Files
ng-eventually/docs/fork-inbox-fallback.md
T
Sylvain Duchesne 63ecfeeff8 docs+refactor(client): fidelity pass — id identity, drop connections, no faux-login, accurate NextGraph framing
Align the polyfill's surface and docs with the verified NextGraph reality and
remove application-level concepts:

- Identity is an ID, not a username: AccountRecord.id, shim predicate shim:id,
  normalizeId; accounts core becomes IdentityStore (set/clear/get) — the faux
  login/logout framing is gone (identity is set at wallet-import time).
- Relationship/connection is an application concept, not a platform primitive
  (NextGraph has no bilateral-connection primitive: grantee is unpersisted
  scaffolding, cap-send is unimplemented). Remove connections.ts; caps exposes
  only a directed grantRead(doc, granteeId) + a read-only protectedDocsOf(owner).
  Delete the now-dead isolation.ts social-visibility axis.
- Inbox docs: NextGraph has no separate curator — the recipient's own verifier
  unseals and applies each queued sealed message inline (process_inbox);
  inbox_post_link is a proposed/future API. Stop attributing the emulated
  curator to the platform.
- Read isolation reframed around the outcome: no cap -> empty union read;
  targeted read of an unheld repo -> RepoNotFound; cap introspection
  (canRead/governsRead) is emulation-only with no NextGraph API behind it.
- read-model.md corrected: the listing path is per-doc ANCHORED default-graph
  queries, never the anchorless GRAPH ?g union (that is O(wallet)); the probe
  section no longer claims the opposite.
- README recap table restructured (target | current NextGraph status | current
  emulation); INDEX_ACCOUNT documented as reservedAccount("index") in the
  sentinel namespace; de-domained generic-layer comments; softened tone.

Consumer application (Festipod) rewired separately to own the relationship
concept and feed the lib an id. Lib gates: bun test 83 pass / 0 fail, tsc clean.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-06 14:02:16 +02:00

94 lines
5.1 KiB
Markdown

# Fallback — forking NextGraph to expose the inbox (path not taken)
**Status:** not taken — short-circuited by this lib's emulated inbox
(`inbox.ts`, see [`simulation.md`](./simulation.md)). Kept as the fallback plan if
a native broker inbox ever becomes necessary — chiefly for the crypto
anonymity the emulation does not provide (native `from = None` sealed deposit).
Current NextGraph does not expose the inbox to the JS SDK: the verifier has no
`InboxPost` arm and no wasm helper seals a deposit (see
[`nextgraph-current-state.md`](./nextgraph-current-state.md) § Inbox). Two ways to
get a real inbox: emulate it (what this lib does) or fork the engine (this
document). The emulation was chosen; this is the archived alternative.
## Strategic posture
The fork would be **explicitly temporary, not for upstream**. Hypothesis:
NextGraph will eventually expose its **own** JS-SDK inbox solution, possibly
different. When it lands, drop the fork and adapt. No PR is targeted. This posture
is the reason the emulation was preferred: it avoids maintaining a fork + hosting
a full stack for a feature the upstream will likely ship differently.
## Layer 1 — the Rust patch: 4 files (vanilla broker)
1. **`engine/net/src/types.rs`** — `InboxMsgContent::Link` is a **unit** variant
(stub); give it a payload (or a `Notification` variant) carrying the target
NURI + link to the deposited reference. Add an `InboxPost::new_link(...)`
builder modeled on `new_contact_details`. `from = None` → anonymity.
2. **`engine/verifier/src/request_processor.rs`** — add the missing command arm
(there is no `InboxPost` arm). Ideally a high-level command (`NotifyInbox`) that
builds the post on the Rust side (keeps crypto sealing in Rust). Model on
`SocialQueryStart`.
3. **`sdk/js/lib-wasm/src/lib.rs`** — expose `pub async fn inbox_post_link(
session_id, to_inbox_nuri, to_profile_nuri, link, anonymous)`, modeled on
`social_query_start`.
4. **`engine/verifier/src/inbox_processor.rs`** (`process_inbox`) — a receive arm
where the owner's own verifier unseals the queued message and applies it inline
into the owner's store (model on the `ContactDetails` handler). The app then
reads via ORM/SPARQL — no new inbox-read API.
Identity resolution (known/anonymous): free via app-side SPARQL (JOIN the
sender inbox NURI against `social:contact` docs). Discovering the owner's
inbox: embed the owner's `public_store` inbox NURI in the entity document or
public profile (the QR profile-share flow already carries it).
## Layer 2 — deployment (from the fork)
The patched verifier runs **in the iframe ng-app** (see integration model in
[`nextgraph-current-state.md`](./nextgraph-current-state.md)) → **build and
self-host `ngd` + the ng-app** from the fork, then rebuild Festipod's `@ng-org/web`
with `NG_REDIR_SERVER`/`NG_DEV*` pointing at that ng-app. **No rewrite of the
third-party integration** (stays iframe). The broker's inbox routing is already
native, but since the patched ng-app is self-hosted, the **whole stack ships from
the fork** (one source tree).
### Coolify hosting — 3 web pieces
1. **`ngd`** — stateful WebSocket daemon: container with a **persistent volume**
for `--base-path` (RocksDB + keys + PeerId, never wiped), `--domain` mode behind
Coolify's Traefik. Build: official Dockerfiles are broken → write a **custom
multi-stage Rust Dockerfile** (RocksDB needs llvm/clang). First boot is
**interactive** (admin-wallet invitation link) → script via `ngcli` or do it
once by hand then persist in the volume.
2. **ng-app** (iframe frontend, patched wasm) — **static build**
(`pnpm webfilebuild`). Served statically.
3. **Routing**: one domain serves the ng-app static AND proxies the WebSocket to
`ngd`.
### Layer 1 (JS libs) — patched client npm packages
Maintain patched client packages, not just the wasm. The generic forwarding
*technically* reaches a wasm method without touching the JS, but that's an untyped
**hack** — quick test only. To modify for real:
- **`@ng-org/web`** — modified anyway (broker URL) → add `inbox_post_link` to the
typed API surface + `.d.ts`.
- **Streamed methods** (if inbox *reads* ever stream) — need an entry on both
sides. For write-only (request/response) — unneeded.
- **`@ng-org/orm`** — only if inbox writes join the ORM flow. Otherwise unneeded.
## Layer 3 — consumer application integration
Exposing the method is not enough; the consumer application must model the entity +
its inbox NURI, write the registration, deposit into the host inbox, and read/resolve
notifications. Several of these are already done in the shared-wallet emulation
(registration wired on the emulated `inbox.post`), which is precisely why this fork
was not needed.
## Why this fallback still matters
The emulated inbox stores `from = null` as *absence of a triple*; it does not seal
deposits, so it does not provide the target's crypto anonymity. If a consumer
application needs true anonymous-but-verifiable deposits to a non-connected host,
only a native inbox (`from = None` sealed) delivers it — and this fork is the route.
Until then, the emulation is sufficient.