Plan M4a: gatewayd in 15 tasks, with skeletons and given tests

Each task's tests were run against a reference at its end state; the end states were replayed from master in order with the gate at each step (650 to 762 tests); each skeleton compiles against its tests and fails them. The reference is kept off this machine. Lessons T27 (every wait in a test has a limit) and T28 (mutate the reference before hand-over) come from this work. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-09-23 19:05:44 -07:00
parent f38dc8d474
commit 0339dc13b2
69 changed files with 7790 additions and 3 deletions
@@ -0,0 +1,60 @@
+# M4a task 01: SHA-1 in `proto`
+
+**Branch:** `m4a` (run `git switch m4a`; `git status --short` must be empty, otherwise stop)
+**Commit subject:** `proto: sha1, for the WebSocket handshake check`
+
+## Goal
+
+`gatewayd` speaks WebSocket to Mattermost. The server proves it understood the handshake by
+sending back the SHA-1 of our key (RFC 6455, section 4.2.2), so we need SHA-1. It lives in
+`proto`, which has no dependencies of its own. It is used for that check and nothing else: it must
+never protect anything.
+
+## Files
+
+- Copy: `crates/proto/tests/sha1.rs`, and the skeleton `crates/proto/src/sha1.rs`
+- Modify: `crates/proto/src/lib.rs` (`pub mod sha1;`), `docs/implementer-log.md`
+
+## Interface (in the skeleton)
+
+```rust
+pub fn sha1(data: &[u8]) -> [u8; 20];          // written: new, update, finish
+
+pub struct Sha1 { state: [u32; 5], block: [u8; 64], filled: usize, length: u64 }
+impl Sha1 {
+    pub fn new() -> Sha1;                        // written: the five initial words
+    pub fn update(&mut self, data: &[u8]);       // todo
+    pub fn finish(self) -> [u8; 20];             // todo
+    fn compress(&mut self, block: &[u8; 64]);    // todo
+}
+```
+
+`length` counts **bits** fed so far. Each `todo!()` has its steps above it. No indexing that can go
+out of bounds: `w[i]` on the fixed `[u32; 80]` inside `compress` is fine; slicing `data` or `block`
+by a computed range is not (use `get`, `get_mut`, `split_at`). Use `as_chunks::<4>()` to read
+big-endian words.
+
+## Steps
+
+- [ ] **1. Copy.** `git switch m4a`, then
+  `cp docs/plans/M4a/files/crates/proto/tests/sha1.rs crates/proto/tests/` and
+  `cp docs/plans/M4a/files/crates/proto/src/sha1.rs crates/proto/src/`. Add `pub mod sha1;` to
+  `crates/proto/src/lib.rs`.
+- [ ] **2. See it fail.** `cargo test -p proto --test sha1`. Expected: it compiles, and 4 tests
+  fail with `not yet implemented`.
+- [ ] **3. Fill the three functions**, one at a time: `compress`, then `update`, then `finish`. Run
+  `cargo check -p proto` after each.
+- [ ] **4. See it pass.** `cargo test -p proto --test sha1`. Expected: 4 passed. The tests include
+  FIPS 180 and RFC 3174 vectors, a million `a`s, the lengths around the 64-byte block where the
+  padding changes shape (checked with `sha1sum`), and the RFC 6455 handshake example; every case is
+  also fed in pieces.
+- [ ] **5. Run the gate.** `cargo fmt --all`, then `make gate`. Expected last line: `gate: ok`.
+- [ ] **6. Log and commit.** `git add crates/proto docs/implementer-log.md Cargo.lock && git commit`
+
+## Done when
+
+- `cargo test -p proto --test sha1` passes; `make gate` prints `gate: ok`.
+
+## Stop and report if
+
+- A test vector fails and you cannot find why after checking `compress` against the comment twice.