Commit Graph

3 Commits

Author SHA1 Message Date
e54ae3afb9 feat(rust): mod configuration from the site, and an editor that will not rewrite a float
All checks were successful
PR Checks / server-tests (pull_request) Successful in 18s
PR Checks / frozen-manifest (pull_request) Successful in 51s
PR Checks / client-build (pull_request) Successful in 7m56s
R18's two tiers: a form generated from a config file's own values, and raw JSON
for what a form cannot express. Admin → Rust mod config, one live round trip per
action, nothing cached between a browser and a game host's disk.

`configEdit.js` is the part that could not be done naively. JavaScript cannot
tell `1` from `1.0`, and both mod frameworks deserialize a config into typed C#
classes — so a read-modify-write silently rewrites every whole-numbered float as
an integer on fields nobody touched, and a plugin that then throws at load does
not come back. It never parses, mutates and re-serialises: it records the SOURCE
SPAN of every value and splices literals into them, so an untouched `1.0` is
still `1.0` and a number an admin types travels as text the whole way (D35/D36).

The bridge's own config is editable with `Host`, `Port` and `ServerId` locked,
in the form and in the raw tier, because either would cut the link carrying the
edit or strand every row this site holds (D38). Credentials render masked with a
reveal; the raw tier shows them (D37) and the audit trail never does.

`rust_config_writes` records every save including the refused and the rolled
back — an operator asking why a setting is not what they set needs to see that
somebody tried.

Three defects a browser walk found that 179 green tests did not:

* every save of the bridge's own config was refused while the page said the
  opposite — a `<select>` whose value matches no `<option>` shows the first one,
  so the reload guess `RunicGateway` was on the wire and "nothing" was on the
  screen;
* `btn ghost` is not a class this platform defines (`.btn-ghost` is), so every
  secondary button in this module has rendered as a primary one since phase 7 —
  here it made the open file and the active tier indistinguishable;
* a save's refusal rendered at the top of a long form, far from the button.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMH6bw1jXMgbyF3ZWGEzSM
2026-09-22 08:55:28 -05:00
43147b796a feat(rust): site-owned permissions — the site is the author, the game is the cache
R2, and the first phase where this module WRITES to a game. Groups and grants are
authored on the website and pushed into each server's own permission store, so
every plugin that already calls `UserHasPermission` honours them with no adapter,
and a wipe stops being a data-loss event.

**Seven org-lead decisions (D28-D34).** A grant is keyed to the website USER and
resolved to every Steam id they have linked at push time (D28); every authored row
carries a scope — a server or `*` (D29); groups are mirrored as real groups rather
than flattened (D30); a holder the site did not author is REPORTED, never undone,
with adopt and revoke offered (D31); one verb, with the plugin diffing locally
(D32); a permission no server has registered is reported unresolved and never
self-registered (D33); authoring is people and groups by hand, with rules deferred
(D34).

**Three sets, and every interesting question is a difference between two.**
`desired − pushed` is what to apply; `pushed − desired` is what to RETIRE, because
the site put it there and has since withdrawn it; `present − desired` is drift. The
middle one is why `rust_perm_pushed` exists: a name in the store that is not in the
desired set is either something the site retired or something a human granted, and
those two have opposite correct answers.

**What lands is not what was sent.** A grant naming a permission the server has not
registered did not land — `GrantUserPermission` no-ops silently — and a member the
store has never seen could not be placed. Neither is recorded as pushed, so the
site never believes it gave a privilege it did not.

The loop asks a cheap question every thirty seconds — does the digest of the
desired set still equal what this server last confirmed — and syncs on a change, a
restart, a wipe, a drift hook, a failed attempt past its backoff, or the
fifteen-minute audit that finds drift on a server nobody has touched.

**This module's first admin page**, because a permission model is the first thing
here that has to be composed rather than configured. What is on it is decided by
what an operator can get wrong: four states are invisible from the game and from a
list of grants, and each is a sentence rather than a number.

Walked end to end against a real core at the pinned ref, the real sidecar, and a
stand-in speaking protocol 4 — including a restart that emptied the store and was
fully re-pushed. Four defects the browser found that 133 green tests did not.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMH6bw1jXMgbyF3ZWGEzSM
2026-09-21 18:28:32 -05:00
baffaa46c9 feat(rust): identity — a link code from the game, and the Steam id inside core's user page
R1's identity link, site-side, and R13's first extension slot. A player types
/link in game, the plugin hands them a six-character code privately, and they
enter it here; the site records who owns which Steam account, and an operator
sees that on core's own `/admin/users/:id` page.

**The site is the author of record and the game holds nothing.** There is no
per-account store in Rust that survives a wipe, and phase 7 needs the site
authoritative anyway — it pushes permissions INTO the game keyed by Steam id. A
copy in the game would be a second thing to reconcile every wipe, for no question
it could answer better.

## D24 — a code is minted by ONE server, so every server is asked

Nothing in six characters says where it came from. The fleet is asked in turn and
the first `link.ok` wins; the others answer `unknown` and nothing happens there,
because a code is only spent at the server that actually holds it. Asking the
player to pick was rejected: a wrong pick would come back indistinguishable from
a wrong code, and that is the one refusal which must not be ambiguous.

**"Every reachable server refused" is not the same answer as "a server was
unreachable."** Collapsing them tells a player whose server is down that their
code is wrong — so they run /link again on that same server and are told the same
thing for as long as it stays down. `unsure` is that case, and it says to try
again rather than to fetch a new code.

## D23 — a Steam id another account holds is refused, never moved

The primary key is `steam_id`, and it is load-bearing rather than tidy: phase 7
grants permissions against a link and phase 13 hangs entitlements off it, so a
silent move is an account takeover performed by typing six characters. The
refusal names the holder, because the advice is unusable without it. The INSERT
is a plain INSERT for the same reason — `ON DUPLICATE KEY UPDATE` here would BE
that move — and the duplicate-key error is the refusal for the race the check
above cannot close.

The way out is `/unlink` in game, which reaches the site off the ingest feed
rather than through a route (the plugin has no link to delete). D25 adds the
other way out: staff can sever a link from the admin panel, for a player who
cannot reach that Steam account in game.

## The slot, and the hole it found in this repo's own generator

`admin.users.detail` is declared in `module.json` AND registered in `index.js`
AND filled by the chunk — three places, because the server half and the client
half are different registrations that share one name.

`swaggerFragment.js` knew only about tier routers, so the two routes under
`/admin/users/:id` were generated by nothing: a fragment that was internally
consistent and described two routes fewer than the module serves. A slot's mount
is core's and cannot be derived here, so it is a fourth constant beside
`TIER_BASE` — held to account by the frozen-manifest job, which was verified to
catch exactly this by removing the two paths and watching it fail.

## Smaller things worth knowing

- **Core's `useAsync` has no `refresh`.** A counter in the deps is how a page
  re-reads after its own write; it blanks while it re-reads, which is right here
  and is exactly what made it wrong for a poll.
- **Every player-portal nav row needs an `icon`** — core draws one on every row,
  and the client suite says so. This module had no icons file until now, because
  the public header is text buttons.
- The two new frame kinds are STAFF-only. Neither carries a code, but both name a
  Steam id beside a website account's activity, and that join is not a public
  fact about what happened on a server.
- The link code route carries its own rate limiter rather than core's
  `accountChangeLimiter`: this is guessing somebody else's secret, not changing
  your own password, and a shared counter would let one policy set the other.

Protocol 3 on all three declaration sites; 17 new tests, 136 green.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMH6bw1jXMgbyF3ZWGEzSM
2026-09-21 08:18:48 -05:00