Anticipating gh #502 — TIPC over a WireGuard mesh as our go-to
multihost tpt deployment — plus a cold-start handoff for whoever
(or whatever) picks this up next.
The wg deats, both verified locally,
- a wg iface is L3/`tun` (`POINTOPOINT,NOARP`, `link/none`, no
L2 addr) so TIPC's `eth` media **cannot** bind it; the udp
bearer is *mandatory* over wg, not merely an alternative.
Also its ~1420 MTU sits under ethernet's 1500.
- the composed deployment maddr is
`/ip4/<pub>/udp/51820/wg/u<key>/tipc/<stype>/<inst>/<scope>`.
XXX note the tipc segment has NO locative part unlike tcp's
inner `/ip4/../tcp/..` — a service name is
location-independent, so wg carries routing and tipc carries
identity. That's the argument for one `/tipc` proto w/ a
structured value in the #498 spec proposal.
XXX ALSO correcting a premise: TIPC is **not** unencrypted. It
ships AES-GCM crypto (`tipc node set key`, linux 5.9+) w/
cluster/master/per-node keys + rekeying. Those keys are
symmetric+pre-shared tho, so wg is still preferred for
public-key identity, NAT traversal, and one overlay every tpt
can share.
`01_tipc_HANDOFF.md` is deliberately provider-neutral: env
setup, the hard-won kernel facts table, the two closed design
decisions (+why), what landed, the pre-land TODOs and the repo's
working conventions.
(this patch was generated in some part by `claude-code` using `claude-opus-5` (`anthropic`))
Bring the shared backend contract and TIPC plan in line with
current runtime behavior and downstream implementation evidence.
Rework the QUIC plan around actor-owned endpoint, bootstrap,
stream, listener and cleanup lifecycles, with explicit validation
gates for the still-unverified UniFFI details.
(this patch was generated in some part by `opencode` using `gpt-5.6-sol` (`openai`))
Landing page for `ai/tpt-backends/`: points at the contract spec
as required first reading, tables the 3 plans against their
issues/deps/size, and states the landing order + why.
Deats,
- TIPC first as the cheap proof the table-registration story
generalizes to a genuinely new proto (stdlib-only, and
`trio`'s sock wrappers are family-agnostic).
- `wg` layer-A next since it's deployable-today doc/example work.
- QUIC last, gated on its own prep PR.
- notes that plans 01 and 02 both want the same
`Address.rebind_from_sockname` gate, so whichever lands first
ships it.
(this patch was generated in some part by `claude-code` using `claude-opus-5` (`anthropic`))
Gud Boi