CB-WP-0004 T01: fix environment friction at the root

CB-RES-0003 measured 84 turns and $15.33 — the largest mechanical
category — spent prefixing commands with `cd` and
`export PATH="$HOME/.cargo/bin:$PATH"`. Both causes are now fixed once
instead of per-leaf.

tools/repo.py resolves the repo root from __file__ and cargo from PATH
then the standard rustup locations. Every tool imports ROOT from it, so
the repo path is stated once rather than redefined in four files —
single source of fact, the rule DFD earned in InnerLoop v1.2.
rule-coverage and dep-weight now call enter_root(), which is why their
relative paths did not need rewriting one by one.

The Makefile derives REPO from MAKEFILE_LIST and resolves CARGO the same
way, so `make -C <repo> <target>` works from any directory with no
prefix.

make env-test is the positive control, and is in `make all`: every tool
runs from / with PATH=/usr/bin:/bin. Without it this fix could regress
silently and invalidate T05's measurement — the whole point of the
control loop.

dep-weight's "cargo not on PATH" error is kept rather than deleted. It
should now be unreachable, and --self-test asserts cargo_bin() resolves
unaided; a control that never fires is cheaper than a regression.

loop-lint failed on repo.py on its first run — a reporting tool with a
positive control but no --self-test entry point. Second time the gate
has caught work from its own pass within the hour.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
tegwick 2026-07-31 10:13:52 +02:00
parent 578dcbea78
commit 3f1dbac164
9 changed files with 254 additions and 30 deletions

View file

@ -1,73 +1,112 @@
# One command surface (InnerLoop §agentic-efficiency #3). Deterministic,
# greppable output; precursor of the `cb` CLI.
#
# CB-WP-0004 T01: every target here runs from a clean shell, from any
# directory, with no prefix. Invoke as `make -C <repo> <target>` from
# elsewhere. No target requires `cd` or `export PATH` — CB-RES-0003
# measured 84 turns and $15.33 spent on exactly those two prefixes.
CARGO := cargo
# Absolute path to this Makefile's directory, so recipes never depend on
# the caller's working directory.
REPO := $(patsubst %/,%,$(dir $(abspath $(lastword $(MAKEFILE_LIST)))))
.PHONY: check test sim bench bench-test coverage dep-weight cost cost-test cost-pin cost-budget cost-mix loop-lint self-tests loc all
# Locate cargo instead of requiring it on the inherited PATH. Mirrors
# tools/repo.py:cargo_bin() — kept in sync by `make env-test`.
CARGO := $(firstword $(shell command -v cargo 2>/dev/null) \
$(wildcard $(HOME)/.cargo/bin/cargo) \
$(wildcard /usr/local/cargo/bin/cargo) \
cargo)
export PATH := $(dir $(CARGO)):$(PATH)
PY := python3
TOOLS := $(REPO)/tools
# Every cargo recipe runs at the repo root; the shell does not persist cd.
IN_REPO := cd $(REPO) &&
.PHONY: check test sim bench bench-test coverage dep-weight cost cost-test cost-pin cost-budget cost-mix loop-lint self-tests env-test loc all
## fmt + clippy (deny warnings) + HashMap deny-lint
check:
$(CARGO) fmt --all --check
$(CARGO) clippy --workspace --all-targets -- -D warnings
$(IN_REPO) $(CARGO) fmt --all --check
$(IN_REPO) $(CARGO) clippy --workspace --all-targets -- -D warnings
## unit + scenario-format tests
test:
$(CARGO) test --workspace
$(IN_REPO) $(CARGO) test --workspace
## run all GROUND scenarios through cb-sim
dep-weight:
python3 tools/dep-weight.py
$(PY) $(TOOLS)/dep-weight.py
coverage:
python3 tools/rule-coverage.py
$(PY) $(TOOLS)/rule-coverage.py
# M-D2-CST (specs/CostAccounting.md). cost-test is the positive control and
# runs first: a cost number from an unverified collector is void.
cost: cost-test
python3 tools/cb-cost.py --composition --by-task
$(PY) $(TOOLS)/cb-cost.py --composition --by-task
cost-test:
python3 tools/cb-cost.py --self-test
$(PY) $(TOOLS)/cb-cost.py --self-test
# InnerLoop rules that are mechanically checkable (CB-WP-0003 T01).
loop-lint:
python3 tools/loop-lint.py
$(PY) $(TOOLS)/loop-lint.py
# Positive control for every reporting tool, per InnerLoop v1.1 Step 5.
self-tests:
python3 tools/cb-cost.py --self-test
python3 tools/loop-lint.py --self-test
python3 tools/rule-coverage.py --self-test
python3 tools/dep-weight.py --self-test
$(PY) $(TOOLS)/cb-cost.py --self-test
$(PY) $(TOOLS)/loop-lint.py --self-test
$(PY) $(TOOLS)/rule-coverage.py --self-test
$(PY) $(TOOLS)/dep-weight.py --self-test
$(PY) $(TOOLS)/repo.py --self-test
# T01 positive control: prove the environment fix, do not assume it. Runs
# every tool from a foreign working directory with a PATH that has no
# cargo on it. Before T01 this failed; if it fails again, the friction is
# back and CB-EV-0003's measurement is invalid.
env-test:
@cd / && env PATH=/usr/bin:/bin $(PY) $(TOOLS)/repo.py --self-test
@cd / && env PATH=/usr/bin:/bin $(PY) $(TOOLS)/rule-coverage.py --self-test >/dev/null \
&& echo " [ok ] rule-coverage runs from / with no cargo on PATH"
@cd / && env PATH=/usr/bin:/bin $(PY) $(TOOLS)/dep-weight.py --self-test >/dev/null \
&& echo " [ok ] dep-weight runs from / with no cargo on PATH"
@cd / && env PATH=/usr/bin:/bin $(PY) $(TOOLS)/cb-cost.py --self-test >/dev/null \
&& echo " [ok ] cb-cost runs from / with no cargo on PATH"
@cd / && env PATH=/usr/bin:/bin $(PY) $(TOOLS)/loop-lint.py --self-test >/dev/null \
&& echo " [ok ] loop-lint runs from / with no cargo on PATH"
@$(MAKE) -C $(REPO) coverage >/dev/null \
&& echo " [ok ] make -C <repo> works from any directory"
# CB-01/CB-02: live spend since the last commit.
cost-budget: cost-test
python3 tools/cb-cost.py --budget
$(PY) $(TOOLS)/cb-cost.py --budget
# CB-RES-0003 baseline: mechanical vs judgment turns.
cost-mix: cost-test
python3 tools/cb-cost.py --composition
$(PY) $(TOOLS)/cb-cost.py --composition
cost-pin: cost-test
python3 tools/cb-cost.py --pin fc76445 --composition --by-task
$(PY) $(TOOLS)/cb-cost.py --pin fc76445 --composition --by-task
sim:
$(CARGO) run -q -p cb-sim -- scenarios/ground/*.yaml
$(IN_REPO) $(CARGO) run -q -p cb-sim -- $(REPO)/scenarios/ground/*.yaml
## Criterion benches (AM-6/AM-7)
bench:
$(CARGO) bench -p games-ground
$(IN_REPO) $(CARGO) bench -p games-ground
## InnerLoop positive control: run every bench once, no measurement.
## Fails if a workload stalls or produces the wrong event count.
bench-test:
$(CARGO) bench -p games-ground --bench synthetic -- --test
$(IN_REPO) $(CARGO) bench -p games-ground --bench synthetic -- --test
## AM-2/AM-3 input: source LOC per crate (excludes tests would need tokei)
loc:
@for d in crates/cb-kernel crates/cb-events crates/cb-game-runtime games/ground tools/cb-sim; do \
@$(IN_REPO) for d in crates/cb-kernel crates/cb-events crates/cb-game-runtime games/ground tools/cb-sim; do \
printf '%-28s %s\n' $$d "$$(find $$d/src -name '*.rs' | xargs cat | grep -vcE '^\s*(//|$$)')"; \
done
all: check test sim coverage dep-weight self-tests loop-lint bench-test
all: check test sim coverage dep-weight self-tests env-test loop-lint bench-test

View file

@ -4,4 +4,29 @@ A rebuild from scratch simulation and games engine framework set up to assimilat
Licensed under the Target Revenue Source License (TRSL V1C1) — see [LICENSE](LICENSE); canonical text lives in the org's `target-revenue` repository.
## Running the gates
```sh
make all # every gate, from a clean shell
make -C /path/to/clay-borg all # …or from any other directory
```
**There is no environment setup step.** No `cd`, no `export PATH`, no
activation script. `make` locates the repo from its own path and `cargo`
from the standard rustup locations; the Python tools do the same via
`tools/repo.py`. The only prerequisites are a rustup toolchain and Python
3.11+.
This is deliberate and enforced: `make env-test` runs every tool from `/`
with a PATH containing no cargo, and `make all` includes it. CB-RES-0003
measured 84 agent turns and $15.33 spent prefixing commands with `cd` and
`export PATH` before that friction was fixed at the root (CB-WP-0004 T01).
Other useful targets: `make cost` (spend per task), `make cost-budget`
(spend since the last commit), `make cost-mix` (mechanical vs judgment
turns), `make loop-lint` (executable InnerLoop rules), `make self-tests`
(every tool's positive control).
## GROUND
The first product vertical is a virtual tabletop implementation of **GROUND — A Game of Bonds and Rivalry: DARVO Edition**. The boardgame itself (rules, editions, content) is at home in the sister repository **`ground-game`** — that repo is authoritative for what GROUND *is*; clay-borg implements the engine that runs it.

Binary file not shown.

View file

@ -40,7 +40,8 @@ except ModuleNotFoundError: # pragma: no cover - py<3.11
print("ERROR: needs Python 3.11+ for tomllib", file=sys.stderr)
sys.exit(1)
REPO = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
from repo import ROOT as REPO # noqa: E402 (single source of fact, T01)
PRICES = os.path.join(REPO, "benchmarks", "baselines", "model-prices.toml")
COMPONENTS = ("input", "output", "cache_read", "write_5m", "write_1h")
TASK_RE = re.compile(r"\bT\d\d\b")

View file

@ -22,10 +22,11 @@ Usage: python3 tools/dep-weight.py [--json] [--self-test]
import glob
import json
import os
import shutil
import subprocess
import sys
from repo import cargo_bin, enter_root
PACKAGE = "games-ground"
CONFIGS = {
"shipped-runtime": ["--no-default-features"],
@ -42,14 +43,19 @@ TARGETS = {
def crates(extra_args):
"""Third-party crates in the normal (non-dev) dependency graph."""
if not shutil.which("cargo"):
# T01: locate cargo rather than demanding the caller export PATH. The
# error below is kept as a positive control — it should now be
# unreachable on a machine with rustup installed, and a control that
# never fires is still cheaper than a regression.
cargo = cargo_bin()
if not cargo:
print(
"ERROR: cargo not on PATH. Try: export PATH=\"$HOME/.cargo/bin:$PATH\"",
"ERROR: cargo not found on PATH or in ~/.cargo/bin — is rustup installed?",
file=sys.stderr,
)
sys.exit(1)
out = subprocess.run(
["cargo", "tree", "-p", PACKAGE, "--edges", "normal", "--prefix", "none"]
[cargo, "tree", "-p", PACKAGE, "--edges", "normal", "--prefix", "none"]
+ extra_args,
capture_output=True,
text=True,
@ -114,6 +120,11 @@ def self_test():
# Targets must be present and numeric — a missing target would make
# the breach check vacuous.
# T01: this tool is useless without cargo, and used to demand the caller
# put it on PATH. Assert it resolves unaided.
check("cargo resolves without caller PATH setup", bool(cargo_bin()),
cargo_bin() or "NOT FOUND")
check("targets defined for every configuration",
set(TARGETS) == set(CONFIGS) and all(
isinstance(v, int) and v > 0 for v in TARGETS.values()),
@ -129,6 +140,8 @@ def self_test():
def main():
# T01: `cargo tree` and the own-source walk are both repo-relative.
enter_root()
if "--self-test" in sys.argv:
return self_test()

View file

@ -24,7 +24,7 @@ import os
import re
import sys
REPO = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
from repo import ROOT as REPO # noqa: E402 (single source of fact, T01)
LOADABILITY_LIMIT = 400
# Artifact classes the loop produces. history/ is an append-only trail

124
tools/repo.py Normal file
View file

@ -0,0 +1,124 @@
#!/usr/bin/env python3
"""Repo location and toolchain discovery — the root fix for CB-WP-0004 T01.
CB-RES-0003 measured 84 turns and $15.33 spent on `cd` and
`export PATH="$HOME/.cargo/bin:$PATH"`. That friction had two causes:
1. tools resolved their inputs relative to the *caller's* working
directory, so every invocation had to be preceded by a `cd`;
2. `cargo` is not on the default PATH, so every Rust-touching command had
to be preceded by an `export`.
Both are fixed here, once, rather than at each leaf. `tools/dep-weight.py`
previously carried its own "cargo not on PATH" message evidence the
friction was noticed and patched in the wrong place.
Single source of fact (InnerLoop v1.2): ROOT is derived here and imported;
it is not recomputed per tool.
"""
import os
import shutil
# tools/repo.py -> tools -> repo root.
ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
# Where a rustup install puts the toolchain when PATH has not been set up.
CARGO_FALLBACKS = (
os.path.expanduser("~/.cargo/bin"),
"/usr/local/cargo/bin",
)
def enter_root():
"""Run from the repo root regardless of where the caller invoked us.
Every tool calls this first. After it, plain relative paths
(`specs/...`, `scenarios/...`) are correct, which is why the tools
below did not need rewriting path-by-path.
"""
os.chdir(ROOT)
return ROOT
def cargo_bin():
"""Absolute path to cargo, or None.
Checks PATH first so an explicitly-configured toolchain wins, then the
standard rustup locations. Returning None rather than exiting keeps the
policy (abort? skip?) with the caller.
"""
found = shutil.which("cargo")
if found:
return found
for d in CARGO_FALLBACKS:
cand = os.path.join(d, "cargo")
if os.path.isfile(cand) and os.access(cand, os.X_OK):
return cand
return None
def cargo_env():
"""Environment with the cargo bin dir prepended to PATH.
For subprocesses that shell out to cargo indirectly. Returns a copy;
never mutates os.environ.
"""
env = dict(os.environ)
cargo = cargo_bin()
if cargo:
env["PATH"] = os.path.dirname(cargo) + os.pathsep + env.get("PATH", "")
return env
def self_test():
"""Positive control: this module must actually locate things.
A resolver that silently returns a wrong-but-plausible answer is the
harness-does-nothing class applied to paths every downstream tool
would then measure the wrong tree while reporting success.
"""
results = []
def check(name, ok, detail=""):
results.append((name, ok, detail))
check("ROOT is a directory", os.path.isdir(ROOT), ROOT)
# Identified by content, not by name: a ROOT that pointed at the parent
# directory would still be "a directory".
check("ROOT is *this* repo",
os.path.isfile(os.path.join(ROOT, "Cargo.toml"))
and os.path.isdir(os.path.join(ROOT, "specs"))
and os.path.isfile(os.path.join(ROOT, "INTENT.md")))
# ROOT must not depend on the caller's cwd — the defect this fixes.
here = os.getcwd()
try:
os.chdir("/")
again = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
check("ROOT is independent of the caller's cwd", again == ROOT)
finally:
os.chdir(here)
cargo = cargo_bin()
check("cargo is locatable without PATH setup", bool(cargo), cargo or "NOT FOUND")
check("cargo_env puts cargo on PATH",
not cargo or os.path.dirname(cargo) in cargo_env()["PATH"].split(os.pathsep))
print("repo self-test (positive control)")
ok = True
for name, passed, detail in results:
print(f" [{'ok ' if passed else 'FAIL'}] {name}"
+ (f"{detail}" if detail else ""))
ok &= passed
return 0 if ok else 1
if __name__ == "__main__":
import sys
# This module has nothing to report but its own controls, so bare
# invocation and --self-test are the same run. The flag exists because
# `make self-tests` and loop-lint both look for it by name.
if len(sys.argv) > 1 and sys.argv[1] not in ("--self-test",):
print(f"usage: {sys.argv[0]} [--self-test]", file=sys.stderr)
sys.exit(2)
sys.exit(self_test())

View file

@ -24,6 +24,8 @@ import glob
import re
import sys
from repo import enter_root
RULE_RE = r"\*\*(GR-[A-Z]+\d+)"
COVERS_RE = r"covers: \[(.*?)\]"
AGGREGATE = "games/ground/src/lib.rs"
@ -107,6 +109,9 @@ def self_test():
def main():
# T01: inputs are repo-relative, so anchor to the repo rather than
# requiring the caller to `cd` first.
enter_root()
if "--self-test" in sys.argv:
return self_test()

View file

@ -1,7 +1,7 @@
---
id: CB-WP-0004
title: "Move mechanical turns off the token budget, and prove it worked"
status: proposed
status: in_progress
state_hub_workstream_id: "6880ac78-d817-41b9-b267-f12ff9deea28"
---
@ -41,7 +41,7 @@ unless the instrument disproved it (§Step 4, correction vs retarget).
```task
id: CB-WP-0004-T01
status: todo
status: done
priority: high
state_hub_task_id: "3ddfd3e2-8596-4969-a069-09577933fcbc"
```
@ -64,6 +64,23 @@ still cheaper than a regression.
**Predicted:** environment-setup turns → **< 10** (from 84),
**$1215** recovered. Highest confidence in the review.
**Delivered.** `tools/repo.py` resolves the repo root from `__file__` and
`cargo` from PATH-then-rustup-locations; every tool imports it, so `REPO`
is now stated once rather than four times. The Makefile derives `REPO`
from `MAKEFILE_LIST` and resolves `CARGO` the same way. `make env-test`
is the positive control — it runs every tool from `/` with
`PATH=/usr/bin:/bin`, and is wired into `make all`, so this cannot
silently regress and invalidate T05's measurement.
The leaf workaround in `dep-weight.py` was kept, not deleted: it is now
unreachable on a rustup machine, and `--self-test` asserts `cargo_bin()`
resolves unaided. Per the task text, a control that never fires is
cheaper than a regression.
`loop-lint` failed on `repo.py` immediately — a new tool with a positive
control but no `--self-test` entry point. Second time the gate has caught
its own pass's work within the hour.
## Task: `make task-done` — one command for a task close
```task