+1 As I noted before, reclaiming values would require extensive verification that they are not in use, and would, *at best*, result in their being “reserved” until the pool of unassigned (but never used) values was nearly exhausted. And, as I recently reminded others elsewhere, the IETF does NOT maintain a list of “currently active” anything. RFCs reflect their status when published; code points reflect their status when assigned. The principle was first encoded into TCP - abandoned connection state sticks around until a new connection clears it out. We don’t “garbage collect” until we need new space. Joe > On Sep 30, 2019, at 7:47 PM, John C Klensin <john-ietf@xxxxxxx> wrote: > > Bob, > > I agree. Let me add one other thing that has not come up in > this discussion. Jon allocated some port numbers under NDAs > that obligated IANA to keep the purpose/description, and > sometimes the requester, private. In some cases, those code > point assignments were kept private only for a while, e.g., > until a planned protocol or product was mature enough to expose > to the community. Others, well, I don't know. I don't know if > any of the code points with restrictions on disclosure were > allocated in the low-order range, but it is a plausible > explanation for code points that are shown as allocated but > without any real description. > > Taking back and reusing port numbers, addresses, or any other > parameter that was (as far as we know) properly allocated at the > time, and allocated without an expiration date, and doing so on > the basis of a newly-invented principle, is bad business and, > IMO, to be avoided if possible. > > The right thing to do now is, as you suggest, almost certainly > nothing. Sadly, I also agree with your second reason even > though "new assignments... blocked by firewalls and middleboxes" > sounds to me like an admission that the Internet has evolved to > the point that we have abandoned one of the most important early > design principles, that of not requiring permission to introduce > new applications and other innovations. > > best, > john > > > --On Monday, September 30, 2019 14:36 -0700 Bob Hinden > <bob.hinden@xxxxxxxxx> wrote: > >> Eric, >> >>> On Sep 30, 2019, at 11:50 AM, Eric Vyncke (evyncke) >>> <evyncke@xxxxxxxxx> wrote: >>> >>> Masataka, Joe and Bob, >>> >>> I think we agree even if my wording was ambiguous: the >>> community should define 'what to do' with those 'any *' IP >>> protocols that are not specified anywhere. And the definition >>> could be "do not use" but follow the process to get a new IP >>> protocol with some 'fences' to avoid wasting the remaining >>> 42% of those IP protocol numbers. >>> >>> => the current 'ambiguous' situation does not seem too good >>> to me >> >> My take is doing anything isn't necessary. Two reasons: >> >> 1) We aren't close to running out. The registry shows: >> >> 143-252 Unassigned >> >> That a lot of room in the registry given the current >> assignment rate. >> >> 2) The second reason is that I think the reason for few IANA >> allocation requests in this registry is that it is likely that >> packets containing any new assignments will be blocked in >> firewalls and middle boxes. It's hard to get a new >> protocol deployed. I am doubtful this will change anytime >> soon. I suspect we will never run out, unless the Internet >> changes significantly. >> >> The most I can see doing is to ask IANA to let the IETF >> community know when we have reached some milestone, like 90% >> of the space has been assigned. >> >> Thanks, >> Bob >> >> >> >> >>> >>> -éric >>> >>> On 30/09/2019, 12:0䨳㸀㸀 ကഀഀ >
