
> There is a README file in the same directory that lists the files as being in the public domain.


These are text files (as well as AUTHORS), they should include the notices. README is only a fallback for binary files.

WRT the GFDL, I'm not sure. Perhaps you'll have to speak to licensing@fsf.org.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Fri 04 May 2018 09:08:00 AM UTC, comment #32:

> While mathematical equations may not be copyrigtable, their expressions are.


The text around the equations are copyrightable, but the expression of the formula itself is not. The formulae expressed in the relax manual are standard forms of mathematics that any 1st year university student would be able to derive themselves, and express in LaTeX in exactly the same way. This is covered by a concept called the idea-expression divide. A practitioner would come up with exactly the same expressions in the relax manual without ever seeing the manual, and would come up with the same LaTeX code for it.

For example, according to the US Copyright Office (my emphasis added):

"Copyright law does not protect ideas, methods, or systems. Copyright protection is therefore not available for ideas or procedures for doing, making, or building things; scientific or technical methods or discoveries; business operations or procedures; mathematical principles; formulas or algorithms; or any other concept, process, or method of operation."

Due to the limited and standardised way of presenting mathematical equations, the formula and its expression are not dividable. So it does not satisfy the idea-expression divide.

>> Using the GPLv3+ was
>> determined to be far less complicated than having a FDLed document
>> interspersed with a huge number of GPLed text, graphic, and data components.
>> So we chose licensing simplicity for legal clarity.
>
> Does it invalidate Savannah policies? The GNU guidelines say when some code
> may be useful both in program and in documentation, it should be released
> in parallel under the GPL and the FDL.


I believe that this would be abused by some in our field to make changes that cannot be reincorporated into relax. A small subset of scientists are rather unscrupulous and only consider immediate personal gain rather than mutual benefit. If dual licensed, they could take one of the complex scripts in the manual (these are direct copies from the relax sample_scripts/ directory) and release their changes solely under the FDL licence, as is allowed with dual licensing. These changes hence cannot be reincorporated into relax as they must in turn be dual licensed by the 3rd party. So we cannot take the FDL changes and incorporate them into our GPL scripts, then copy that into the manual and dual FDL+GPL license it.

This is a real issue and the pressures of academia will almost guarantee that it will be exploited!

This, together with the other large quantity of GPLv3+ material in the manual, including a large percentage of auto-generated content from GPLv3+ Python source code, is why we have licensed the relax manual and API documentation GPLv3+ instead of FDLv1.3+. This concern does not apply to our webpages (excluding the HTML versions of the API and manual), hence they are FDLv1.3+ licensed.

We also calculated that the GPL free software licence would have been acceptable for documentation at Savannah. The following text appeared to us to be non-exclusive for the FDL:

"GNU GPL-compatible license: your license should be listed as compatible at http://www.gnu.org/licenses/license-list.html. You can also use the GNU Affero GPL, since it is effectively compatible with GPLv3. For documentation, we accept the GNU Free Documentation License (and compatible), even though is not compatible with the GNU GPL. Always use the "or any later version" wording in your notices, as otherwise future compatibility problems are crated. (Aside: for the LGPL, we can technically accept LGPL(star)-only since it can be converted to any version of the GPL, but we nevertheless strongly recommend against using LGPL(star)-only.)"

After listing the GPL, the text "we accept the [FDL], even though is not compatible with the GNU GPL" reads to be discretionary. I.e. Use the GPL or FDL. We also assumed that the optional nature of the FDL in this text is because there are cases where the GPL is a better choice than the FDL. For example our auto-generated API documentation which includes annotated copies of all our source files (e.g. http://www.nmr-relax.com/api/4.0/auto_analyses.dauvergne_protocol-pysrc.html).

> I don't think this addresses my concerns: the license for those files isn't
> expressed consistently, and it isn't clear what tools you used for relicensing.


That was fixed when I posed the previous comment. They are now consistently labelled as being LGPLv3+ licensed everywhere.

> If it was a student who created the file, the chances are that the copyright
> holder is their university...


In some cases yes, in most cases though it is not. See for example https://www.jisc.ac.uk/guides/copyright-guide-for-students. It depends on the University's policy, as well as the copyright law in each country. In any case, it is not possible determine who the original copyright holder is with any certainty. So we cannot credit the original copyright holders in the public domain PDB 3D structure files. Only the authors listed in the headers of the files themselves are credited as being authors (but not necessarily copyright holders).

> ...but anyway, when the user looks at these PDB files, their copyright
> status should be clear, without asking the developers of the package.
> When I look at these files, their status isn't clear for me even after
> your explanation (I'm sorry, I couldn't find the statement about the data
> being in public domain on Protein Data Bank websites, I admit I didn't
> look for it very well).


There is a README file in the same directory that lists the files as being in the public domain.

>>> and why opening SVG files in an editor like Inkscape shows that they have a proprietary license,
>>
>> This new Inkscape "feature" of listing the licence, which I've only just
>> found out about, did not exist when these graphics were created. It seems to
>> default to "proprietary" in the Inkspace application for any SVG without
>> <cc:license/> tags. But that does not make our vector graphics "proprietary".
>
>No, but it raises an issue.
>
>The task is to prominently and unambiguously notify the user
>about the status of each file.
>
>Inkscape is one of the natural ways to edit SVG files, and it doesn't show
>your embedded comments; moreover, it may even not preserve them after minor
>unrelated edits. This is not quite good.
>
>> Should we be adding "Creative Commons" tags to our SVG files just for the
>> benefit of the current version of the Inkscape application?
>
>CC? Your comments say "GPLv3+".


The XML SVG tag invented by the Inkscape developers is <cc:license/>. The "cc" in this tag refers to "Creative Commons". I think that the Inkscape developer who came up with this needs to be educated about free software! Also, the <cc:license/> tag is not part of the SVG standard, as defined by the W3C.

As <cc:license/> is not part of the SVG standard, editing the SVG in any other vector graphics editor will cause the Inkscape-Creative-Commons-specific tag to be lost. A quick test using LibreOffice Draw to open and save such an Inkscape SVG shows this to be the case. And it may change in a newer Inkscape version, as it is not part of the SVG standard. I also cannot find any mention of this specific tag on the Inkscape mailing lists. I cloned their repository and can see it was added prior to their first git commit back in 2006. But their old SVN repository no longer seems to be online so I cannot chase down the origin of this tag.

So again I would prefer not to add such narrow, software and CC license specific tags. This does not advance the free software cause. The fact that they list a file as being "proprietary" if that file does not adopt their non-SVG-standard Inkscape-specific and Creative-Commons-specific XML tag is clearly a bug/failure on their part.

> I can't see the second page in PDFs; would the copyright line be valid
> for them all?


The manual is only generated with each relax release. We used to have about 6-10 releases per year. Since the Gna! shutdown over a year ago, we do not have the required free software infrastructure behind the project to make any releases. The copyright line now added to the source code repository will appear in the PDFs once we can perform our long overdue next release.

The copyright statement has now been slightly clarified:

"""
Copyright (C) 2001-2018 the relax development team

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU General Public License (GPL), Version 3 or any later version published by the Free Software Foundation.

The Oxygen Icons used herein are licensed under the terms of the GNU Lesser General Public License (GPL), Version 3 or any later version published by the Free Software Foundation.
"""

> Anything more than 15 lines long is presumably copyrightable; if not,
> it makes sense to add an explanation why it isn't.


As I am not a lawyer who can say that a list of names is not copyrightable, I have modified the graphics/oxygen_icons/README file to list the AUTHORS files under the same LGPLv3+ licensing and authorship as the rest of the Oxygen Icon files.

Edward d'Auvergne <bugman>
Tue 01 May 2018 01:49:45 PM UTC, comment #31:

> The PDF manual and auto-converted HTML manual are GPLv3+ licensed. These
> contain lots of source code fragments, full scripts, UI screenshots,
> documentation auto-generated from the code (215 pages from 712), and a huge
> number of non-copyrightable mathematical equations.


While mathematical equations may not be copyrigtable, their expressions are.

> Using the GPLv3+ was
> determined to be far less complicated than having a FDLed document
> interspersed with a huge number of GPLed text, graphic, and data components.
> So we chose licensing simplicity for legal clarity.


Does it invalidate Savannah policies? The GNU guidelines say when some code
may be useful both in program and in documentation, it should be released
in parallel under the GPL and the FDL.

>> (e.g. what license actually applies to graphics/oxygen_icons? GPL? LGPL? may
>> you really relicense it under GPLv3 "or later"?;
>
> The authors listed in the AUTHORS file originally licensed the graphics under
> the LGPLv3+. They use the text "or (at your option) any later version" in
> their COPYING file. The original AUTHORS and COPYING file have been placed
> into the graphics/oxygen_icons/ directory.
>
> We originally just had the LGPLv3 licence text in the
> graphics/oxygen_icons/COPYING file, rather than their "annotated" version.
> The new README files contained a cut and paste error that is now fixed (the
> copied GPL notice should have been modified to be LGPL).


I don't think this addresses my concerns: the license for those files isn't
expressed consistently, and it isn't clear what tools you used for relicensing.

>> and public domain files should also have notices saying who was the original copyright holder;
>
> This is the case for all the public domain content except for PDB (Protein
> Data Bank) 3D protein structures.

...

> The "AUTHOR" entry is sometimes the "original copyright
> holder", sometimes the professor who does not have copyright ownership rather
> than their student who created the file, and sometimes all of the authors on
> the publication.


If it was a student who created the file, the chances are that the copyright
holder is their university...

> Upon submission of a 3D structure to the PDB, the "original
> copyright holder" accepts to release the work as public domain. Therefore
> there is no legal question about the copyright-free status of the contents in
> the Protein Data Bank.


...but anyway, when the user looks at these PDB files, their copyright
status should be clear, without asking the developers of the package.
When I look at these files, their status isn't clear for me even after
your explanation (I'm sorry, I couldn't find the statement about the data
being in public domain on Protein Data Bank websites, I admit I didn't
look for it very well).

>> and why opening SVG files in an editor like Inkscape shows that they have a proprietary license,
>
> This new Inkscape "feature" of listing the licence, which I've only just
> found out about, did not exist when these graphics were created. It seems to
> default to "proprietary" in the Inkspace application for any SVG without
> <cc:license/> tags. But that does not make our vector graphics "proprietary".


No, but it raises an issue.

The task is to prominently and unambiguously notify the user
about the status of each file.

Inkscape is one of the natural ways to edit SVG files, and it doesn't show
your embedded comments; moreover, it may even not preserve them after minor
unrelated edits. This is not quite good.

> Should we be adding "Creative Commons" tags to our SVG files just for the
> benefit of the current version of the Inkscape application?


CC? Your comments say "GPLv3+".

> This is not even
> used by the Creative Commons organisation themselves (see
> https://mirrors.creativecommons.org/presskit/logos/cc.logo.svg).


Creative Commons has no aim to follow Savannah hosting requirements.

>> and why normally reading DVI and PDF files people don't see the notices) and
>> one has to figure out why files like graphics/oxygen_icons/AUTHORS are
>> missed.
>
> I'm not sure what this means? Should the GPLv3+ licensing of the PDF and HTML
> manual be better advertised? If so, I have just added a second title page
> with this info. It adds the following text to the bottom of page 2:


"""
Copyright (C) 2001-2018 the relax development team

Permission is granted to copy, distribute and/or modify this document under the
terms of the GNU General Public License (GPL), Version 3 or any later version
published by the Free Software Foundation.
"""

I can't see the second page in PDFs; would the copyright line be valid
for them all?

> For graphics/oxygen_icons/AUTHORS, this simply a list of authors. Does it
> require a copyright notice? I thought this type of content is not
> copyrightable.


Anything more than 15 lines long is presumably copyrightable; if not,
it makes sense to add an explanation why it isn't.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Thu 26 Apr 2018 10:18:25 AM UTC, comment #30:

>> Does this mean that we have to wait ~4 months for any progress on this task?
>
> Strictly speaking, you don't have: you could make sure that Savannah hosting requirements (for example, the documentation should be released in a FDL1.3+-compatible way) are met by then (i.e. June).


As far as I was aware, that is what we have always strived to do:

  • The API documentation is GPLv3+, matching the source code as expected (it is 100% auto-generated from the source code).
  • The PDF manual and auto-converted HTML manual are GPLv3+ licensed. These contain lots of source code fragments, full scripts, UI screenshots, documentation auto-generated from the code (215 pages from 712), and a huge number of non-copyrightable mathematical equations. Using the GPLv3+ was determined to be far less complicated than having a FDLed document interspersed with a huge number of GPLed text, graphic, and data components. So we chose licensing simplicity for legal clarity.

> Scripts are nice, but they rarely catch things like licensing inconsistency


The relax project was started in 2001 with an aim of adhering to all of Richard Stallman and the FSF goals to be truly free software. That is why we chose Gna! hosting originally (see this discussion about the Affero licence). Any licensing inconsistencies will be due to honest mistakes. With such an old project written by scientists (i.e. non-programmers) for scientists, and including a large quantity of data files for complete test coverage, some issues might creep in. But I have tried to be vigilant and have caught and fixed most of these issues as they are committed.

The script I wrote was to help find any inconsistencies (but not all). As many other GPLv3+ licensed projects do, we also did not attach copyright notices to non-source-code files (data and graphics). This is under the assumption that those copying any source or content would look at the repository(s) to obtain that information (otherwise they would not have a legal licence for their distributed copy). We have changed that policy now so that all files have a copyright notice, and the script is a useful tool for identifying those with missing notices.

The script will also be very useful for automated checking in the future in case other developers do not meet the FSF licensing standards. The developer can run the script themselves to see the issue.

> (e.g. what license actually applies to graphics/oxygen_icons? GPL? LGPL? may you really relicense it under GPLv3 "or later"?;


The authors listed in the AUTHORS file originally licensed the graphics under the LGPLv3+. They use the text "or (at your option) any later version" in their COPYING file. The original AUTHORS and COPYING file have been placed into the graphics/oxygen_icons/ directory.

We originally just had the LGPLv3 licence text in the graphics/oxygen_icons/COPYING file, rather than their "annotated" version. The new README files contained a cut and paste error that is now fixed (the copied GPL notice should have been modified to be LGPL).

>and public domain files should also have notices saying who was the original copyright holder;


This is the case for all the public domain content except for PDB (Protein Data Bank) 3D protein structures. For the structures listed in test_suite/shared_data/structures/README and elsewhere as being public domain, it is impossible to determine the "original copyright holder". The PDB file lists all authors on the scientific paper. However the "original copyright holder" is only one of those authors and that author is not identified. See the "AUTHOR" and "JRNL AUTH" entries in these files for the author lists. The "AUTHOR" entry is sometimes the "original copyright holder", sometimes the professor who does not have copyright ownership rather than their student who created the file, and sometimes all of the authors on the publication. Upon submission of a 3D structure to the PDB, the "original copyright holder" accepts to release the work as public domain. Therefore there is no legal question about the copyright-free status of the contents in the Protein Data Bank.

> and why opening SVG files in an editor like Inkscape shows that they have a proprietary license,


This new Inkscape "feature" of listing the licence, which I've only just found out about, did not exist when these graphics were created. It seems to default to "proprietary" in the Inkspace application for any SVG without <cc:license/> tags. But that does not make our vector graphics "proprietary". Should we be adding "Creative Commons" tags to our SVG files just for the benefit of the current version of the Inkscape application? This is not even used by the Creative Commons organisation themselves (see https://mirrors.creativecommons.org/presskit/logos/cc.logo.svg).

> and why normally reading DVI and PDF files people don't see the notices) and one has to figure out why files like graphics/oxygen_icons/AUTHORS are missed.


I'm not sure what this means? Should the GPLv3+ licensing of the PDF and HTML manual be better advertised? If so, I have just added a second title page with this info. It adds the following text to the bottom of page 2:

"""
Copyright (C) 2001-2018 the relax development team

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU General Public License (GPL), Version 3 or any later version published by the Free Software Foundation.
"""

For graphics/oxygen_icons/AUTHORS, this simply a list of authors. Does it require a copyright notice? I thought this type of content is not copyrightable.

Edward d'Auvergne <bugman>
Wed 25 Apr 2018 04:43:28 PM UTC, comment #29:

> Does this mean that we have to wait ~4 months for any progress on this task?


Strictly speaking, you don't have: you could make sure that Savannah hosting requirements (for example, the documentation should be released in a FDL1.3+-compatible way) are met by then (i.e. June).

Scripts are nice, but they rarely catch things like licensing inconsistency (e.g. what license actually applies to graphics/oxygen_icons? GPL? LGPL? may you really relicense it under GPLv3 "*or later*"?; and public domain files should also have notices saying who was the original copyright holder; and why opening SVG files in an editor like Inkscape shows that they have a proprietary license, and why normally reading DVI and PDF files people don't see the notices) and one has to figure out why files like graphics/oxygen_icons/AUTHORS are missed.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Tue 06 Mar 2018 08:38:07 AM UTC, comment #28:

> See you in June.


Does this mean that we have to wait ~4 months for any progress on this task?

Edward d'Auvergne <bugman>
Sun 04 Mar 2018 11:43:10 AM UTC, comment #27:

Actually, the error was rather in the FSF Copright Validation script configuration file for relax.

Edward d'Auvergne <bugman>
Sun 04 Mar 2018 11:37:14 AM UTC, comment #26:

Sorry, that was a single mistake in my automated FSF Copyright Validation script, specifically:

DIR_SKIP = [
'.git',
'.svn',
'graphics/oxygen_icons', # External source, copyright documented as much as possible.
]

This is now fixed and README files have been added to all of the Oxygen Icon directories. I have rerun this script and checked every last file in the relax source repository, and all copyrights notices appear to be up to the FSF standard.

Cheers,
Edward

Edward d'Auvergne <bugman>
Sun 04 Mar 2018 07:47:05 AM UTC, comment #25:

Unfortunately, some files still aren't compliant, for example, graphics/oxygen_icons/scalable/actions/chronometer.svgz has no notices, and that directory has no README that would list it with its copyright notice and distribution terms. (I don't think saying like "all files here are copyrighted by RJH and distributed under the terms of *PL" is really sufficient.)

I understand that the work is enormous, however, it isn't complete yet. I hope you'll understand that it would be hard for me to check it too often: the rules are clear, and you could do it yourself before asking me next time.

See you in June.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Fri 02 Mar 2018 05:04:35 PM UTC, comment #24:

For the record, here was our previous Gna! homepage, archived on the Internet Archive Wayback Machine: https://web.archive.org/web/20170301004608/https://gna.org/projects/relax

Edward d'Auvergne <bugman>
Fri 02 Mar 2018 04:51:50 PM UTC, comment #23:

Hi Pavel! Have you had a chance to look at our project again? We are really looking forward to hosting our project on Savannah non-GNU, and finally finding a new home after a year of uncertainty.

Cheers,
Edward

Edward d'Auvergne <bugman>
Fri 26 Jan 2018 03:24:37 PM UTC, comment #22:

Sorry, I've been quite busy lately with RL and haven't had a chance to respond. But the relax users are still waiting for me to set up mailing lists, bug trackers, and the rest of the infrastructure. I have checked all files and, to me, everything looks to be licensed as specified by the FSF. This was a lot of work for such an established piece of software, so finally having free software infrastructure behind this project would be greatly appreciated. Since the inception of the project in 2001, I and the other relax developers have strived to make this software truly free software. I hope it can now be classified as such. Cheers!

Edward d'Auvergne <bugman>
Fri 26 Jan 2018 09:17:39 AM UTC, comment #21:

If there is no further interest, I'll cancel this submission.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Sat 25 Nov 2017 03:36:26 PM UTC, comment #20:

Hello, Edward!

I think the notices in extern/sobol are fine.

Did you check all other files in your tarball?

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Thu 23 Nov 2017 09:28:25 PM UTC, comment #19:

Hi, I was wondering if you had had a chance to look at the manually reconstructed copyright notices in the extern/sobol Python package? This reconstruction by code comparison was a relatively easy task. Cheers!

Edward d'Auvergne <bugman>
Wed 25 Oct 2017 08:02:52 AM UTC, comment #18:

>> The files already contain notices, but these are actually placed as comments
>> after every single function, e.g.:
>> https://sourceforge.net/p/nmr-relax/code/ci/master/tree/extern/sobol/sobol_lib.py
>
>
> There are no copyright notices after functions, there are information
> about the authors; the license notice is too ambiguous: it doesn't
> mention the allowed versions of the LGPL.


I have now noticed that this has changed over time. When I originally bundled the Sobol Python package, the text after each function was for example:

# Licensing:
#
# This code is distributed under the GNU LGPL license.
#
# Modified:
#
# 22 February 2011
#
# Author:
#
# Original MATLAB version by John Burkardt.
# PYTHON version by Corrado Chisari

I see now that the licence has changed:

# Licensing:
#
# This code is distributed under the MIT license.
#
# Modified:
#
# 22 February 2011
#
# Author:
#
# Original MATLAB version by John Burkardt.
# PYTHON version by Corrado Chisari

The original MATLAB code from John Burkardt is still LGPL licensed. From memory, there was also previously text on the source webpage or a linked page that the GNU LGPL licence was the current version. For 2011 that would be version 3. However I cannot find that text any more, either on the webpage or on the Internet Archive Wayback Machine.

>> I have nevertheless added FSF recommended notices to the top of all of the
>> files
>
>
> Is that copyright notice valid? It only list one copyright holder, though
> there are much more authors. Did they assign their copyright to one
> person?


It looks like Corrado Chisari wrote the LGPL Python code. The original MATLAB version by J. Burkardt was LGPL licensed. Looking at the code side-by-side:

http://people.sc.fsu.edu/~jburkardt/m_src/sobol/
http://people.sc.fsu.edu/~jburkardt/py_src/sobol/

There should clearly be the copyright notice dates from the .m files in the .py files. Looking at all of the code on the webpages there, the history of the code appears to be:

FORTRAN77 version by Bennett Fox
FORTRAN90 version by John Burkardt
MATLAB version by John Burkardt
Python version by Corrado Chisari

None of the original FORTRAN77 code referenced in some functions appears in the FORTRAN90 code. So the history looks is FORTRAN90 -> MATLAB -> Python. I have never encountered a mention of copyright transfers, so I would assume that both John Burkardt and Corrado Chisari have copyright ownership. Though Chisari's MATLAB -> Python language translation would not have produced much new copyrightable material.

>> (excluding the auto-generated output files which remain in the README file).
>
>
>Those are text files. They should themselves contain the notices.


I have updated the Sobol package in relax to the newest MIT licensed version, changed the licence from LGPLv3 to MIT, and tracked down all of the dates throughout the FORTRAN90 -> MATLAB -> Python chain to create the copyright notices. See:

https://sourceforge.net/p/nmr-relax/code/ci/master/tree/extern/sobol/
https://sourceforge.net/p/nmr-relax/code/ci/master/tree/extern/sobol/sobol_lib.py

Edward d'Auvergne <bugman>
Wed 11 Oct 2017 01:47:17 PM UTC, comment #17:

> The files already contain notices, but these are actually placed as comments
> after every single function, e.g.:
> https://sourceforge.net/p/nmr-relax/code/ci/master/tree/extern/sobol/sobol_lib.py


There are no copyright notices after functions, there are information
about the authors; the license notice is too ambiguous: it doesn't
mention the allowed versions of the LGPL.

> I have nevertheless added FSF recommended notices to the top of all of the
> files


Is that copyright notice valid? It only list one copyright holder, though
there are much more authors. Did they assign their copyright to one
person?

> (excluding the auto-generated output files which remain in the README file).


Those are text files. They should themselves contain the notices.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Fri 22 Sep 2017 01:52:14 PM UTC, comment #16:

I have added the LGPLv3 licence text to the extern/sobol package:

https://sourceforge.net/p/nmr-relax/code/ci/master/tree/extern/sobol/COPYING.LESSER

For the copyrights in the README, I simply added these for clarification. The files already contain notices, but these are actually placed as comments after every single function, e.g.:

https://sourceforge.net/p/nmr-relax/code/ci/master/tree/extern/sobol/sobol_lib.py

I have nevertheless added FSF recommended notices to the top of all of the files (excluding the auto-generated output files which remain in the README file). The new code can be downloaded from one of the following mirrors:

https://gitlab.com/nmr-relax/relax/repository/master/archive.tar.bz2
https://github.com/nmr-relax/relax/archive/master.zip
https://bitbucket.org/nmr-relax/relax/get/master.tar.bz2
https://sourceforge.net/p/nmr-relax/code/ci/master/tarball

Edward d'Auvergne <bugman>
Wed 13 Sep 2017 01:02:37 PM UTC, comment #15:

Thank you!

Format of files in extern/sobol allow comments; copyright and license notices should be put in the files directly rather than in README.

Then, README says that some of those files are under the LGPLv3+, but no license text is provided.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Mon 11 Sep 2017 08:34:59 AM UTC, comment #14:

As numdifftools is now quite easy to install on Python using pip, I have simply removed the extern/numdifftools package from relax:

https://sourceforge.net/p/nmr-relax/code/ci/174fd5a42b54c08202eb8306be828f9125f86a95/

This was only bundled in relax for stability with the early numdifftools packages, and to ensure that relax always had the ability to use this code. Note that numdifftools is not a relax dependency, but rather an extremely useful tool for testing new code. Hopefully the relax sources mirrored in zip/tar form at one of:

https://github.com/nmr-relax/relax/archive/master.zip
https://sourceforge.net/p/nmr-relax/code/ci/master/tarball

or in source form at one of:

https://sourceforge.net/p/nmr-relax/code/ci/master/tree/
https://github.com/nmr-relax/relax
https://gitlab.com/nmr-relax/relax

will now meet the requirements for a non-GNU project.

Cheers,
Edward

Edward d'Auvergne <bugman>
Fri 08 Sep 2017 08:14:43 AM UTC, comment #13:

I have submitted an issue for numdifftools for licence clarity:

https://github.com/pbrod/numdifftools/issues/34

And for downloads, another automatically created archive is:

https://github.com/nmr-relax/relax/archive/master.zip

This interface is a little simpler.

Edward d'Auvergne <bugman>
Tue 05 Sep 2017 05:07:38 AM UTC, comment #12:

> The <ORGANIZATION> part seems to be there because they copied the boiler plate ...
> We cannot judge their intent.


Probably we can't; I think this means that you should ask them to explain their intent (preferably in a public statement). In corner cases the interpretation may render their package nonfree (for example, if they prohibit using names of users' organizations to endorse users' own products).

> This link should always have the newest changes:
>
> https://sourceforge.net/p/nmr-relax/code/ci/master/tarball


This doesn't work for me very well: first, it says "we have problems with finding this tarball" and shows a button "request a snapshot again", when I push it, it says "Generating snapshot..." without any further result.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Sun 03 Sep 2017 01:33:21 PM UTC, comment #11:

Note the latest relax code is mirrored - for backup purposes - in three different locations:

https://sourceforge.net/p/nmr-relax/code/ci/master/tree/
https://github.com/nmr-relax/relax
https://gitlab.com/nmr-relax/relax

The main branch from which all releases are tagged is 'master'.

Edward d'Auvergne <bugman>
Sun 03 Sep 2017 01:28:30 PM UTC, comment #10:

>> ... a numdifftools README added.


> This is not sufficient, all files should have notices (and the license isn't clear itself: what is <ORGANIZATION>?)


README files have now been placed in all directories listing the copyrights. Note this is a direct bundling of the numdifftools package from:

https://pypi.python.org/pypi/Numdifftools

Specifically the old 0.6 version which we have extensively tested. We may update this in the future, hence the placement into README files and not in the source files themselves. The <ORGANIZATION> part seems to be there because they copied the boiler plate 3-clause new BSD licence and didn't change this text:

https://en.wikipedia.org/wiki/BSD_licenses#3-clause_license_.28.22BSD_License_2.0.22.2C_.22Revised_BSD_License.22.2C_.22New_BSD_License.22.2C_or_.22Modified_BSD_License.22.29

This text is still in their newest version:

https://github.com/pbrod/numdifftools/blob/05fd7d190837ba78b415338eb5b0f0bae411960b/LICENSE.txt

I do not believe we have the right to change their licencing text, even though this is clearly an 'error'? We cannot judge their intent.

> Also, could you provide the resulting tarball?


This link should always have the newest changes:

https://sourceforge.net/p/nmr-relax/code/ci/master/tarball

Edward d'Auvergne <bugman>
Sun 03 Sep 2017 11:36:35 AM UTC, comment #9:

> ... a numdifftools README added.


This is not sufficient, all files should have notices (and the license isn't clear itself: what is <ORGANIZATION>?)

Also, could you provide the resulting tarball?

Thank you!

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Thu 31 Aug 2017 09:44:55 AM UTC, comment #8:

&gt; Thank you, great work!
&gt;
&gt; Unfortunately, some issues still need resolving.
&gt;
&gt; Licensing of some files in extern/ is not very clear,
&gt; for example, extern/sobol/sobol_lib.py has no copyright notices,
&gt; and it could have more unambiguous license notice;
&gt; extern/sobol/test.py has no notices at all.

The Sobol package is fixed and a numdifftools README added.

&gt; Images like graphics/screenshots/dispersion_analysis/disp_64_N.eps.gz
&gt; and docs/latex/images/pec_diag.eps suggest they were exported from GIMP
&gt; or created with gnuplot, so your users would have to provide
&gt; the source code for those files.

These are mostly found in the test_suite/shared_data directories, and pointed to in the manual as these are used for student tutorials. For example disp_64_N.eps.gz is used at:

http://www.nmr-relax.com/manual/Dispersion_GUI_mode_inspection_of_the_results.html

This is a section from:

http://www.nmr-relax.com/manual/The_relaxation_dispersion_auto_analysis_in_the_GUI.html

And the data, distributed with relax, to load and auto-generate this graph is referenced at:

http://www.nmr-relax.com/manual/Dispersion_GUI_mode_loading_the_data.html

where it says &quot;In this tutorial, the Sparky formatted peak lists in the test_suite/shared_data/dispersion/Hansen/500_MHz and test_suite/shared_data/dispersion/Hansen/800_MHz directories will be loaded.&quot; This is the base data for relax for generating the plots.

The docs/latex/images/pec_diag.eps diagram is different in that it was plotted in Mathematica. The base data generation was considered so trivial that the ~5 line script and 1-2 line Mathematica notebook was not kept. I have recreated a small relax script to quickly regenerate the data.

&gt; Likewise, graphics/screenshots/noe_analysis/grace.svg seems generated
&gt; from a different source.

Such plots should have had the base plotting files added. I'm not sure why some were missed. This example is now included.

In addition, the Xmgrace plot file for graphics/screenshots/xmgrace_peak_intensities.agr has been added as well. As far as I can tell, everything else is either in test_suite/shared_data/ or the original sources are there. Since the start of the project, we have made a lot of effort to obtain all original sources to include within the relax distribution. Any discrepancies are by accident.

&gt; Then, the documentation has some terminological inconsistencies:
&gt; in files like docs/latex/develop.tex the operating system sometimes is
&gt; called &quot;Linux&quot; rather than &quot;GNU/Linux&quot;; also, some TeX files speak
&gt; about &quot;open source&quot;; Savannah supports
&gt; free software.

Our intent was to use &quot;GNU/Linux&quot;, and most references use this. This is now updated. All other references are either non-mutable quotes, old commit messages, or the index reference back to &quot;GNU/Linux&quot;.

We have also switched all &quot;open source&quot; references, excluding the non-mutable ones, with &quot;free sofware&quot;.

Edward d'Auvergne <bugman>
Thu 31 Aug 2017 07:36:06 AM UTC, comment #7:

Thank you, great work!

Unfortunately, some issues still need resolving.

Licensing of some files in extern/ is not very clear,
for example, extern/sobol/sobol_lib.py has no copyright notices,
and it could have more unambiguous license notice;
extern/sobol/test.py has no notices at all.

Images like graphics/screenshots/dispersion_analysis/disp_64_N.eps.gz
and docs/latex/images/pec_diag.eps suggest they were exported from GIMP
or created with gnuplot, so your users would have to provide
the source code for those files.

Likewise, graphics/screenshots/noe_analysis/grace.svg seems generated
from a different source.

Then, the documentation has some terminological inconsistencies:
in files like docs/latex/develop.tex the operating system sometimes is
called "Linux" rather than "GNU/Linux"; also, some TeX files speak
about "open source"; Savannah supports
free software.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Mon 28 Aug 2017 09:36:28 AM UTC, comment #6:

After a lot of work - the svn to git migration while preserving all branch and non-standard trunk history, as well as the process of adding valid license and copyright notices to all files was brutal - this is now ready:

https://sourceforge.net/p/nmr-relax/code/ci/master/tarball

Cheers! For the copyrights, I developed a FSF copyright notice compliance checking script to check all files tracked in the repository. This script should help keep all copyright notices compliant in the future.

Edward d'Auvergne <bugman>
Thu 01 Jun 2017 07:50:33 AM UTC, comment #5:

Please let us know when a fixed tarball is ready.

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Wed 31 May 2017 08:05:19 PM UTC, comment #4:

No problem. I'm in the process of trying to convert the old svn repository to git, preserving history, and if successful I'll make the changes in the new git repository (which will hopefully be hosted here).

Edward d'Auvergne <bugman>
Wed 31 May 2017 03:22:36 PM UTC, comment #3:

All nontrivial files, including changelogs and images, should have valid license and copyright notices; some files e.g. in docs/ and graphics/ have not.

Could you fix that?

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Tue 30 May 2017 08:53:27 AM UTC, comment #2:

We are in the process of looking at a svn to git migration, or uploading the last good backup of the svn backend before the unfortunate Gna! shutdown. We will probably post an archive of the SVN repository to https://sourceforge.net/projects/nmr-relax/, as the Gna! servers, where the files were attached, are gone for good. We also have mirrors of our distribution files, for example the last release at https://sourceforge.net/projects/nmr-relax/files/4.0.3/.

Edward d'Auvergne <bugman>
Tue 30 May 2017 08:43:57 AM UTC, comment #1:

I can't download the tarball; could you attach it in a comment?

Ineiev <ineiev>
Site AdministratorIn charge of this item.
Mon 29 May 2017 09:28:33 AM UTC, original submission:

A new project has been registered at Savannah
This project account will remain inactive until a site admin approves or discards the registration.

Registration Administration

While this item will be useful to track the registration process, approving or discarding the registration must be done using the specific Group Administration page, accessible only to site administrators, effectively logged as site administrators (superuser):

Registration Details

  • Name: relax
  • System Name: relax
  • Type: non-GNU software and documentation
  • License: GNU General Public License v3 or later

Description:

The software package 'relax' is designed for the study of molecular dynamics through the analysis of experimental NMR data. Organic molecules, proteins, RNA, DNA, sugars, and other biomolecules are all supported. It supports exponential curve fitting for the calculation of the R1 and R2 relaxation rates, calculation of the NOE, reduced spectral density mapping, the Lipari and Szabo model-free analysis, study of domain motions via the N-state model and frame order dynamics theories using anisotropic NMR parameters such as RDCs and PCSs, the investigation of stereochemistry in dynamic ensembles, and the analysis of relaxation dispersion data.

Other Software Required:

Python, Python Software Foundation License, https://www.python.org/

numpy, BSD licence, http://www.numpy.org/

Optional dependencies listed at https://web.archive.org/web/20161104044122/http://www.nmr-relax.com:80/download.html#Dependencies

Other Comments:

This is an attempt to migrate from http://gna.org/projects/relax :S

Tarball URL:

http://download.gna.org/relax/relax-4.0.3.src.tar.bz2

Edward d'Auvergne <bugman>

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach Files:
   
   
Comment:
   

No files currently attached

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by ineiev (Posted a comment)
  • -email is unavailable- added by bugman (Submitted the item)
  •  

    Do you think this task is very important?
    If so, you can add your encouragement to it.
    This task has 0 encouragements so far.

    Only logged-in users can vote.

     

     

     

    Follow 9 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2018-05-01 ineiev Should Start On- => 2017-05-28
        Should be Finished on- => 2019-05-01
    2018-05-01 ineiev Should Start On- => -
        Should be Finished on- => -
    2018-05-01 ineiev Should Start On2017-05-28 => -
        Should be Finished on2017-06-07 => -
    2017-05-31 ineiev StatusNeed Info => In Progress
    2017-05-30 ineiev StatusNone => Need Info
        Assigned toNone => ineiev

    Back to the top


    Powered by Savane 3.3