Re: Reverse Engineering was: Re: pdepe and pdeval

[Jordi Gutiérrez Hermoso]

Note: the following is not legal advice. I am not a lawyer. You should
consult a lawyer if you want a more informed answer than mine.

On Fri, 2013-11-29 at 19:52 -0700, ghaverla wrote:
> On Fri, 29 Nov 2013 21:17:36 -0500
> Jordi Gutiérrez Hermoso <address@hidden> wrote:
> 
> > I only know of one instance of this code for Octave, and there
> > were copyright problems with it, because the code was based on
> > Matlab, so we had to take it down.
> > 
> > It doesn't seem like the functions are difficult to implement, but
> > if you choose to do so, it's important that you do NOT consult
> > Matlab code while writing them.
> 
> I would like more information/feedback on this.
> 
> Somebody in the geosciences wrote a matlab utility of marginal use
> to me (for an application that has nothing to do with the
> geosciences) . 

To be clear, the problem isn't if someone writes Matlab code, but who
wrote the Matlab code and if they provided a free license for it. If
you find code written for Matlab by someone who grants you a free
license to it, then there is no problem. The Mathworks cannot claim
copyright on the code merely because it's written in the Matlab
language.

When I said that the pdepe code was "based on Matlab", I meant that
the person in question had read Matlab's own source code, as written
and copyrighted by The Mathworks, and attempted to give us an
implementation of pdepe after reading this. This is dangerous enough
for us, because it is quite possible that the resulting code bears
strong similarity to Matlab's own source code.

> If I do a global search and replace on variable names, the code is
> still the same.  To me.  The core of the code is useful.

Still the same code to you and to everyone else.

Such minor tricks are not usually enough to declare that your code is
not derivative work of the original one under copyright law. Judges
are able to see through such tricks. The copyrightability of things is
always a fuzzy thing, up to interpretation of fickle judges and
juries. It is not a merely matter of running your code through some
simple program that transforms it trivially in order to declare that
your work is an independent creation not based on someone else's.

What we do know is safe is clean-room reverse engineering:

    http://en.wikipedia.org/wiki/Clean_room_design

The techniques in pdepe are not secrets, and even if they were, you
can't copyright an algorithm, only a particular expression of it. So
all we need is someone who can figure out how pdepe works without
reading Matlab's source code and then provide an independent
implementation of it.

HTH,
- Jordi G. H.