From Bernard Lang <Bernard.Lang@inria.fr>  Sat Nov  3 10:50:26 2001
From: Bernard Lang <Bernard.Lang@inria.fr> (Bernard Lang)
Date: Sat, 3 Nov 2001 11:50:26 +0100
Subject: [Patents] Re: object code.
In-Reply-To: <01110311135900.19454@francesca.cphil.net>; from Philippe Coulonges on Sat, Nov 03, 2001 at 11:13:59AM -0500
References: <200111021901.OAA09833@world.std.com> <01110300064801.14191@francesca.cphil.net> <000b01c163fb$842a7280$418e193e@oemcomputer> <01110311135900.19454@francesca.cphil.net>
Message-ID: <20011103115026.A30295@margaux.inria.fr>

If object code is to be considered on the same level as source code,
as a mode of expression, then the trade secret was revealed by these
companies that published object code for reading DVDs on Windows and
Mac.

  Furthermore, one should not be fooled by reverse engineering
technology. It can bring the code back to a somewhat readable version.
But if the code has been produced by a sophisticated optimizing
compiler, no known technology will be able in gneral to retrieve the
original conceptual structure.  For example, techniques based on
finite differencing can improve performance considerably (even
reducing algorithmic complexity), while removing perspicuity of the
underlying ideas.  Same is true when separate modules (at source
level) are folded together in order to merge and optimize computation
steps.  Or, when dealing with more sophisticated copiler/generator
techniques ...  how do you retrieve the original context-free grammars
from an optimized version of the corresponding parser (I am not
talking about a table interpreter), especially if you do not know the
parser generator.  Or worse, how do you retrieve an original Esterel
program from the generated code.

   I am sure reverse engineering does miracles ... but only so far.
And in many cases, even if one can think the technology is achievable,
it has not been done, and is probably too expensive and not useful
enough to be done.

Bernard

On Sat, Nov 03, 2001 at 11:13:59AM -0500, Philippe Coulonges wrote:
> Le Vendredi  2 Novembre 2001 19:07, Nick Fleming a crit :
> > Philippe Coulonges wrote :
> > > That decision was right, there are no ideas disclosed in object codes.
> >
> > > Source codes are ideas and as such can't be patented and can't be 
> restricted
> > > from researches. Let us contribute in order to fund our best advocates to
> > > defend that point of view.
> >
> >     That was my first thought too. Then I realised that modern object files
> > can
> > contain so much stuff, especially files with debugging and browsing
> > information
> > in them that they could easily be converted to a more human readable form,
> > even source code. However, because of widespread use of macro's and
> > preprocessors and even errors in software, I think it is more correct to
> > say that object files may not contain a complete description of an idea or
> > algorithm.
> 
> I disagree. It does contain that complete description. Otherwise the program 
> could not run. But it does not disclose it. For, has it has been noted, there 
> may be a long way from an object code to the idea. You have to search that 
> idea, possibly with sophiticated tools.
> 
> The important point being the admittance that the ideas in fact dwell in the 
> source code, and the benefits we can get from this.
> - protection of disclosure under free speech
> - legalization of research of ideas by mean of reverse engineering
> - a strong argument against software patents.
> 
> We have to admit a limit between the ideas that need to stay free and the 
> implementation on a stupid machine. The source code/object code limit seems 
> natural to me.
> 
> CU
> CPHIL
> 
> _______________________________________________
> Patents maillist  -  Patents@liberte.aful.org
> http://liberte.aful.org/mailman/listinfo/patents

-- 
         Non aux Brevets Logiciels  -  No to Software Patents
           SIGNEZ    http://petition.eurolinux.org/    SIGN

Bernard.Lang@inria.fr             ,_  /\o    \o/    Tel  +33 1 3963 5644
http://pauillac.inria.fr/~lang/  ^^^^^^^^^^^^^^^^^  Fax  +33 1 3963 5469
            INRIA / B.P. 105 / 78153 Le Chesnay CEDEX / France
         Je n'exprime que mon opinion - I express only my opinion
                 CAGED BEHIND WINDOWS or FREE WITH LINUX


