trait | HeavyGuardHost | pub trait HeavyGuardHost: vox_cli_contracts::GateStatusWriter | crates/vox-cli-ci/src/lib.rs#L8-L16 | Seam for the Tier-2 vox ci guards that stay in vox-cli because they reach into its internals (command_catalog, command_registry_model, utils, fs_utils, commands::runtime). The moved dispatcher (run::run) calls dispatch_heavy(&cmd) first; vox-cli implements it on VoxCliProviders. Extends GateStatusWriter so the dispatcher can also record per-gate status through the same host object. |
fn | affected::is_sentinel | pub fn is_sentinel(path: &str) -> bool | crates/vox-cli-ci/src/affected.rs#L28-L30 | — |
fn | affected::contracts_outside_graph_force_full | pub fn contracts_outside_graph_force_full(changed_files: &[String]) -> bool | crates/vox-cli-ci/src/affected.rs#L39-L45 | Non-graph contracts/** edits can change SSOT surfaces workspace-wide; force a full PR gate. Generated reports are the exception (see [GENERATED_CONTRACT_REPORTS]); every other contracts path still forces a full run, so a new reader of a contract can never be skipped. |
fn | affected::ci_workflow_force_full | pub fn ci_workflow_force_full(changed_files: &[String]) -> bool | crates/vox-cli-ci/src/affected.rs#L48-L52 | CI workflow edits can change gate behavior workspace-wide; force a full PR gate. |
fn | affected::file_to_crate | pub fn file_to_crate(path: &str) -> Option<&str> | crates/vox-cli-ci/src/affected.rs#L54-L58 | — |
struct | affected::PathFlags | pub struct PathFlags | crates/vox-cli-ci/src/affected.rs#L61-L68 | — |
fn | affected::compute_path_flags | pub fn compute_path_flags(changed_files: &[String]) -> PathFlags | crates/vox-cli-ci/src/affected.rs#L70-L101 | — |
fn | affected::invert | pub fn invert(graph: &BTreeMap<String, Vec<String>>) -> BTreeMap<String, BTreeSet<String>> | crates/vox-cli-ci/src/affected.rs#L103-L112 | — |
fn | affected::reverse_closure | pub fn reverse_closure( | crates/vox-cli-ci/src/affected.rs#L114 | — |
fn | affected::set_includes_compiler | pub fn set_includes_compiler(set: &BTreeSet<String>) -> bool | crates/vox-cli-ci/src/affected.rs#L136-L138 | — |
enum | affected::Affected | pub enum Affected | crates/vox-cli-ci/src/affected.rs#L141-L145 | — |
fn | affected::compute_affected | pub fn compute_affected( | crates/vox-cli-ci/src/affected.rs#L147 | — |
fn | affected::shadow_misses | pub fn shadow_misses(junit_xml: &str, affected: &BTreeSet<String>) -> Vec<String> | crates/vox-cli-ci/src/affected.rs#L175-L203 | Parse nextest JUnit for failing tests; return crate names not in affected. |
struct | affected_cmd::CrateGraph | pub struct CrateGraph | crates/vox-cli-ci/src/affected_cmd.rs#L5-L8 | — |
fn | affected_cmd::graph_from_metadata | pub fn graph_from_metadata(meta: &serde_json::Value) -> CrateGraph | crates/vox-cli-ci/src/affected_cmd.rs#L10-L50 | — |
fn | affected_cmd::regen_graph | pub fn regen_graph(out_path: &std::path::Path) -> Result<(), String> | crates/vox-cli-ci/src/affected_cmd.rs#L52-L70 | — |
fn | affected_cmd::check_graph | pub fn check_graph(committed: &std::path::Path) -> Result<(), String> | crates/vox-cli-ci/src/affected_cmd.rs#L72-L90 | — |
fn | affected_cmd::partition_crates | pub fn partition_crates(weights: &[(String, u64)], n: usize) -> Vec<Vec<String>> | crates/vox-cli-ci/src/affected_cmd.rs#L103-L124 | Split crates into at most n groups of near-equal weight (longest-processing-time greedy: heaviest first, each into the currently lightest group). Each CI shard then compiles and runs only its own group’s test binaries — compiling every test binary in every shard (hash sharding) cost 17-19 min per shard. Empty groups are dropped; ties break by name. |
fn | affected_cmd::run_affected_cmd | pub fn run_affected_cmd(args: &[String]) -> i32 | crates/vox-cli-ci/src/affected_cmd.rs#L181 | — |
fn | agentskills_compliance::run | pub fn run() -> Result<()> | crates/vox-cli-ci/src/agentskills_compliance.rs#L86 | — |
fn | ai_fixtures_coverage::run | pub fn run(repo_root: &Path) -> Result<()> | crates/vox-cli-ci/src/ai_fixtures_coverage.rs#L15-L55 | Verify AI fixture catalog parity against shipped parser + HIR surfaces. |
fn | attention_ledger_parity::run | pub fn run(root: &Path) -> Result<()> | crates/vox-cli-ci/src/attention_ledger_parity.rs#L16-L54 | — |
fn | attention_parity::run | pub fn run(root: &Path) -> Result<()> | crates/vox-cli-ci/src/attention_parity.rs#L5-L38 | — |
fn | benchmark_telemetry::record_opt | pub async fn record_opt(name: &str, metric_value: Option<f64>, details: Option<serde_json::Value>) | crates/vox-cli-ci/src/benchmark_telemetry.rs#L74-L76 | Async: connect and write one benchmark row (no-op unless VOX_BENCHMARK_TELEMETRY is set). |
fn | benchmark_telemetry::record_opt_with_unit | pub async fn record_opt_with_unit( | crates/vox-cli-ci/src/benchmark_telemetry.rs#L79 | Like [record_opt] but forwards metric_value_unit (e.g. "seconds") to Codex. |
fn | benchmark_telemetry::record_syntax_k_opt | pub async fn record_syntax_k_opt( | crates/vox-cli-ci/src/benchmark_telemetry.rs#L128 | Async: connect and write one syntax-k telemetry row (no-op unless telemetry env is enabled). |
fn | benchmark_telemetry::record_opt_blocking | pub fn record_opt_blocking( | crates/vox-cli-ci/src/benchmark_telemetry.rs#L164 | Blocking variant for sync CLI paths (no Tokio handle required). |
fn | benchmark_telemetry::record_opt_with_unit_blocking | pub fn record_opt_with_unit_blocking( | crates/vox-cli-ci/src/benchmark_telemetry.rs#L173 | Blocking variant for sync CLI paths that includes a metric unit. |
fn | benchmark_telemetry::record_syntax_k_opt_blocking | pub fn record_syntax_k_opt_blocking( | crates/vox-cli-ci/src/benchmark_telemetry.rs#L231 | Blocking variant for sync syntax-k telemetry paths. |
struct | build_bench::TimingEntry | pub struct TimingEntry | crates/vox-cli-ci/src/build_bench.rs#L18-L21 | — |
fn | build_bench::extract_unit_data | pub fn extract_unit_data(html: &str) -> Vec<TimingEntry> | crates/vox-cli-ci/src/build_bench.rs#L30-L60 | Parse the embedded UNIT_DATA JS array from a cargo —timings HTML file. cargo emits const UNIT_DATA = [ {..}, .. ]; where each entry’s duration is float seconds (e.g. 9.37). We convert to whole milliseconds and drop zero-duration units (cached / no-rebuild). We do NOT filter on mode: in a cached/sccache build the real compile rows carry mode:"todo" with a nonzero duration, so a mode filter would discard exactly the data we want. |
fn | build_bench::find_newest_timings_html | pub fn find_newest_timings_html(repo_root: &Path) -> Option<std::path::PathBuf> | crates/vox-cli-ci/src/build_bench.rs#L63-L74 | Return the most-recently-modified cargo-timing-*.html under target/cargo-timings/. |
struct | build_bench::HistoryRecord | pub struct HistoryRecord | crates/vox-cli-ci/src/build_bench.rs#L77-L80 | — |
fn | build_bench::append_to_history | pub fn append_to_history(history_path: &Path, run_id: &str, entries: &[TimingEntry]) -> Result<()> | crates/vox-cli-ci/src/build_bench.rs#L83-L101 | Append one JSONL record to the history file (create if absent). |
struct | build_bench::Scenario | pub struct Scenario | crates/vox-cli-ci/src/build_bench.rs#L104-L108 | — |
struct | build_bench::ScenarioFile | pub struct ScenarioFile | crates/vox-cli-ci/src/build_bench.rs#L111-L114 | — |
struct | build_bench::BenchRecord | pub struct BenchRecord | crates/vox-cli-ci/src/build_bench.rs#L117-L121 | — |
struct | build_bench::Snapshot | pub struct Snapshot | crates/vox-cli-ci/src/build_bench.rs#L124-L132 | — |
fn | build_bench::current_host | pub fn current_host() -> String | crates/vox-cli-ci/src/build_bench.rs#L139-L141 | Host identifier for a snapshot written now, e.g. "x86_64-windows". arch-os rather than the full target triple: std::env::consts gives these without a build script, and the distinction that matters for wall-clock comparison is the machine class, not the ABI suffix. |
fn | build_bench::host_mismatch_warning | pub fn host_mismatch_warning(baseline_host: Option<&str>, current: &str) -> Option<String> | crates/vox-cli-ci/src/build_bench.rs#L150-L160 | Warn when a baseline was timed on a different host than the current run. Scenario wall-clock is host-bound far more strongly than per-crate self-time: a Windows-measured baseline compared against a Linux CI run produces a large delta on every scenario that reflects the machine, not the code. Returns None when the baseline predates provenance, so pre-existing files do not emit a warning on every run — the dynamic that gets warnings ignored. |
fn | build_bench::by_id | pub fn by_id(&self) -> BTreeMap<&str, &BenchRecord> | crates/vox-cli-ci/src/build_bench.rs#L163-L165 | — |
struct | build_bench::DeltaRow | pub struct DeltaRow | crates/vox-cli-ci/src/build_bench.rs#L169-L175 | — |
fn | build_bench::compute_deltas | pub fn compute_deltas(base: &Snapshot, new: &Snapshot) -> Vec<DeltaRow> | crates/vox-cli-ci/src/build_bench.rs#L177-L204 | — |
fn | build_bench::baseline_is_unpopulated | pub fn baseline_is_unpopulated(base: &Snapshot) -> bool | crates/vox-cli-ci/src/build_bench.rs#L213-L222 | True when base is a committed placeholder rather than a real measurement: it has at least one ok record and every ok record is wall_ms == 0. This matters because [compute_deltas] defines pct as 0.0 whenever the base is zero, so an unpopulated baseline makes every scenario report a 0% delta and the gate can never fail. A real cargo check scenario cannot measure 0 ms, so all-zero is unambiguously “never filled in”. |
fn | build_bench::format_delta_markdown | pub fn format_delta_markdown(label: &str, rows: &[DeltaRow]) -> String | crates/vox-cli-ci/src/build_bench.rs#L224-L235 | — |