plan: 0086 cost-flag CLI ergonomics

Four tasks from spec 0086: (1) named non-negativity diagnostic via a parse_nonneg_rate helper + perturbed-test fix; (2) non-R-harness guard (the spec_gap, reject/exit-2); (3) COST_FLAGS_NOTE appended to the usage + --help surfaces; (4) the C10 ledger note + full-suite/clippy gate. Confined to crates/aura-cli; the eight stage1-r cost goldens stay byte-green. refs #153
2026-06-29 11:35:03 +02:00
parent f26b33f345
commit dfc01bf837
1 changed files with 422 additions and 0 deletions
@@ -0,0 +1,422 @@
+# Cost-flag CLI ergonomics — Implementation Plan
+
+> **Parent spec:** `docs/specs/0086-cost-flag-cli-ergonomics.md`
+>
+> **For agentic workers:** REQUIRED SUB-SKILL: use the `implement` skill to run
+> this plan. Steps use `- [ ]` checkboxes for tracking.
+
+**Goal:** Make the `aura run` cost-flag surface honest — units discoverable, a
+named non-negativity diagnostic, and a refuse-don't-guess rejection of cost flags
+on a non-R harness — with one C10 ledger note, changing no cost-model/R math.
+
+**Architecture:** All code lands in the pure `parse_run_args` grammar (and two
+print sites) in `crates/aura-cli/src/main.rs`; every new rejection is a pure
+`Result::Err` printed by `main`'s existing `eprintln!("aura: {msg}")` + `exit(2)`
+path (main.rs:3273-3279). One realization note is added to contract C10 in the
+ledger. The accepted stage1-r cost path and its eight `net_expectancy_r`/bleed
+goldens stay byte-identical.
+
+**Tech Stack:** Rust; `crates/aura-cli/src/main.rs` (the `aura` binary + its
+`mod tests`), `crates/aura-cli/tests/cli_run.rs` (integration), `docs/design/INDEX.md`.
+
+---
+
+**Files this plan creates or modifies:**
+
+- Modify: `crates/aura-cli/src/main.rs` — new private `parse_nonneg_rate` fn + new
+  `const COST_FLAGS_NOTE`; rewire the three cost-flag arms (:3185-3211); insert the
+  non-R-harness guard (between :3215 and :3224); append the note to the `usage`
+  closure (:3137-3140) and the `--help` print (:3265).
+- Test: `crates/aura-cli/src/main.rs` `mod tests` (:3338) — pure `parse_run_args`
+  units (negative-rate, non-R-harness, accepted-path, malformed-regression).
+- Modify: `crates/aura-cli/tests/cli_run.rs` — update the perturbed negative-carry
+  test (:1825-1841); add non-R-harness, negative-cost, and `--help`-note integration
+  tests beside the cost block (:1631-1888) and the help block (:595-663).
+- Modify: `docs/design/INDEX.md` — one C10 realization note at the cost-section tail
+  (after `sweep-path cost. Decision log: #148.`, :831).
+
+---
+
+### Task 1: Named non-negativity diagnostic (spec item 2)
+
+**Files:**
+- Modify: `crates/aura-cli/src/main.rs` (new `parse_nonneg_rate`; arms :3185-3211)
+- Test: `crates/aura-cli/src/main.rs` `mod tests` (:3338)
+- Test: `crates/aura-cli/tests/cli_run.rs` (update :1825-1841; add one negative-cost test)
+
+- [ ] **Step 1: Write the failing unit tests**
+
+Add to `mod tests` in `crates/aura-cli/src/main.rs`, beside the existing
+`parse_run_args_*` units (near :4480):
+
+```rust
+#[test]
+fn parse_run_args_negative_cost_rates_are_named_not_usage() {
+    for flag in ["--cost-per-trade", "--slip-vol-mult", "--carry-per-cycle"] {
+        let err = parse_run_args(&["--harness", "stage1-r", flag, "-0.5"])
+            .expect_err("a negative cost rate must be refused");
+        assert!(err.contains("must be non-negative"), "{flag}: {err}");
+        assert!(err.contains(flag), "message must name the offending flag: {err}");
+        assert!(err.contains("-0.5"), "message must carry the value: {err}");
+    }
+}
+
+#[test]
+fn parse_run_args_malformed_cost_rate_stays_grammar_usage() {
+    let err = parse_run_args(&["--harness", "stage1-r", "--cost-per-trade", "x"])
+        .expect_err("a malformed rate is a grammar error");
+    assert!(err.contains("usage"), "a malformed value stays the usage path: {err}");
+}
+```
+
+- [ ] **Step 2: Run the unit tests to verify they fail**
+
+Run: `cargo test -p aura-cli parse_run_args_negative_cost_rates_are_named_not_usage`
+Expected: FAIL — the current arms `return Err(usage())` on a negative value, so the
+message is the usage string and `err.contains("must be non-negative")` is false.
+
+- [ ] **Step 3: Add the `parse_nonneg_rate` helper**
+
+Insert immediately above `fn parse_run_args` (currently `crates/aura-cli/src/main.rs:3136`):
+
+```rust
+/// Parse a cost-rate flag value and enforce non-negativity, naming the offending
+/// flag so a sign typo is distinguishable from a mistyped flag name. A *malformed*
+/// value stays a grammar usage error (unchanged); only a parsed-but-negative value
+/// yields the named message. Pure — no I/O, no exit.
+fn parse_nonneg_rate(flag: &str, value: &str, usage: &impl Fn() -> String) -> Result<f64, String> {
+    let v: f64 = value.parse().map_err(|_| usage())?;
+    if v < 0.0 {
+        return Err(format!("{flag} must be non-negative, got {v}"));
+    }
+    Ok(v)
+}
+```
+
+- [ ] **Step 4: Rewire the three cost-flag arms to call the helper**
+
+In `parse_run_args`, replace the `--cost-per-trade` arm (`main.rs:3185-3193`):
+
+```rust
+            "--cost-per-trade" if cost.is_none() => {
+                let (value, t) = t.split_first().ok_or_else(usage)?;
+                cost = Some(parse_nonneg_rate("--cost-per-trade", value, &usage)?);
+                tail = t;
+            }
+```
+
+the `--slip-vol-mult` arm (`main.rs:3194-3202`):
+
+```rust
+            "--slip-vol-mult" if slip_vol_mult.is_none() => {
+                let (value, t) = t.split_first().ok_or_else(usage)?;
+                slip_vol_mult = Some(parse_nonneg_rate("--slip-vol-mult", value, &usage)?);
+                tail = t;
+            }
+```
+
+and the `--carry-per-cycle` arm (`main.rs:3203-3211`):
+
+```rust
+            "--carry-per-cycle" if carry_per_cycle.is_none() => {
+                let (value, t) = t.split_first().ok_or_else(usage)?;
+                carry_per_cycle = Some(parse_nonneg_rate("--carry-per-cycle", value, &usage)?);
+                tail = t;
+            }
+```
+
+- [ ] **Step 5: Run the unit tests to verify they pass**
+
+Run: `cargo test -p aura-cli parse_run_args_negative_cost_rates_are_named_not_usage`
+Expected: PASS.
+Run: `cargo test -p aura-cli parse_run_args_malformed_cost_rate_stays_grammar_usage`
+Expected: PASS (a malformed value still routes through `usage()`).
+
+- [ ] **Step 6: Update the now-perturbed integration test**
+
+The existing `stage1_r_negative_carry_per_cycle_refused_with_usage_exit_2`
+(`crates/aura-cli/tests/cli_run.rs:1825-1841`) asserts `stderr.contains("usage")`,
+which the new named message no longer satisfies. Update its assertion and name.
+
+Replace the assertion line (`cli_run.rs:1838`):
+
+```rust
+    assert!(stderr.contains("must be non-negative"), "negative carry stderr must name the cause: {stderr:?}");
+```
+
+Rename the function (`cli_run.rs:1826`) and adjust the doc line that says
+"usage on stderr":
+
+```rust
+/// Property (cycle 0085 + #153, cost-flag validation symmetry): a NEGATIVE
+/// `--carry-per-cycle` is refused at the binary boundary — exit 2, a named
+/// non-negativity diagnostic on stderr, nothing on stdout, no `runs/` write — the
+/// same shape the sibling `--cost-per-trade`/`--slip-vol-mult` flags enforce. The
+/// `v < 0.0` guard is load-bearing: without it a negative carry reaches
+/// `CarryCost::new` and PANICs, and a negative price-unit carry would CREDIT R —
+/// inverting the accrual sign. Deterministic; the refusal precedes any data access.
+#[test]
+fn stage1_r_negative_carry_per_cycle_refused_non_negative_exit_2() {
+```
+
+- [ ] **Step 7: Add a sibling negative-cost integration test**
+
+Add beside it in `crates/aura-cli/tests/cli_run.rs` (after the carry test, ~:1842):
+
+```rust
+/// Property (#153): a negative `--cost-per-trade` is refused at the binary
+/// boundary — exit 2, a named non-negativity diagnostic on stderr, no `runs/`
+/// write — the sibling of the carry guard, exercising the per-trade arm.
+#[test]
+fn run_negative_cost_per_trade_refused_non_negative_exit_2() {
+    let dir = temp_cwd("cost-per-trade-neg");
+    let out = Command::new(BIN)
+        .current_dir(&dir)
+        .args(["run", "--harness", "stage1-r", "--cost-per-trade", "-0.5"])
+        .output()
+        .expect("spawn aura run --cost-per-trade -0.5");
+    assert_eq!(out.status.code(), Some(2), "negative cost must exit 2; stderr: {}",
+        String::from_utf8_lossy(&out.stderr));
+    assert!(out.stdout.is_empty(), "a refused run must not emit a report: {:?}", out.stdout);
+    let stderr = String::from_utf8_lossy(&out.stderr);
+    assert!(stderr.contains("must be non-negative"), "stderr must name the cause: {stderr:?}");
+    assert!(!dir.join("runs").exists(), "a refused run must not write the registry");
+    let _ = std::fs::remove_dir_all(&dir);
+}
+```
+
+- [ ] **Step 8: Run the integration tests to verify they pass**
+
+Run: `cargo test -p aura-cli --test cli_run stage1_r_negative_carry_per_cycle_refused_non_negative_exit_2`
+Expected: PASS.
+Run: `cargo test -p aura-cli --test cli_run run_negative_cost_per_trade_refused_non_negative_exit_2`
+Expected: PASS.
+
+---
+
+### Task 2: Non-R-harness guard (spec item 3 — the spec_gap)
+
+**Files:**
+- Modify: `crates/aura-cli/src/main.rs` (insert guard after :3215)
+- Test: `crates/aura-cli/src/main.rs` `mod tests` (:3338)
+- Test: `crates/aura-cli/tests/cli_run.rs` (new integration test beside :1574-1589)
+
+- [ ] **Step 1: Write the failing unit tests**
+
+Add to `mod tests` in `crates/aura-cli/src/main.rs`, beside the Task-1 units:
+
+```rust
+#[test]
+fn parse_run_args_cost_flags_require_r_harness() {
+    let cases: [(&[&str], &str); 3] = [
+        (&["--harness", "sma"], "--cost-per-trade"),
+        (&["--harness", "macd"], "--slip-vol-mult"),
+        (&["--harness", "sma"], "--carry-per-cycle"),
+    ];
+    for (harness, flag) in cases {
+        let mut argv: Vec<&str> = harness.to_vec();
+        argv.extend_from_slice(&[flag, "0.001"]);
+        let err = parse_run_args(&argv).expect_err("cost flag on a non-R harness must be refused");
+        assert!(err.contains("R-evaluator harness"), "{harness:?}+{flag}: {err}");
+    }
+}
+
+#[test]
+fn parse_run_args_cost_flag_without_harness_defaults_sma_and_is_refused() {
+    let err = parse_run_args(&["--carry-per-cycle", "0.5"])
+        .expect_err("the default harness is sma → cost flag refused");
+    assert!(err.contains("R-evaluator harness"), "{err}");
+}
+
+#[test]
+fn parse_run_args_cost_flags_accepted_on_stage1_r() {
+    let a = parse_run_args(&["--harness", "stage1-r", "--cost-per-trade", "0.001"])
+        .expect("cost flags are accepted on the R harness");
+    assert!(matches!(a.harness, HarnessKind::Stage1R));
+    assert_eq!(a.cost, Some(0.001));
+}
+```
+
+- [ ] **Step 2: Run the unit tests to verify they fail**
+
+Run: `cargo test -p aura-cli parse_run_args_cost_flags_require_r_harness`
+Expected: FAIL — today a cost flag on `--harness sma` is accepted (silently
+ignored downstream), so `parse_run_args` returns `Ok` and `expect_err` panics.
+
+- [ ] **Step 3: Insert the non-R-harness guard**
+
+In `parse_run_args`, immediately after `let harness = harness.unwrap_or(HarnessKind::Sma);`
+(`main.rs:3215`) and before the `// --from/--to require --real.` comment:
+
+```rust
+    let harness = harness.unwrap_or(HarnessKind::Sma);
+    // Cost flags are defined only against the R chain (gross-R -> net-R); a non-R
+    // harness produces no R to charge against, so a cost flag there is a usage error,
+    // not a silent no-op (refuse-don't-guess, C10).
+    if !matches!(harness, HarnessKind::Stage1R)
+        && (cost.is_some() || slip_vol_mult.is_some() || carry_per_cycle.is_some())
+    {
+        return Err(
+            "cost flags require an R-evaluator harness (stage1-r); \
+             --harness sma/macd produces no R to charge against"
+                .to_string(),
+        );
+    }
+```
+
+- [ ] **Step 4: Run the unit tests to verify they pass**
+
+Run: `cargo test -p aura-cli parse_run_args_cost_flags_require_r_harness`
+Expected: PASS.
+Run: `cargo test -p aura-cli parse_run_args_cost_flags_accepted_on_stage1_r`
+Expected: PASS (the accepted path is unchanged).
+
+- [ ] **Step 5: Write the failing integration test**
+
+Add to `crates/aura-cli/tests/cli_run.rs`, beside `run_harness_sma_is_pip_only_no_r_block`
+(~:1589):
+
+```rust
+/// Property (#153): a cost flag on a non-R harness (sma/macd, which produce no R)
+/// is a usage error — exit 2 with a named diagnostic, no silent no-op, no `runs/`
+/// write. The pre-#153 behaviour silently ignored the flag and exited 0.
+#[test]
+fn run_cost_flag_on_non_r_harness_is_refused_exit_2() {
+    let dir = temp_cwd("sma-cost-refused");
+    let out = Command::new(BIN)
+        .current_dir(&dir)
+        .args(["run", "--harness", "sma", "--cost-per-trade", "0.001"])
+        .output()
+        .expect("spawn aura run --harness sma --cost-per-trade");
+    assert_eq!(out.status.code(), Some(2), "cost on a non-R harness must exit 2; stderr: {}",
+        String::from_utf8_lossy(&out.stderr));
+    assert!(out.stdout.is_empty(), "a refused run must not emit a report: {:?}", out.stdout);
+    let stderr = String::from_utf8_lossy(&out.stderr);
+    assert!(stderr.contains("R-evaluator harness"), "stderr must name the cause: {stderr:?}");
+    assert!(!dir.join("runs").exists(), "a refused run must not write the registry");
+    let _ = std::fs::remove_dir_all(&dir);
+}
+```
+
+- [ ] **Step 6: Run the integration test to verify it passes**
+
+Run: `cargo test -p aura-cli --test cli_run run_cost_flag_on_non_r_harness_is_refused_exit_2`
+Expected: PASS.
+
+---
+
+### Task 3: Cost-flag units note (spec item 1)
+
+**Files:**
+- Modify: `crates/aura-cli/src/main.rs` (new `const COST_FLAGS_NOTE`; usage closure :3137; `--help` print :3265)
+- Test: `crates/aura-cli/tests/cli_run.rs` (new `--help` note test beside :595-663)
+
+- [ ] **Step 1: Write the failing integration test**
+
+Add to `crates/aura-cli/tests/cli_run.rs`, beside the help tests (~:663):
+
+```rust
+/// Property (#153): the cost-flag units/constraints are discoverable from
+/// `aura run --help` — the appended note states the in-R charge, the per-ENGINE-
+/// cycle carry semantics, the `>= 0` constraint, and that cost flags are stage1-r
+/// only. Pre-#153 the usage listed the flags with no unit or scale guidance.
+#[test]
+fn run_help_surfaces_cost_flag_units_note() {
+    let out = Command::new(BIN).args(["run", "--help"]).output().expect("spawn aura run --help");
+    assert_eq!(out.status.code(), Some(0), "run --help exits 0: {:?}", out.status);
+    let stdout = String::from_utf8(out.stdout).expect("utf-8 stdout");
+    assert!(stdout.contains("per ENGINE cycle"), "help must explain carry units: {stdout}");
+    assert!(stdout.contains("stage1-r only"), "help must state the R-harness requirement: {stdout}");
+}
+```
+
+- [ ] **Step 2: Run the integration test to verify it fails**
+
+Run: `cargo test -p aura-cli --test cli_run run_help_surfaces_cost_flag_units_note`
+Expected: FAIL — the current `--help` output (`USAGE`) carries no cost-flag note,
+so `stdout.contains("per ENGINE cycle")` is false.
+
+- [ ] **Step 3: Add the `COST_FLAGS_NOTE` const**
+
+Insert immediately above `const USAGE: &str` (`main.rs:3244`):
+
+```rust
+/// One-line note appended to the `aura run` usage surfaces (the grammar-error
+/// usage closure and `--help`) so the cost-flag units, scale, per-engine-cycle
+/// semantics, non-negativity, and R-harness requirement are discoverable at the
+/// point of use (#153). A leading newline puts it on its own line under the usage.
+const COST_FLAGS_NOTE: &str = "\ncost flags (stage1-r only; all >= 0, charged in R as cost/|entry−stop|): \
+--cost-per-trade = per-trade cost in price units; --slip-vol-mult = slippage multiplier on realized vol; \
+--carry-per-cycle = carry in price units per ENGINE cycle (not per day, not an overnight swap)";
+```
+
+- [ ] **Step 4: Append the note to the `parse_run_args` usage closure**
+
+Replace the `usage` closure (`main.rs:3137-3140`):
+
+```rust
+    let usage = || {
+        format!(
+            "usage: aura run [--harness <sma|macd|stage1-r>] [--real <SYMBOL> [--from <ms>] [--to <ms>]] [--trace <name>] [--cost-per-trade <f64>] [--slip-vol-mult <f64>] [--carry-per-cycle <f64>]{COST_FLAGS_NOTE}"
+        )
+    };
+```
+
+- [ ] **Step 5: Append the note to the `--help` print**
+
+Replace the `--help` short-circuit print (`main.rs:3265`):
+
+```rust
+    if args.iter().any(|a| a == "--help" || a == "-h") {
+        println!("{USAGE}{COST_FLAGS_NOTE}");
+        return;
+    }
+```
+
+- [ ] **Step 6: Run the integration test to verify it passes**
+
+Run: `cargo test -p aura-cli --test cli_run run_help_surfaces_cost_flag_units_note`
+Expected: PASS.
+
+- [ ] **Step 7: Regression — the uniform help/usage pins still hold**
+
+Run: `cargo test -p aura-cli --test cli_run per_subcommand_help_is_uniform_stdout_exit_zero`
+Expected: PASS (the appended note keeps `contains("usage")` true and stderr empty).
+
+---
+
+### Task 4: C10 ledger note + full-suite gate
+
+**Files:**
+- Modify: `docs/design/INDEX.md` (C10 cost-section tail, after :831)
+
+- [ ] **Step 1: Add the C10 realization note**
+
+In `docs/design/INDEX.md`, replace the boundary between the cycle-0085 cost note
+and the #117 HISTORY reframe (`:831-833`):
+
+```markdown
+sweep-path cost. Decision log: #148.
+
+**Realization (cost-flag harness scoping, #153).** Cost flags are defined only
+against an R-evaluator harness (the gross-R → net-R chain). On a non-R harness
+(`--harness sma`/`macd`, which produce no R) a cost flag is a **usage error
+(exit 2)**, not a silent no-op — refuse-don't-guess. Negative cost rates are
+likewise rejected (exit 2) with a named diagnostic that identifies the offending
+flag. CLI ergonomics only; the cost-model graph and the R math are untouched.
+
+**Reframe (2026-06-23, #117 — exposure → bias, R as the signal-quality unit).**
+```
+
+- [ ] **Step 2: Full aura-cli suite gate — new tests pass, goldens byte-green**
+
+Run: `cargo test -p aura-cli`
+Expected: PASS — the whole `aura-cli` suite is green, including the eight accepted-path
+cost goldens (`stage1_r_flat_cost_net_expectancy_r_golden` et al., cli_run.rs
+:1686/1714/1736/1757/...) byte-unchanged, since only the grammar/usage strings and the
+previously-unguarded sma-cost path changed.
+
+- [ ] **Step 3: Lint gate**
+
+Run: `cargo clippy -p aura-cli --all-targets -- -D warnings`
+Expected: PASS — no warnings (the new helper, const, and guard are idiomatic).