benilla-octo/docs/METHOD.md

METHOD — how benilla is built

benilla is a from-scratch World of Warcraft 1.12.1 client in Rust and Bevy. It reads the game's data from the player's own install, speaks the original protocol to any 1.12.1 server, and does not build a server or simulate the game. The reference client is the spec: benilla does what 1.12.1 does, as a modern, idiomatic client.

The rule for every change

A change is right when it makes benilla more like 1.12.1 or fixes a bug, with evidence from the reference, in one small piece, with the gates green. A deviation from the reference is the maintainer's call alone, and each one is written down where it lives: a CVar's Deviates row, a comment naming the reference fact and why we differ. Anything else is a fork.

The loop

1. The reference first. For anything fidelity-defining, get the mechanism from the real client before building: its behaviour, a DBC field, a FrameXML line, a packet capture. The names and shapes under reference/ are the surface benilla tracks.
2. Build it idiomatically. Modern Bevy and Rust, performance woven in from the start. Implement the mechanism well; do not ape a quirk that is a bug.
3. Measure, never eyeball. Timing, delay, ordering and feel are settled with instruments: trace timestamps, byte-derived numbers, a numeric probe. A capture confirms an existence fact; it cannot measure time. Never search for a viewpoint: take the one simplest shot, and if it does not settle the question, hand over a precise script ("log in, look at X, does Y?").
4. Close the loop on the symptom. A bug reported as seen is closed by reproducing it at the reported spot first, fixing, and showing the same instrument clean there plus one adjacent state. Green gates and an existing mechanism are never that evidence.
5. Prove the run before reading the result. A live result is evidence once the run is valid from numbers: preflight banner clean, camera and body where intended, the subject in frame, the window long enough. A negative from an unproven run is not evidence.
6. Commit small and often; land by pull request. Atomic commits, explicit paths, on a branch; main moves only by a squash-merged pull request, one commit per piece of work. No "done" and no commit without saying what was verified and how the result was judged.

Hard rules

• Never commit anything from the install. Art, models, sounds, maps, data: the install is read at runtime from a gitignored path, and everyone provides their own. The one exception is interface code: FrameXML and GlueXML run off the player's own patch chain, and our own counterparts under assets/ui stay until they retire.
• The install is read-only. benilla never writes into the WoW folder: no screenshot, log, cache or scratch file. A hook blocks such a write at the tool call, and scripts/smoke.sh fails a run that leaves the install changed.
• Local state lives in one folder, benilla-config/ beside the binary, and every path to it resolves through crate::local_state. Player settings are CVars persisted as a diff in benilla-config/config.toml.
• UI is stock-first: a window is migrated, not authored. The end state is the stock 1.12 FrameXML executed off the player's own chain. assets/ui does not grow: a test names its files and fails on a new one. A window is built by pointing benilla.toc at the stock file, deleting ours and building the engine verbs it calls, never stubbing one to make the file load.
• A setting's default is the reference's default. Every option boots at the stock 1.12 value. Shipping another value costs an explicit Deviates row with the reason, and a test fails a row that drifts either way.
• Our code is MIT OR Apache-2.0, which is a claim about provenance: no original client code, no bundled assets.

Where things live

• docs/METHOD.md, this file: rules and method. Never state, never a log.
• docs/MAP.md: what is built, generated from the code by scripts/genmap.sh at every land. Read it to orient; never edit or commit it by hand.
• git: the history. No changelogs, no status docs, no "what we did" narratives.
• The code: a comment says what the code does and the 1.12 fact behind it, in a line. No history, no stories.
• .claude/skills/: procedures with steps. probe runs the client unattended against the local server.

Gates

• The land gates: cargo fmt --all -- --check, cargo clippy --workspace --all-targets -- -D warnings, cargo test --workspace, the doc-link and render-pass lints, the player build without the dev feature with its own tests, and the engine boot checks, as scripts/gates.sh. It runs on the tree that lands, before it lands; it is memoized on the tree. The tests that read the install or the addon corpus skip where the data is absent, and the gate prints the count; where the data is, a skip is a failure (BENILLA_REQUIRE_DATA=1). The workspace tests then run a second time with WOW_DATA= (no install), so a test that reads the install without opening with wow_data_or_skip! fails here and not on a clone.
• Per round: scripts/check.sh, scoped to the changed crates and their dependents. Run each gate once per round; tee the output to a file and grep the file.
• A clean run is the fourth gate: scripts/smoke.sh, a live login and logout against the local server. For wire work the gate is a live run with a trace (WOW_MOVE_TRACE=<path>, tags filtered), never a capture: capture mode has no network.
• Platform seams: a cfg(target_os), a [target.'cfg(…)'] dependency, a #[link] or an extern "system" is invisible to the gates of the platform you are on. Say in the commit which platforms you built.
• One hook: no writes into the install. Everything else is a gate or discipline.

The local server

benilla talks to any 1.12.1 server: WOW_HOST names it (default localhost), WOW_USER and WOW_PASS together log in without typing, WOW_CHAR picks the character. There is no default account. These bind every session and every agent:

• A scripted run logs in as the account its checkout declares in .probe-identity (the three variables, one per line, never committed), or as the three variables name; never as a player's, because a login kicks whoever holds the account, and the client refuses a scripted login from a checkout on any account but the declared one.
• Every unattended run says so: WOW_UNATTENDED=1. An agent's run is silent: WOW_NOSOUND=1.
• No unattended combat probes.
• Anything about hostility, reaction colour, nameplates, threat, damage or breath runs with WOW_GM=off.
• Read the preflight banner before debugging anything else.

The mechanics are the probe skill.