Re: [Autoconf] issues with draft-baccelli address model

[Alexandru Petrescu <alexandru.petrescu at gmail.com>]

Ulrich Herberg a écrit :
> On Thu, Oct 29, 2009 at 12:06 AM, Charles E. Perkins 
> <charles.perkins at earthlink.net> wrote:
> > [..] Of course!  They may apply only to a restricted subset of the 
> > wireless space of interest, and thus not solve the whole problem, 
> > but why wouldn't they be acceptable for the realm of their 
> > applicability?
>
> I fully agree with Charlie. For a subset of deployment scenarios, we 
> can very well use LLs (for example, if they guarantee unique MACs, or
>  if we guarantee a good source of entropy). And this is said in the
> DT draft:
>
> "When more specific assumptions can be made regarding the
> connectivity between interfaces, these SHOULD be considered when
> configuring subnet prefixes."
>
> So for _specific_ settings (like Alex's testbed), using LL will work 
> fine. However, the address model of the DT tries to be as _general_
> as possible and NOT to focus on a subset with specific assumptions.

My question is whether or not this type of use LL addresses working fine
in _some_ applications (in a restricted subset of deployment scenarios
in the wireless space of interest) is a valid case for AUTOCONF or not.

Alex

> Ulrich