On Thu, Mar 20, 2008 at 7:04 PM, Paolo Bonzini <bonzini at gnu.org> wrote:
> > he wrote the new code that probably has the same content as the
>  > original one.  [...]
>
>  >
>  > we simply have to trust people that they will not cheat. And they should
>  > be ready to sign it.
>
>  Of course.  But the bigger problem is the first.
>
>  In the case of accessors, for example, the original contributor could
>  have contributed a single change to add those accessors, or these
>  methods could be part of a bigger change.  The first case probably is
>  not even copyrightable; in the second the overall changeset is
>  copyrightable, and I wonder if a judge would be convinced  if they were
>  shown the exact same source code up to the MD5 checksum.
>
>  After all saying "a method that returns instance variable foo" is not
>  different from saying "a method that does ^foo".  It is not talking
>  about an idea, the idea *is* the implementation (all this of course
>  holds because the "first team" does see the source code, it does not
>  apply if there was English documentation that the "second team" could
>  directly look at).
>
>  In both cases, it looks like chasing windmills; you should get clear
>  advice if it isn't better to just leave these simple methods aside, with
>  attribution to the original author.
>
>  Paolo
>

The most clean way is to write more tests, delete SqueakL code and
then fix it. The problem here is that we are playing with the kernel
and we should do it very canny and we cannot do it at once.

-- Pavel