[IPOL discuss] Patents and IPOL code
José Luis Lisani
joseluis.lisani at uib.es
Sun Apr 10 23:40:49 CEST 2011
the text is OK for me
José Luis
El 10/04/2011 16:23, Jean-Michel Morel escribió:
> Hi,
>
> Would the following guidelines for authors of IPOL algorithms using
> patented algorithms be O.K.? (They are a synthesis of your proposals
> and comments). Your corrections please? We need a short clear text.
>
> Best,
> Jean-Michel
>
> oooo Guidelines for authors publishing at IPOL patented algorithms oooo
>
> 1) Add the patent references in the references
> 2) In the source code linked to the patent, add the following header:
>
> _____________________________________________________
> Copyright (c) 2011, Your Name Here
> All rights reserved.
>
> The source code files <NAMES OF THE FILES> implement as tightly as
> possible algorithms described in this IPOL article. They are made
> available to the exclusive aim of serving as scientific tools enabling
> the verification of the soundness and completeness of the algorithmic
> descriptions.
>
> These source codes implement an algorithm possibly linked to the
> patent <REFERENCE OF THE PATENT OR PENDING PATENT GIVEN IN THE TEXT>.
> Compiling or running this code may
> violate patents in certain countries. For this reason, the use of the
> source files <NAMES OF THE FILES> may be restricted in certain
> countries to non profit research
> and non profit educational purposes. In certain countries,
> redistribution or commercial use of these source files may require the
> consent of the patent owner.
> In short, be careful before you download, compile, use, modify, or
> redistribute these source codes. The situation being different for
> every country and changing over time, it is your responsibility to
> check that you are not infringing a patent by using this source code.
> IPOL therefore invites potential users to consult a patent lawyer. If
> and only if you are free from any patent restriction, then you can
> enjoy the <NAME OF LICENSE> license terms.
>
> With the exception of the files mentioned above, redistribution and
> use in source and binary forms, with or without modification, are
> permitted
> provided that the following conditions are met:
> (the rest of usual <NAME OF LICENSE> license follows)
> _________________________________________
>
> Pierre Moulon a écrit :
>> Hi,
>>
>> What occur for partially patented algorithm ?
>>
>> I think to all the method that exist and that will be added to IPOL
>> that use
>> SIFT as a black box for matching.
>>
>> => It falls in the Choice 3, with a part of the source code that is
>> patented.
>> Or for other choice => Potentially only a subpart of the algorithm
>> could be
>> published...
>>
>> Regards,
>> Pierre
>>
>> 2011/4/8 Jean-Michel Morel <morel at cmla.ens-cachan.fr>
>>
>>> Dear all,
>>>
>>> The final decision lies with the IPOL scientific committee and the IPOL
>>> editorial board. We will have to submit them the following alternative
>>> which summarizes our debate:
>>>
>>> Choice 1: For patented algorithms, the source code will not be
>>> available
>>> to the public, but only to the referees. There will be no online
>>> demo for
>>> the patented algorithms.
>>>
>>> Choice 2: For patented algorithms, the source code will not be
>>> available
>>> to the public but only to the referees. The online demo will be
>>> maintained.
>>>
>>> Choice 3: For patented algorithms, the source code will be
>>> available, but
>>> the users will be warned of infringement risks. The journal will
>>> argue that
>>> the source code is made available as a scientific proof of
>>> completeness and
>>> consistency of the described algorithm. The users will also be
>>> warned that
>>> the source code may be withdrawn from IPOL in case of a request by the
>>> patent owners.
>>>
>>> I can prepare a text to submit this decision to both committees, but I
>>> would like to have the final draft from José Luis, Nicolas and
>>> Pascal for
>>> the warnings in the source codes (which will be at least delivered
>>> to the
>>> referees).
>>>
>>> Best,
>>> JM
>>>
>>>
>>>
>>>
>>> José Luis Lisani a écrit :
>>>
>>>> Dear all,
>>>>
>>>> I don't think authors should have the right to decide not to make
>>>> public
>>>> the code of an algorithm.
>>>> If it is IPOL's policy to make publicly available the source codes,
>>>> they
>>>> have to accept this policy
>>>> or withdraw the submission.
>>>>
>>>> This been said, and concerning the particular submission in which I'm
>>>> involved, there is a misunderstanding
>>>> in what has been my intention with respect to the publication of the
>>>> source code.
>>>>
>>>> We are in the process of discussing the convenience of publishing
>>>> patented
>>>> algorithms and I have expressed
>>>> my opinion (I think it is safer for IPOL not to make publicly
>>>> available
>>>> the source code), and I have acted in
>>>> consequence with the publication I'm currently editing (I'm talking in
>>>> quality of editor, not as a co-author).
>>>>
>>>> Since the discussion seemed open to me, I opted for the conservative
>>>> approach of not providing the source
>>>> code in the web page. My intention was, once the discussion was
>>>> over, to
>>>> follow the policy of IPOL on this
>>>> matter.
>>>>
>>>> After Jean-Michel's contribution to the discuss list it seems that
>>>> indeed
>>>> the discussion is closed and the
>>>> conclusion is that source code should be published and some legal
>>>> warnings
>>>> must be added to the code.
>>>> Therefore, as editor of the paper, I will follow this guideline and
>>>> I will
>>>> add a link to the source code in the
>>>> web site, just as in any regular IPOL paper.
>>>>
>>>>
>>>> Best,
>>>> José Luis
>>>>
>>>>
>>>> El 07/04/2011 17:09, Jean-Michel Morel escribió:
>>>>
>>>>> Dear all,
>>>>>
>>>>> Reading José Luis message and his submission together with Juan Biel
>>>>> Gomila of a patented algorithm from which he has withdrawn public
>>>>> access to
>>>>> the code I realize that the authors can decide what they want on this
>>>>> matter. José Luis and Juan Biel Gomila can of course decide not to
>>>>> disclose
>>>>> the source code at IPOL. But I certainly do not support this
>>>>> decision, and
>>>>> I hope it will not happen too often.
>>>>>
>>>>> My opinion matters in that I and probably also my institution
>>>>> might be
>>>>> held personally responsible in case of patent infringement at IPOL.
>>>>>
>>>>> I had mentioned that IPOL could give up the disclosure of source
>>>>> code of
>>>>> patented algorithms. But this was not proposed as a a rule, it was
>>>>> just
>>>>> mentioned as a possible retreating line in case of trouble. For
>>>>> the time
>>>>> being there is no such trouble.
>>>>>
>>>>> The patented algorithms published in IPOL have been disclosed by
>>>>> their
>>>>> authors themselves in the patent AND in journal papers. The
>>>>> description of
>>>>> the algorithm that IPOL does is a public discussion of the public
>>>>> paper and
>>>>> is therefore not a clear infringement, even if it certainly
>>>>> provides help
>>>>> for infringement. But any patent discussion does, and there are even
>>>>> journals dedicated to the analysis of patents, which to the best
>>>>> of my
>>>>> knowledge are not illegal. As Pascal Getreuer pointed out, the
>>>>> source code
>>>>> by itself is not a direct infringement, but may be interpreted as
>>>>> a help to
>>>>> infringement. Thus, it is all a question of degree and of
>>>>> jurisprudence.
>>>>> From my scientific point of view the scientific paper published by
>>>>> the
>>>>> patent authors themselves coul also be interpreted as a help to
>>>>> infringe,
>>>>> and IPOL simply follows suit.
>>>>>
>>>>> IPOL is a purely scientific endeavor. We view the source code as a
>>>>> more
>>>>> detailed description of a published algorithm, permitting to check
>>>>> that the
>>>>> algorithm description given by IPOL is complete. Together with
>>>>> the demo, it
>>>>> permits users to verify this completeness and the reality of the
>>>>> experiments shown by the authors of the algorithm.
>>>>>
>>>>> This is why IPOL gives little or no value in the evaluation to well
>>>>> written code or to a smart implementation: we furnish to the
>>>>> public an
>>>>> implementation That helps them checking that the algorithm works,
>>>>> and seeing
>>>>> how it works. This is therefore not aimed at fostering any patent
>>>>> infringement, but just at informing better the scientific public
>>>>> on the
>>>>> meaning of an invented algorithm.
>>>>>
>>>>> The source code plays therefore more or less the same role as the
>>>>> publication of a proof for a theorem. It gives evidence linked to the
>>>>> algorithm, which makes people understand it better. Thus, it is a
>>>>> pity to
>>>>> withdraw the source code of a patented algorithm, and it will be a
>>>>> pity if
>>>>> we ever have to stop an online demo.
>>>>>
>>>>> On the other hand, I definitely agree with all warning texts that
>>>>> have
>>>>> been proposed, including the last one written by José. We must
>>>>> indeed inform
>>>>> the public that certain algorithms are patented, as soon as we get
>>>>> aware
>>>>> that they are, and we must inform the public that what we publish is
>>>>> published to foster image science, and nothing else.
>>>>>
>>>>> I hope that José Luis, Nicolas and Pascal will propose a final
>>>>> version
>>>>> warning text that we could include in the Zhang-Wu, NL-means,
>>>>> local color
>>>>> correction, SIFT and ASIFT demos.
>>>>>
>>>>> A last point. Since the source code is just on IPOL as an
>>>>> illustration of
>>>>> the algorithm, we cannot put any license when the source code
>>>>> implements a
>>>>> patented algorithm. I do not see how we can license a source code,
>>>>> even
>>>>> conditionally, even with warnings, if we are not the patent
>>>>> owners. I
>>>>> consider it safer to just warn users about the patent, as is done
>>>>> in the
>>>>> text thay you are maturing.
>>>>>
>>>>> As Nicolas pointed out we are no patent lawyiers and it is not our
>>>>> job to
>>>>> enforce patents or to make an inquiry about how far one can go or
>>>>> not go in
>>>>> the scientific exploration of a patent.
>>>>>
>>>>> We are on scientific terrain so far, and nobody has sued us. The
>>>>> day it
>>>>> happens we will hire a lawyer and find the best strategy. But I
>>>>> simply
>>>>> refuse to do this preventively.
>>>>>
>>>>> At this point, could Pascal propose a new version of the text? I
>>>>> enclose
>>>>> the last current proposals:
>>>>>
>>>>> ooooooooooo Pascal Proposal ooooooooooooooooooo
>>>>>
>>>>> Copyright (c) 2011, Your Name Here
>>>>> All rights reserved.
>>>>>
>>>>> The source file "dmha.c" implements the algorithm covered by U.S.
>>>>> Patent
>>>>> 5629734. For this reason, the source files "dmha.c", "dmha.h",
>>>>> "dmhacli.c" and its binary program "dmha" may only be used for
>>>>> research
>>>>> and educational purposes. Redistribution or commercial use is not
>>>>> allowed without the consent of the patent owner.
>>>>>
>>>>> With the exception of the files mentioned above, redistribution and
>>>>> use in
>>>>> source and binary forms, with or without modification, are permitted
>>>>> provided that the following conditions are met:
>>>>> (the rest of usual BSD license follows)
>>>>> _____________________________________________________
>>>>>
>>>>> For files implementing the patent:
>>>>> "may only be used for research and educational purposes" and
>>>>> "redistribution or commercial use is not allowed without the consent
>>>>> of the patent owner."
>>>>>
>>>>> For all the other files:
>>>>> "With the exception of the files implementing the patent mentioned
>>>>> above, ..." (BSD/GPL license)
>>>>>
>>>>>
>>>>> oooooooooooo José Luis text oooooooooooooooooooooooooooooooo
>>>>> # LEGAL WARNING:
>>>>> The source file "localcolorcorrection.cpp" implements the ALGORITHM
>>>>> COVERED BY U.S. PATENT 6,822,762. For this reason, this source file
>>>>> and its binary program "localcolorcorrection" may only be used for
>>>>> non-profit research and non-profit educational purposes.
>>>>> Redistribution
>>>>> or commercial use is not allowed without the consent of the patent
>>>>> owner.
>>>>>
>>>>> With the exception of the files mentioned above, redistribution
>>>>> and use
>>>>> in
>>>>> source and binary forms, with or without modification, are permitted
>>>>> under the terms of the GNU General Public License as published by
>>>>>
>>>>> the Free Software Foundation, either version 3 of the License, or
>>>>>
>>>>> (at your option) any later version. The program is distributed in
>>>>> the hope that it will be useful, but WITHOUT ANY WARRANTY;
>>>>> without even
>>>>> the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
>>>>> PURPOSE.
>>>>> See the GNU General Public License for more details
>>>>> http://www.gnu.org/licenses/.
>>>>>
>>>>> oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
>>>>>
>>>>> José Luis Lisani a écrit :
>>>>> Dear all,
>>>>>
>>>>> I think that the proposal of Jean-Michel of not making public the
>>>>> source
>>>>> codes
>>>>> of patented algorithms is the safest for IPOL, in front of
>>>>> hypothetical
>>>>> suits.
>>>>> Users can still run the algorithms on-line (are we sure that this
>>>>> does
>>>>> not
>>>>> infringe patent laws?) and they will have the guarantee by IPOL that
>>>>> what
>>>>> they run does indeed correspond to what is described in the web
>>>>> site.
>>>>>
>>>>> Moreover, I think that, as a matter of courtesy, the Editor of IPOL
>>>>> should send
>>>>> the authors of the patent a letter announcing them that there is an
>>>>> online version
>>>>> of their algorithm. I don't think we should ask for specific
>>>>> consent of
>>>>> the authors,
>>>>> since, in most cases, they don't even own the invention, their
>>>>> companies
>>>>> do.
>>>>>
>>>>> José Luis
>>>>>
>>>>>
>>>>>
>>>>> El 06/04/2011 7:42, Jean-Michel Morel escribió:
>>>>>
>>>>> Dear all,
>>>>>
>>>>>
>>>>> Should we ask the advice of a patent attorney as suggests Nicolas?
>>>>> It is probably a good idea.
>>>>> However, unfortunately, I know some of these guys. They always give
>>>>> the most conservative advice. According to the CNRS patent attorney,
>>>>> to publish the NL-means code without the owner's consent IS an
>>>>> infringement because we help companies to infringe it, even
>>>>> though we
>>>>> mention that it is patented.
>>>>>
>>>>> We need a conservative solution showing our good faith. Here is
>>>>> what I
>>>>> propose:
>>>>>
>>>>> -systematically, the patented algorithms published in IPOL will be
>>>>> sent to their authors as referees. That way we will get a first hand
>>>>> information about whether and how the patent is enforced. We will
>>>>> have
>>>>> to follow the authors' advice.
>>>>>
>>>>> In case of touchy patent owners, we will only transmit the code
>>>>> to the
>>>>> referees under NDA. IPOL will therefore run online a code
>>>>> certified by
>>>>> referees, but which will not be public.
>>>>>
>>>>> In other terms IPOL will only publish (actually following its
>>>>> definition) "Algorithms", with as a bonus an implementation
>>>>> certified
>>>>> by referees, and running on line. The code of this implementation
>>>>> will
>>>>> be certified but not public. The online facility will be
>>>>> installed in
>>>>> such a way (limited size, etc.) as to prevent any real or massive
>>>>> use.
>>>>>
>>>>> Still, with all this, a patent owner could object that we are
>>>>> helping
>>>>> counterfactors who could do online reverse engineering with our
>>>>> code.
>>>>> But, at this point, I think our display of good faith will be solid,
>>>>> and it will all remain compatible with the IPOL definition that IPOL
>>>>> publishes algorithms and shows how they work.
>>>>>
>>>>> -Meanwhile, we never met the bad guys and we can continue disclosing
>>>>> the code. I agree with Pascal's proposal to add the text (enclosed
>>>>> below) to every patented algorithm.
>>>>>
>>>>> I do not think we can put any license on patented code without
>>>>> incurring in some risk. Thus I agree that we should separate the
>>>>> patented part and put on it the warning proposed by Pascal.
>>>>>
>>>>>
>>>>> Nevertheless, it is not enough to write "for research and
>>>>> educational purposes": we must write "for non profit research and
>>>>> nonprofit educational purposes", because there is lucrative research
>>>>> and there is lucrative education.
>>>>>
>>>>> I disagree with Nicolas' proposal to put a BSD or other licenses,
>>>>> and
>>>>> invite users to decide by themselves if it is legal or not depending
>>>>> where they live. Putting a license at this point would be confusing.
>>>>> As a publisher we are supposed to know the law. It is our
>>>>> responsibility to avoid any ambiguity about rights.
>>>>>
>>>>> Best, JM
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Pascal Getreuer a écrit :
>>>>> _____________________________________________________
>>>>> Copyright (c) 2011, Your Name Here
>>>>> All rights reserved.
>>>>>
>>>>> The source file "dmha.c" implements the algorithm covered by U.S.
>>>>> Patent
>>>>> 5629734. For this reason, the source files "dmha.c", "dmha.h",
>>>>> "dmhacli.c" and its binary program "dmha" may only be used for
>>>>> research
>>>>> and educational purposes. Redistribution or commercial use is not
>>>>> allowed without the consent of the patent owner.
>>>>>
>>>>> With the exception of the files mentioned above, redistribution and
>>>>> use in
>>>>> source and binary forms, with or without modification, are permitted
>>>>> provided that the following conditions are met:
>>>>> (the rest of usual BSD license follows)
>>>>> _____________________________________________________
>>>>>
>>>>> For files implementing the patent:
>>>>> "may only be used for research and educational purposes" and
>>>>> "redistribution or commercial use is not allowed without the consent
>>>>> of the patent owner."
>>>>>
>>>>> For all the other files:
>>>>> "With the exception of the files implementing the patent mentioned
>>>>> above, ..." (BSD/GPL license)
>>>>>
>>>>> So the BSD/GPL only applies to the files that are not covered by
>>>>> patents.
>>>>>
>>>>> Having this separate distinction is useful for example for the
>>>>> self-similar demosaicking code, where Hamilton-Adams is used as an
>>>>> initialization but actually the majority of the code is for the
>>>>> self-similar demosaicking algorithm itself. The license specifies
>>>>> just the Hamilton-Adams part as "research and education use only, no
>>>>> redistribution" and everything else as GPL.
>>>>>
>>>>> Best,
>>>>> Pascal
>>>>> _______________________________________________
>>>>> discuss mailing list
>>>>> discuss at list.ipol.im
>>>>> http://tools.ipol.im/mailman/listinfo/discuss
>>>>>
>>>>> _______________________________________________
>>>>> discuss mailing list
>>>>> discuss at list.ipol.im
>>>>> http://tools.ipol.im/mailman/listinfo/discuss
>>>>>
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> discuss mailing list
>>>>> discuss at list.ipol.im
>>>>> http://tools.ipol.im/mailman/listinfo/discuss
>>>>> _______________________________________________
>>>>> discuss mailing list
>>>>> discuss at list.ipol.im
>>>>> http://tools.ipol.im/mailman/listinfo/discuss
>>>>>
>>>> _______________________________________________
>>>> discuss mailing list
>>>> discuss at list.ipol.im
>>>> http://tools.ipol.im/mailman/listinfo/discuss
>>>>
>>>> _______________________________________________
>>> discuss mailing list
>>> discuss at list.ipol.im
>>> http://tools.ipol.im/mailman/listinfo/discuss
>>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> discuss mailing list
>> discuss at list.ipol.im
>> http://tools.ipol.im/mailman/listinfo/discuss
> _______________________________________________
> discuss mailing list
> discuss at list.ipol.im
> http://tools.ipol.im/mailman/listinfo/discuss
More information about the discuss
mailing list