[IPOL discuss] Patents and IPOL code

Jean-Michel Morel morel at cmla.ens-cachan.fr
Wed Apr 6 07:42:02 CEST 2011


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
> 


More information about the discuss mailing list