Re: A simple question

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



i'm sorry to come into this late.  something went by a while ago that's
quite interesting.  sherman, set the wayback machine for...

> > % >>> So, unless there's some routing revolution, you're just going to
> > % >>> have to learn to deal with multiple addresses, and unless LL
> > % >>> addresses go away, you're also going to have to deal with scoped
> > % >>> addresses.

having recently done some work to describe and differentiate what i now
call a "dns-aware application", the above statement strikes me as more
relevant than most of the discussion that has followed it.

an application which has to be aware of ip address content in any way is
a different beast altogether than one which treats them as opaque.  from
a C programmer's point of view, the difference can be determined by
searching the module's source for "s6_addr" and "sin6_".  if neither
appears then you're probably not dealing with an "address aware" module.
if either appears, then you'll need to look carefully at it to find out.

an "address aware" program (module, library, application, whatever) has
to know about things like LL scopes, and currently, about SL issues.  any
other, non-"address aware" program, just receives "address blobs" from one
place and carries them around opaquely until it's time to offer them up on
some other altar, never knowing or caring what's "in the bag."

that being the case, i think it's "just plain wierd" that anybody is
arguing about the conceptual or architectural or complexity cost of SL
since the number of "address aware" programs is expected to remain quite
low in v6, perhaps lower as an overall share than in v4.  and for those
few programs, LL scopes and SL issues will not be a noticeable share of
the overall cost or complexity of the program.

this is not to paint myself as a supporter (nor detractor) of SL.  it's
possible that SL is as bad as people say, or as misunderstood, or whatever.
i just don't think it's reasonable to use application complexity as an
indicator of SL's cost:benefit ratio, since most applications Just Won't
Care.

(when i first heard of SL i thought, oh good, it's like rfc1918 but at the
campus level rather than at the autonomous system level.  them someone told
me it wasn't really like that, so now i'm opinionless.)
-- 
Paul Vixie


[Index of Archives]     [IETF Annoucements]     [IETF]     [IP Storage]     [Yosemite News]     [Linux SCTP]     [Linux Newbies]     [Fedora Users]