Re: DogCows or Polymorphism in the Hurd

[Marcus Brinkmann]

At Tue, 7 Feb 2006 12:17:26 +0100,
Patrick Negre <address@hidden> wrote:
> 
> I have read all the thread on DogsCows, and i can't see why it's not possible 
> to avoid binding views to new names.

Nobody said it is.  You have to bind a view to a name if you want the
view to be nameable.  You suggest making the binding process-local,
ie, making the file name space per-process.  That's something you can
do, but this just means that you make the bindings process-local, it
does not mean that you avoid binding views to names.  In particular,
you do not avoid creating "new names" this way: Every process gets new
names for all files.  So, in your solution, you always create new names.

The DogCow discussion was primarily and originally about the
interfaces of an object, not about namespaces.  The question how the
namespace is organized, and if it is per-process or shared, is
orthogonal to the question how the interfaces of an object are
organized.  However, it is true that the namespace discussion is
important and related.

Thanks,
Marcus