Summary

  • Peter Hessler welcomed a reference-only index of RIPE policy and asked that Database Working Group material be included.
  • He argued for a second, separate step: identify conflicting definitions before anyone edits accepted policy text.

A policy index sounds like filing. The hard part begins when two documents point in different directions and someone has to decide what the wording means.

Remco van Mook brought that problem to the RIPE 93 Address Policy Working Group with what he called a “new kitchen”: an index to sort the policy set by subject and make its structure easier to see. He described a tangle of old and new documents, inconsistent terms and proposals caught between two familiar objections. One draft is too complicated; the next version leaves out somebody’s corner case. Both objections can be true, and the proposal can sit there while the room argues over which failure matters more.

Van Mook said his first step would be modest. The index would point to existing RIPE documents; it would not move or rewrite their policy text. Chapters could group IPv4, IPv6, ASNs, transfers, special-purpose resources and related subjects. An appendix could later collect history and operational context. The session transcript records an invitation to discuss that framework, not a formal adoption, owner or delivery date.

Peter Hessler, speaking as co-chair of the RIPE Database Working Group, welcomed a place for database documentation. He also drew a boundary that a filing plan can easily blur. Collecting references is straightforward. Reconciling conflicting definitions is substantive work. He wanted conflict collection treated as an early, separately visible step, then repeated that collation and correction should remain separate. A list of documents answers “where is the rule?” A conflict record answers “which texts collide, and who needs to resolve the collision?”

That distinction matters because editing a policy can change how people apply it. In the same discussion, Tobias Fiebig raised minimum IPv6 assignment sizes as an example of definitions that may differ across documents. The transcript names no document numbers or clauses; comparing those exact provisions would establish whether the example is a direct conflict or a difference that needs explanation.

RIPE’s current Policy Development Process, RIPE-781, sets four phases for policy development: proposal, discussion, review and conclusion. It describes an open, transparent process that reaches decisions by consensus and places policies in RIPE Documents. That process gives a route for changing policy. A well-maintained index would give readers a route to the documents before they propose a change.

The distinction was already contested in the room. Van Mook called the reference index a sequence of “baby steps.” Maynard Koch warned that even a later clarification could change how policy is applied, while Gert Döring said rewritten language needs confirmation that it preserves the original intent. The argument was not about whether a better catalogue would help. It was about who could decide that two old texts meant the same thing.

Hessler’s intervention was practical. The official Database Working Group page describes the group’s remit around database changes, features, distribution and security. Bringing those documents into the index makes the map more useful. Recording conflicts separately gives each working group a visible handoff instead of asking a maintainer to quietly harmonise language across documents.

The session record ends without a named owner or date for that conflict inventory. The next useful artifact is easy to describe: keep the reference index faithful to source documents, list each suspected conflict with exact document and section references, then send interpretation or correction to the appropriate process. Hessler’s line protects both halves of the job. A map should stay a map; the decision about what the map reveals must be made in the open.

Sources