Re: [Int-dir] side note RFC 4291 2nd par sec. 2.1 LL on loopback

[Alexandre Petrescu <[e-mail]>]

Le 17/04/2019 à 17:20, 神明達哉 a écrit :
> At Wed, 17 Apr 2019 16:33:53 +0200,
> Alexandre Petrescu <[e-mail] 
> <mailto:[e-mail]>> wrote:
> 
>  > >  > [...]> Note also that MacOS is derived from BSD in case you don't
>  > >  > > remember/know it.  And, in fact it also generates both 
> "fe80::1" and
>  > >  > > "::1" on the "lo0" interface.
>  > >  >
>  > >  > Right, I forgot that.
>  > >  >
>  > >  > But there is something I never knew for sure.
>  > >  >
>  > >  > Are BSD flavors supporting OCB options?
>  > >
>  > > I don't know, but I don't think it matters here anyway:
>  >
>  > Maybe it matters, maybe it does not matter.
>  >
>  > If BSD does not support OCB, why should we write an IP-over-OCB spec
>  > along the lines of BSD which does not accept fe80:1::1 in the first 
> place?
> 
> I don't get it.  This sub-thread is not about fe80:1::1 but about
> whether to assign an fe80 address on a loopback interface.
> 
> But to answer the question: because we don't know if that assumption
> holds in future or we don't know if there are other platforms that
> strictly assume the RFC4291 format but yet implement the IP-over-OCB
> spec.

Right, we may expect a future platform to stricly abide to RFC4291, and 
implement OCB too.

For now, we know that linux supports OCB and fe80:1::1.  My trial with 
multiple single-link subnets uses fe80:1::/32, fe80:2::/32 and 
fe80:3::/32.  They are all on linux-based openwrt.

I would like this to interoperate with others and in the future too.

> We should either make IP-over-OCB compliant/consistent with
> existing standards, or make it clear that IP-over-OCB makes an
> exception to the existing standards by formally updating the latter.
> That's, in my understanding, what a responsible spec author of an IETF
> document is supposed to do, especially if they really respect the
> concept of interoperability.  And that's the point of my very first
> message in this entire int-dir thread.

I am interested in interoperability.

I would like to test linux talk to BSD in OCB mode.

The first thing that blocks is lack of OCB support in BSD.  After that 
is fixed we can talk fe80:1::1 in BSD.

Another first thing that blocks is that I do not know how to make an 
"Update: 4291" statement on the first page of this document.

Also, the AD suggestion is to stay silent about plen like 64 or 118 or 
so, and deal with this elsewhere.

Alex
> 
> Frankly, I'm surprised to hear that an author of a would-be standards
> track document says something like this:
> 
>  > If BSD does not support OCB, why should we write an IP-over-OCB spec
>  > along the lines of BSD which does not accept fe80:1::1 in the first 
> place?
> 
> To me, this sounds like the author saying they don't care about
> interoperabitlity beyond the platforms they are interested in.
> 
> --
> JINMEI, Tatuya