`wg` multiaddr protocol: upstream spec submission plan

[goodboy]

wg multiaddr protocol: upstream spec submission plan

Plan for registering the wg (WireGuard) protocol upstream
once the py-multiaddr impl (PR
multiformats/py-multiaddr#108) is settled.

Settled impl semantics (post-review)

• protocol name: wg
• code: 0x01c7 (decimal 455) — the unassigned slot
  between noise (0x01c6) and shs (0x01c8) in the
  multiaddr tag range.
• binary form: exactly 32 bytes — a raw Curve25519 public
  key; fixed size 256 bits, NO varint length prefix and no
  multibase envelope on the wire (verified: ...c703 varint
  code + 32 raw bytes).
• string form: multibase base64url — u prefix + 43
  unpadded chars (44 total), matching the certhash
  convention in py-/go-multiaddr.
  • standard base64 (as printed by wg(8)) can't appear in
    the /-delimited textual form; convert via
    multibase.encode('base64url', base64.b64decode(k)).
  • padded (u...=) and non-u multibase prefixes are
    rejected by the py impl — spec text should state whether
    canonical-only-u is the contract or decoders should
    accept any multibase and re-emit u (open Q for
    maintainers).
  • key semantics: impl length-checks only; spec should say
    whether degenerate keys (e.g. all-zero low-order point)
    are the codec's concern (suggest: no — semantics stay
    out of the codec, like other maddr protocols).
• example:
  /ip4/1.2.3.4/udp/51820/wg/urCzgw_sszhsg-SpR8V5IPmVnYPg2PliWW739_SWemwY

Step 1: multicodec PR (reserves the code)

Repo: https://github.com/multiformats/multicodec

Add one row to table.csv (cols:
name, tag, code, status, description), keeping the file's
column alignment, sorted by code next to noise/shs:

wg,                             multiaddr,      0x01c7,         draft,      WireGuard tunnel endpoint - 32-byte Curve25519 public key

• run the repo's table validation locally before pushing
  (make / npm test per their CONTRIBUTING docs).
• per their addition process, draft status entries for
  unclaimed codes are routinely accepted via small PRs.
• PR body: link WireGuard cryptokey-routing docs, py-multiaddr
  PR Multiple issues on the code examples present in the README file. #108 as the first implementation, and the overlay-network
  use case from issue General formatting  #107.

Step 2: multiaddr spec PR (defines the protocol)

Repo: https://github.com/multiformats/multiaddr

1. Add a row to protocols.csv (cols:
   code,\tsize,\tname,\tcomment), sorted by code after
   noise (454):

   455,	256,	wg,	WireGuard tunnel endpoint (Curve25519 public key)

2. If requested by maintainers, add a short protocol
   description to the README/spec covering:

   • value = 32-byte Curve25519 public key (binary, fixed
     size, no length prefix); multibase base64url u<43>
     (string).
   • rationale: / in std base64 collides with the maddr
     delimiter; multibase matches the certhash precedent
     across implementations. Conversion from wg(8)
     output: multibase.encode('base64url', base64.b64decode(key)).
   • canonical composition: /ip{4,6}/<host>/udp/<port>/wg/<key>.
   • open Qs to settle in review: canonical-only-u vs
     accept-any-multibase decode; degenerate-key semantics
     (see gh/pr108_review_followups.md).

Open this PR referencing the multicodec PR from step 1 so
both land with the same code.

Step 3: circle back to py-multiaddr

After (or alongside) the upstream PRs:

• update multiaddr/codecs/wg.py module docstring + PR Multiple issues on the code examples present in the README file. #108
  body to point at the upstream PRs instead of calling the
  code a speculative draft.
• tick the "Submit a draft multicodec addition PR upstream"
  TODO checkbox in the PR Multiple issues on the code examples present in the README file. #108 description.

Follow-ups (optional)

• propose the same protocol to go-multiaddr /
  js-multiaddr once the spec rows land, so the string/binary
  forms stay interoperable across impls.