[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