Move the canonical golangci-lint to v2.15.0 and bring back canonicalheader

v2.15.0 (released 2026-10-16, built with go1.27.0 like v2.14.0) bundles
canonicalheader with its upstream fix for non-deterministic results, the
defect behind issue 105. Linted 20 times each on a scratch package with
two planted findings beside a ResponseWriter.Header() call, v2.14.0
reported them on 8 runs and nothing on 12; v2.15.0 reported them on all
20. REPO_POLICIES.md names the new digest; .golangci.yml no longer
disables canonicalheader and is otherwise unchanged, since it verifies
against v2.15.0 as it was and default: all switches on no new linter.

.golangci.yml

@@ -25,8 +25,6 @@ linters:
     # silenced by disabling that name, not by enabling the successor.
     - wsl              # Deprecated, replaced by wsl_v5
     - gomodguard       # Deprecated, replaced by gomodguard_v2
-    # Misses findings at random in v2.14.0; back once a pinned release fixes it
-    - canonicalheader
   settings:
     lll:
       line-length: 88

TODO.md

@@ -21,15 +21,15 @@ fmt-check, and commit.
 
 # Completed Steps
 
-- 2026-11-03: A first test run that timed out now gets no verbose rerun
+- 2026-11-01: Moved the canonical golangci-lint to v2.15.0 (released
-  (https://git.eeqj.de/sneak/prompts/issues/61), on sneak's approval at
+  2026-10-16), still built with go1.27.0, so the `go` directive rule is
-  https://git.eeqj.de/sneak/prompts/issues/61#issuecomment-165195. The Go recipe
+  unchanged. The canonical `.golangci.yml` passes `golangci-lint config verify`
-  in `REPO_POLICIES.md` looks in the first run's output for
+  on v2.15.0, and `default: all` switches on no new linter. `canonicalheader` is
-  `panic: test timed out` or `*** Test killed`, and exits 1 when either is
+  enabled again (issue 105): v2.15.0 carries the upstream commit "fix:
-  there. The Python `make test` example takes the same `timeout 90 sh -c`
+  non-deterministic results", and on a scratch package with two planted findings
-  wrapper as its `Dockerfile` example. The 60 and 90 second rule and the
+  beside a `ResponseWriter.Header()` call, linted 20 times with each pinned
-  5-minute rule now say a hung package uses the backstop once. Consuming
+  image, v2.14.0 reported them on 8 runs and nothing on 12, while v2.15.0
-  repositories pick this up on their next re-vendor.
+  reported them on all 20.
 - 2026-10-27: Added `prompts/API_POLICIES.md`
   (https://git.eeqj.de/sneak/prompts/issues/137). It holds sneak's rule from
   https://git.eeqj.de/sneak/routewatch/issues/199 that a value in an API

prompts/REPO_POLICIES.md

@@ -1,6 +1,6 @@
 ---
 title: Repository Policies
-last_modified: 2026-11-03
+last_modified: 2026-11-01
 ---
 
 This document covers repository structure, tooling, and workflow standards. Code
@@ -233,7 +233,6 @@ style conventions are in separate documents:
     # Test phase. -race needs cgo and so a C compiler, which the Debian Go
     # image ships and the alpine one does not. -count=1 because the mounted
     # cache also holds test results, which go test would otherwise replay.
-    # A first run that timed out gets no rerun: its output has the stacks.
     # golang:1.x, YYYY-MM-DD
     FROM golang@sha256:... AS test
     WORKDIR /src
@@ -242,16 +241,9 @@ style conventions are in separate documents:
     ARG CACHEBUST
     COPY . .
     RUN --mount=type=cache,target=/root/.cache/go-build \
-        output=$(go test -count=1 -timeout 90s -race -cover ./... 2>&1); \
+        go test -count=1 -timeout 90s -race -cover ./... || \
-        status=$?; \
+        { echo "--- Rerunning with -v for details ---"; \
-        printf '%s\n' "$output"; \
+          go test -count=1 -timeout 90s -race -v ./...; exit 1; }
-        test "$status" -eq 0 && exit 0; \
-        case "$output" in \
-            *"panic: test timed out"*|*"*** Test killed"*) exit 1 ;; \
-        esac; \
-        echo "--- Rerunning with -v for details ---"; \
-        go test -count=1 -timeout 90s -race -v ./...; \
-        exit 1
 
     # Build stage. Nothing is wanted from either phase above; the copies
     # are what make BuildKit build them first, so nothing after them can
@@ -421,8 +413,7 @@ style conventions are in separate documents:
     `pipenv install --dev --deploy`:
 
     ```dockerfile
-    # Test phase. Both pytest runs share one 90-second timeout, so a hang
+    # Test phase. The 90-second timeout covers both pytest runs.
-    # uses it once and gets no rerun.
     # python:3.x-slim, YYYY-MM-DD
     FROM python@sha256:... AS test
     # make and git are base tools that script/bootstrap installs when they
@@ -493,25 +484,20 @@ style conventions are in separate documents:
   improvement bug against that repo. Add a 90-second timeout to the test
   invocation (`go test -timeout 90s`). The backstop deliberately sits above the
   hard cap so that it catches a genuinely hung test rather than a merely slow
-  one. A hung package uses the backstop once, because a first run that timed out
+  one.
-  gets no rerun. `-timeout` applies to each package, so several hung packages
-  can each use it.
 
 - **The test command should use the conditional verbose rerun pattern.** Run
   tests without `-v` (verbose) first. If tests fail, automatically rerun with
   `-v` to show full output. This keeps CI logs and `docker build` output clean
   on success (just package/suite summaries) while providing full diagnostic
-  detail on failure (every test case, every assertion). A first run that timed
+  detail on failure (every test case, every assertion). The command lives in the
-  out gets no rerun, because the rerun would only wait for the same hang. The
+  `test` phase of the `Dockerfile`, since `script/test` builds that phase; the
-  command lives in the `test` phase of the `Dockerfile`, since `script/test`
+  Makefile form below is the same pattern for any repo-local invocation:
-  builds that phase; the Makefile form below is the same pattern for any
-  repo-local invocation:
 
     ```makefile
     test:
     	@<test-command> || \
-    		{ <exit 1 if that run timed out>; \
+    		{ echo "--- Rerunning with -v for details ---"; \
-    		  echo "--- Rerunning with -v for details ---"; \
     		  <test-command-with-v>; exit 1; }
     ```
 
@@ -519,29 +505,11 @@ style conventions are in separate documents:
 
     ```makefile
     test:
-    	@output=$$(go test -count=1 -timeout 90s -race -cover ./... 2>&1); \
+    	@go test -count=1 -timeout 90s -race -cover ./... || \
-    		status=$$?; \
+    		{ echo "--- Rerunning with -v for details ---"; \
-    		printf '%s\n' "$$output"; \
+    		  go test -count=1 -timeout 90s -race -v ./...; exit 1; }
-    		test "$$status" -eq 0 && exit 0; \
-    		case "$$output" in \
-    			*"panic: test timed out"*|*"*** Test killed"*) exit 1 ;; \
-    		esac; \
-    		echo "--- Rerunning with -v for details ---"; \
-    		go test -count=1 -timeout 90s -race -v ./...; \
-    		exit 1
     ```
 
-    The first run's output is kept so that the recipe can look in it for what Go
-    prints for a hang:
-    - `panic: test timed out` means the test's own timer fired.
-    - `*** Test killed` means `go test` stopped a test that ignored that timer,
-      about a minute later.
-
-    Go prints the goroutine stacks with either line, as long as the test can
-    still respond. A rerun would only wait for the same hang, so the recipe
-    exits 1. Any other failure reruns with `-v`. The kept output prints when the
-    first run ends, not while it runs.
-
     `-count=1` is required on both invocations: it defeats Go's test _result_
     cache, so neither run can report a stored pass in place of running the
     tests. It leaves the build cache alone, so it costs the runtime of the suite
@@ -560,22 +528,17 @@ style conventions are in separate documents:
 
     ```makefile
     test:
-    	@timeout 90 sh -c 'python -m pytest || { \
+    	@python -m pytest || \
-    		echo "--- Rerunning with -v for details ---"; \
+    		{ echo "--- Rerunning with -v for details ---"; \
-    		python -m pytest -v; exit 1; }'
+    		  python -m pytest -v; exit 1; }
     ```
 
-    Both pytest runs share that one 90-second timeout, as in the `Dockerfile`
-    example above, so a hang uses it once and gets no rerun.
-
     The `exit 1` ensures the target always fails after a rerun — the first run
     already proved the tests are broken, so the build must not pass even if a
     flaky test happens to succeed on the second attempt. The rerun exists solely
     for diagnostic output.
 
-- Docker builds must complete in under 5 minutes. A hung package uses the
+- Docker builds must complete in under 5 minutes.
-  90-second test backstop once, because a first run that timed out gets no
-  rerun.
 
 - `make check` must not modify any files in the repo. Tests may use temporary
   directories.
@@ -695,10 +658,10 @@ style conventions are in separate documents:
   `test-support` depguard rule, where a repo names its own test-support packages
   by full import path. A repo adds entries there and changes nothing else, and a
   re-vendor carries its entries forward. The canonical golangci-lint version is
-  v2.14.0 (released 2026-09-24), pinned as the digest of the lint phase's base
+  v2.15.0 (released 2026-10-16), pinned as the digest of the lint phase's base
   image
-  (`golangci/golangci-lint@sha256:ad862ba6b3798cbe0fd9fd7408d498fd74fbd2623a92406b2fd3898faf0bf98f`,
+  (`golangci/golangci-lint@sha256:fbfcfd12a6f8d7f1b97b3362ca80fc2a244d46863148c73cb88e7793af96c481`,
-  which reports `2.14.0 built with go1.27.0 from 114493f9`). A module's `go`
+  which reports `2.15.0 built with go1.27.0 from 25c5369a`). A module's `go`
   directive must not name a newer Go minor version than the one golangci-lint
   was built with, or golangci-lint refuses to lint it: this release lints
   `go 1.27.1` but not `go 1.28`. That digest is the only pin, since no repo