ci: re-run make check on every cibuild instead of serving it from the layer cache (closes #115)

`script/cibuild` was plain `docker build .`. The Dockerfile does
`COPY . .` and then `RUN make check`, and Docker invalidates `COPY . .`
only on a content change, so on a byte-identical tree the check layer
was reused and the suite never ran. The script's header comment claimed
that a successful build implies all checks pass, which was false
whenever the cache was warm. Reproduced on this branch's parent: a
second consecutive run returned success in 283 ms with
`#13 [builder 9/10] RUN make check` reported `CACHED`.

That matters more here than in a typical repo. DNS is never mocked in
this repository, so the suite queries live DNS and its outcome varies
with real-world conditions; caching the verdict of a non-deterministic
check replays a stale result in exactly the case where re-running is
most valuable. It is also the gate every PR is verified through.

Fix: declare `ARG CHECK_EPOCH` immediately above the check step and
expand it into the command, with `script/cibuild` passing a fresh
`$(date +%s%N)` per invocation. A build argument's value participates in
the cache key of later instructions in the stage even when they do not
reference it, so a fresh value busts this layer either way; the value is
expanded into the command deliberately, which makes the invalidation a
property of the command string itself rather than of how a given builder
treats unreferenced args, and surfaces the epoch in the build log as a
diagnostic. Placing the ARG here and no earlier keeps the pinned
toolchain installs and `go mod download` above the invalidation line, so
only the check and the steps after it re-run. The epoch is nanosecond
granular so that two concurrent invocations starting in the same second
cannot share a value.

A plain `docker build` without the argument caches as before; nothing
outside the CI entrypoint changes behaviour.

Verified by experiment, not inspection:

- Two consecutive runs on an unchanged tree: 55.2 s and 42.2 s, both
  exit 0, with distinct epochs. The second run shows
  `RUN echo "check epoch: ..." && make check` executing for 36.0 s and
  216 passing tests across all eight packages, while `apk add`, both
  pinned `go install` steps, `go mod download`, `COPY go.mod go.sum` and
  `COPY . .` all report `CACHED`.
- Negative control: planted `internal/config/zz_negative_control_test.go`
  calling `t.Fatal("NEGATIVE-CONTROL-115: planted failure, cache did not
  serve this layer")`. The build failed in 24.7 s with exit 1, printing
  that exact message and `--- FAIL: TestNegativeControlIssue115`, and the
  check step exited with code 2. A cached layer cannot produce a failure
  predicted in advance, so this establishes the suite ran. The file was
  then removed, `git status` confirmed clean, and the tree built green
  again in 48.1 s.
- Total build time 42-55 s against the policy's 5-minute ceiling.
- `make check` green. No pin touched: the `golang` and `alpine` sha256
  digests, golangci-lint `c0d3ddc9`, and goimports `009367f5` are
  unchanged, and `.golangci.yml` still hashes to `021cc83f4e6f...`.

Dockerfile

@@ -15,8 +15,25 @@ RUN go mod download
 
 COPY . .
 
-# Run all checks - build fails if any check fails
+# Run all checks - build fails if any check fails.
-RUN make check
+#
+# CHECK_EPOCH is a cache-busting build argument. Without it, an
+# unchanged tree leaves this layer's cache key identical and Docker
+# serves the previous verdict instead of re-running the suite, so the
+# build reports a green it did not earn. A build argument's value
+# participates in the cache key of later instructions in the stage even
+# when they do not reference it, so a fresh value busts this layer
+# either way. It is expanded into the command deliberately: that makes
+# the invalidation a property of the command string itself rather than
+# of how a given builder treats unreferenced args, and it surfaces the
+# epoch in the build log as a diagnostic.
+#
+# Placing the ARG here and nowhere earlier keeps everything above it
+# (toolchain install, go mod download) cached, so only the check and the
+# steps after it re-run. script/cibuild passes a fresh value per run; a
+# plain `docker build` without it caches as before.
+ARG CHECK_EPOCH
+RUN echo "check epoch: ${CHECK_EPOCH}" && make check
 
 # Build the binary
 RUN make build

README.md

@@ -393,7 +393,10 @@ them. We provide:
 - `script/check` — run test, lint, and fmt-check
 - `script/docker` — build the Docker image tagged via
   `script/projectname`
-- `script/cibuild` — CI entrypoint: plain `docker build .`
+- `script/cibuild` — CI entrypoint: `docker build .` with a fresh
+  `CHECK_EPOCH` build argument, so the Dockerfile's `make check` layer
+  is never served from the cache and a green build always means the
+  checks ran on this invocation
 - `script/precommit` — run by the git pre-commit hook; `go mod tidy`
   guard, then `script/check`
 - `script/install-precommit` — install the git pre-commit hook

TODO.md

@@ -25,6 +25,16 @@ confirm make check still passes.
 
 # Completed Steps
 
+- 2026-08-09: `script/cibuild` can no longer report a green it did not
+  earn. The Dockerfile declares `ARG CHECK_EPOCH` immediately above the
+  check step and expands it into the `RUN` command, and `script/cibuild`
+  passes a fresh `$(date +%s%N)` per invocation, so the `make check`
+  layer is always re-executed while the pinned toolchain install and
+  `go mod download` stay cached. Verified by experiment: before the fix
+  a second run on an unchanged tree returned in 283 ms with the check
+  layer `CACHED`; after it the check runs every time, and a deliberately
+  planted always-failing test made the build fail with exactly that
+  test's message
 - 2026-08-07: golangci-lint bumped to v2.12.2 (commit-pinned installs
   in `Dockerfile` and `script/bootstrap`); `.golangci.yml` set to the
   org-standard v2-schema config used across the org's repos

script/cibuild

@@ -1,13 +1,18 @@
 #!/bin/sh
-# script/cibuild: run the CI build. The Dockerfile runs make check, so
+# script/cibuild: run the CI build. The Dockerfile runs make check, and
-# a successful build implies all checks pass.
+# the CHECK_EPOCH build argument below is fresh on every invocation, so
+# the check layer is never served from the Docker layer cache: a
+# successful build means the checks were executed and passed on this
+# run, not on some earlier one. Only the check step and the steps after
+# it are invalidated; the toolchain install and go mod download stay
+# cached.
 set -eu
 
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    docker build .
+    docker build --build-arg CHECK_EPOCH="$(date +%s%N)" .
 }
 
 main "$@"