Summary

  • RFC 3499 made the nominal range 3400–3499 browsable, but numeric adjacency and inclusion in one summary did not make those publications one architecture, authority or current specification.
  • Its coarse [STANDARDS TRACK] marker, explicit warning that status could change and truthful “never issued” entry for 3400 separated archival identity from present applicability.

Read RFC 3499 from its own entry downward and the catalogue seems complete. One hundred positions fit between 3400 and 3499. Each line offers the confidence of a number, a name and a title. Then the last position refuses the pattern: “RFC 3400 was never issued.” The list is complete as a record of the range precisely because it preserves an absence.

That small act of restraint explains the whole document. Published in December 2003 by Sandy Ginoza at ISI, RFC 3499 called itself a slightly annotated list of the hundred-number interval. It was Informational, specified no Internet standard and included itself with the wonderfully economical description “This memo.” What it standardized was nothing. What it preserved was a way to find things.

The things were wildly unlike one another. The upper end contains a SONET protection-switching MIB, an RTP payload for uncompressed studio video, ATM-aware MPLS signalling and a PacketCable DHCP option. Farther down come IPv6 socket APIs, IDNA, STUN, SIP compression, TCP over wireless links, graceful MPLS restart, XML, DNS roles, mail reports, SNMP architecture and DDDS. Their numbers share a publishing neighbourhood. Their packets do not necessarily meet.

Publication order is therefore not a protocol stack. Adjacency is not dependency. A reader cannot infer that RFC 3498 uses RFC 3497, or that the whole block forms a release train. A listed document may depend on an RFC hundreds of numbers older, be updated by one thousands of numbers newer, or have no operational relationship to its nearest numeric neighbour.

RFC 3499 even warned against treating the snapshot as current authority. Its note said that many, but not all, RFCs were Proposed Standards, Draft Standards or Standards. Because those states could change, the summary recorded only the broad label [STANDARDS TRACK] and referred readers to the latest “Internet Official Protocol Standards” for current state and status. The index announced its own expiry boundary.

That boundary matters because an RFC number is permanent while interpretation moves. Published text is not silently rewritten when a successor arrives. A later RFC can update or obsolete it; errata can be reported and verified; an STD or BCP identifier can continue while its constituent RFCs change. The modern RFC Editor explains these relationships explicitly. A number identifies a publication. It does not promise that publication is the present specification.

Nor does the abstract close the evidence chain. RFC 3499 reproduced compact descriptions, sometimes reflecting corrections made during publication. Those descriptions are valuable for discovery. They are not substitutes for normative sections, referenced definitions, later errata, updated-by edges, implementation choices or packet traces. A search result can lead an engineer to a document. It cannot certify the code derived from it.

The history of the summary product reinforces the point. RFC 1099 began the every-hundred convention in December 1991. RFC 3299 covered 3200–3299. RFC 3399 never appeared; RFC 3499 and RFC 3599 did, and 3599 became the last issued summary. The pattern left holes because editorial habits do not possess the number line.

In 2013 an Internet-Draft proposed formally retiring the xx99 summaries. It said searchable abstracts had removed the need, that no one had complained about the missing product, and that numbers ending in 99 could return to ordinary assignment. The draft expired. It is good historical evidence of a transition, but it never became a binding RFC. Describing it as enacted policy would repeat the very category error this article resists: a record of intention is not the same as an effective rule.

A parallel transition happened to a different series. The xx00 “Internet Official Protocol Standards” snapshots were not the xx99 range summaries. RFC 7100 retired STD 1 as an official-standards summary because it was not current. RFC 7101 recorded replacement of xx00 snapshots by a web page and the release of future xx00 numbers. The shared direction was from periodic static booklets toward maintained online views; the products and authority paths remained distinct.

The web did improve freshness. A current document page can expose stream, status, errata, history, updates and obsolescence relationships without waiting for another hundred numbers. But interface freshness is still not deployment evidence. A perfectly maintained graph can say which specification supersedes another while knowing nothing about the firmware in a router, the option enabled on a server or the outcome seen by a customer.

The useful receipt ladder begins lower. First, a number is assigned or honestly left unused. Then a permanent document is published. A summary records its identity. Current metadata describes status and relations. A reader follows the dependency closure. An implementer selects requirements. Code is built, tested, deployed and configured. Finally packets, logs and service measurements show what ran. No rung inherits the proof of the next.

RFC 3400 is the best emblem for that discipline. A database obsessed with rectangular completeness might invent a placeholder document and call the row repaired. The archive instead records “never issued.” Nothing is missing from the truth. The missing thing is part of the truth.

That is also why RFC 3499 deserves a history of its own, rather than serving only as a container for better-known specifications. It captures an Internet large enough to need navigation but still willing to distinguish navigation from command. The catalogue did not claim that its hundred positions shared one purpose. It did not freeze a living standards process. It did not turn abstracts into running code.

An index is a visibility layer. It reduces the cost of finding evidence while remaining smaller than the reality it describes. Its institutional virtue lies in that thinness. It can preserve identity, sequence and absence without pretending to be a legislature, an architecture diagram or an operations dashboard.

The mistake begins when symbolic order is promoted into operational order: when a number becomes currentness, proximity becomes dependency, [STANDARDS TRACK] becomes an exact present status, or searchability becomes conformance. RFC 3499 supplied enough caveats to prevent each inflation. The rest is reader discipline.

Sources