ci(gate): compile and lint feature-gated code, which nothing did (#1380)

[dc0sk]

Closes #1380.

#[cfg(feature = "x")] code must still parse when the feature is off, so a syntax error was
caught — but nothing after parsing was: type errors, borrow errors, wrong arity, a renamed method.
Both lint passes in gate.sh and the hook build --no-default-features, so nine feature families
gating code (cpal, serial/gpio, gpu, keychain, tokio, serde, gui/serve, hardware-tests, instruments)
were compiled by no automated check. ci.yml's gpu-feature-gates guarded one of them and, being
release/**-scoped (#1120), never ran on an ordinary PR.

It found two live defects, neither in cpal

• apps/openpulse-linksim/tests/serve_integration.rs — E0063. LinkParams gained cessb_enabled
  (b883b5f8) and notch (26696f98) in 2026-06; the test was last touched 2026-06-21. The serve
  feature's only tests had been uncompilable, and therefore never run, for ~3 months. They now
  run: 2 passed, 0 failed.
• apps/openpulse-linksim/src/gui.rs:509 — float-literal-f32-fallback, which rustc says "will
  become a hard error in a future release". The gui binary would have stopped compiling on a
  future toolchain with no warning to anyone.

One rule, not a list

cargo clippy --workspace --all-features --all-targets -- -D warnings, rather than the issue's named
--features cpal-backend: a hand-maintained list of crate+feature pairs is the same rotting mirror
the guard exists to catch. Safe because every feature here is additive (generic-serial = ["serial"]). A per-feature matrix was considered and rejected on measurement — no
cfg(all(feature = A, not(feature = B))) exists anywhere, so a 2¹⁷ powerset would find zero defects.

The preflight is not ceremony

--all-features pulls alsa-sys, libudev-sys and libdbus-sys, whose build scripts call
pkg_config and panic when a .pc file is absent. The ubuntu runner ships none of the three
-dev packages and neither gate-running job installed anything, so as first written this was
red-on-arrival for post-merge-gate.yml — #1074's archetype. My clean local run described this
host only after a root update stamped those .pc files on 2026-09-19; the ledger records
libdbus-sys failing on this same machine four days earlier.

So: a pkg-config preflight in both gate and hook that fails with the Debian package names
rather than skipping, and an apt-get step (deliberately not || true) in both CI jobs that run
gate.sh.

A latent defect the design exposed

gpiocdev is declared under [target.'cfg(target_os = "linux")'] while gpio.rs was gated on
feature = "gpio" alone. Cargo enables a feature whose optional dep is target-filtered out, so
--all-features on macOS compiles that code with no crate. Retargeted to all(target_os = "linux", feature = "gpio") at all five sites. Unverified on darwin — no darwin std on this host.

gpu-feature-gates removed

Subsumed: the new pass covers gpu and hardware-tests across every crate rather than five named
plugins, on every local gate run and in post-merge-gate, where that job never ran at all. Its
reasoning and the PR #424 precedent survive in the new step's comment.

Two stated limits

Not closed over targets (see gpio). Not closed over the shipped recipe: --all-features turns
instruments on, so an instruments-only item called from a cpal-gated production path passes all
three passes and fails only cargo build --release -p openpulse-cli --features cpal-backend. Zero
instances today — a residual, not a claim of completeness. My earlier "2×2 grid" framing was wrong:
the axes are not independent.

Sabotage — #1380's own probe

A planted type error in the cpal-gated run_drive (calibrate.rs:320):

pass1  --no-default-features --all-targets  -> rc=0
pass2  --no-default-features                -> rc=0
pass3  --all-features --all-targets         -> rc=101

The two rc=0 rows are what make the failure attributable to the new pass. The preflight was
separately sabotaged with a nonexistent library and correctly printed SKIPPED (missing pkg-config: zz-nonexistent-lib) with the install line, and failed.

Evidence honesty

Both defects found are in openpulse-linksim, a crate under a 2026-09-10 direction to be replaced,
and the yield on shipped cpal/gpu/serial paths was zero. The justification is the class — three
months undetected, PR #424 precedent — not today's haul.

Test results

scripts/gate.sh on the branch tip, clean tree:

suites=339 tests_passed=2574 tests_failed=0
GATE: PASS bcb23ed5def5098148236a0e1f7175a5f8487400 clean 20260920T093321Z

All 13 steps ok, including the new one.

cargo test -p openpulse-linksim --features serve --test serve_integration → 2 passed, 0 failed.

Reviewer note

The CI evidence that matters here is post-merge-gate.yml, not this PR's checks — the gate does
not run pre-merge (#1144), and the apt step is the thing at risk. I will confirm that run after merge
and report it.

Verification-objective: feature-gated code must be compiled and linted by an automated check, not
only parsed

Review: docs/dev/reviews/review-1380-feature-rot-guard.md

🤖 Generated with Claude Code

https://claude.ai/code/session_0188ATCj6DZ9aRVQ2vSirua6