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

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

Charles E. Perkins a écrit :
> Hello Alex,
>
> Some comments and corrections below...
>
> Alexandru Petrescu wrote:
> > There are number of issues which have been discussed recently about each
> > document.
> >
> > I know for my own point, but not sure how to re-make it.  If I make it
> > again and gets lost then it's a waste of time.
> >
> > http://tools.ietf.org/html/draft-baccelli-autoconf-adhoc-addr-model-03
> >
> > 1.  I think Charles had text proposal to me and onlist about replacing
> > "undetermined properties" and that proposal sounded better to me, except
> > that he assumed that an LL address is derived from link-layer
> > characteristics (the MAC adddress),
> I did not assume this.
> > and thus he discarded LLs
> I did not discard LLs
> > ; at which
> > point I disagreed saying that not discard LL, because LL can be like
> > fe80::1.
> I pointed out that this was totally irrelevant.
>
> > I think he got tired of me at that point so we
> > left it there.
> If I stopped discussing when I was tired, you might
> not have heard from me for many many months.
>
> >   LAter Charles repeatedly said he had nothing particular against LLs.
> Later, and earlier, and even earlier, and ...

You have nothing against IPv6 LLs or nothing against IPv4 LLs or both?

BEcause current draft still says "discourage IPv4 LL".

Or when you have nothing against something it means that you actually 
discourage it?

> > 2.  Also - I strongly oppose "discouraging LL IPv4" which the current
> > baccelli draft says.  Just type ifconfig or ipconfig on your offtheshelf
> > machine with wireless interfaces and see the IPv4 LL in action.  I 
> > think that's support enough.
>
> Is that a valid metric for utility in a wireless ad hoc network?

It's wireless links for adhoc networks - yes.  It's 169.254. yes.  It's 
widely spread on your off-the-shelf machines - yes.

> > 3.  I also oppose in draft-baccelli the exclusive use of /128 prefixes,
> > it makes no sense, I could explain further if listened, or noted in an
> > issue tracker.
>
> It makes a lot of sense.  In fact it makes enough sense so that
> the multiplicity of early proposed solutions to the problem could
> be developed into standards.

Do we make 3 issues out of these or should I consider that you convinced 
me and I give up?

Alex

> If only.



> Regards,
> Charlie P.