Module cli 

Hand-rolled CLI argument parsing.

clap would add a multi-crate dependency tree. The demo binary only takes a handful of flags per subcommand, so a small, declarative parser keeps the surface area observable.

Each subcommand module exposes parse_and_run(args) which receives the tail of the argv (after the subcommand name) and returns an std::process::ExitCode. The parser itself is parse_flags, which walks --key value and --key=value forms and populates a tiny HashMap.

Modules

audit_report

S-REAL.1 — Deterministic, no-JavaScript HTML audit report renderer.

bench

dsfb-gpu-debug bench — measure pipeline wall-clock on CPU and GPU.

bench_gpu_scale

dsfb-gpu-debug bench-gpu-scale — R.7 money-table headline benchmark

compare

dsfb-gpu-debug compare — compare two case files and emit a verdict.

generate_fixture

dsfb-gpu-debug generate-fixture — synthesize the canonical 10 000-event trace fixture and write it to disk in canonical-JSON form.

ingest

S-REAL.1 — Deterministic ingest of # residual-projection v2 TSV fixtures.

run_cpu

dsfb-gpu-debug run-cpu — run the CPU reference pipeline.

run_gpu

dsfb-gpu-debug run-gpu — run the CUDA-accelerated pipeline.

s_real_audit

S-REAL — Real Dataset Audit Gauntlet CLI driver (canonical subcommand: s-real-audit; historical alias: s-real-1-audit).

Functions

parse_flags

Parse --key value, --key=value, and bare---key flag forms into a sorted map.

require_flag

Look up a required flag; emit a helpful error if it’s missing.

usage_error

Print a usage error to stderr and return the standard usage exit code.