Local code intelligence for developers and AI agents. Connect code, docs and infrastructure in a persistent graph, with CBM Lens, MCP tools, agent integrations and change-aware test selection.
C aiderastclaude-codecode-analysiscode-intelligencecodexcursorcypherdeveloper-toolsgemini-cligraph-visualizationkilocodeknowledge-graphmcpmcp-servermodel-context-protocolopencodesqlitetree-sitterwindsurf
test(lint-mem): an unwhitelisted suffixed finding must fail the gate
Martin Vogel24871eed7b62
diff --git a/tests/test_lint_mem_gate.sh b/tests/test_lint_mem_gate.shindex c6a655222..9178bca56 100755--- a/tests/test_lint_mem_gate.sh+++ b/tests/test_lint_mem_gate.sh@@ -2,10 +2,15 @@# Regression contract for clang-tidy's `,-warnings-as-errors` diagnostic suffix# (#2442). With WarningsAsErrors enabled, clang-tidy prints an analyzer check as# [clang-analyzer-<check>,-warnings-as-errors]; a gate that only accepts-# [<check>] drops every such line and reports a clean gate. A suffixed finding-# WITH an argued whitelist entry must be counted under its plain check name and-# suppressed (exit 0, "1 argued false positive"). Every assertion is reported,-# so a regression names what broke.+# [<check>] drops every such line and reports a clean gate. Both directions are+# pinned, and every assertion is reported, so a regression names what broke:+# 1. a suffixed finding WITH an argued whitelist entry is counted under its+# plain check name and suppressed (exit 0, "1 argued false positive");+# 2. a suffixed finding WITHOUT an entry fails the gate (exit 1) and is+# reported under its plain check name. This is the negative assertion: a+# parser that drops the suffixed line reports clean (exit 0) here.+# Reverting the suffix group in FINDING_RE (scripts/lint-mem-gate.py) turns+# both red.## The gate runs from inside the fixture directory with RELATIVE paths. On the# Windows leg python3 is a native MSYS2 build: MSYS converts path ARGUMENTS for@@ -39,6 +44,10 @@ cat > "$FIX/sample.c" <<'EOF'int sample(int *value) {return *value;}++int unlisted(int *value) {+ return *value;+}EOF# A native Windows python3 writes CRLF; strip it before the hash is reused.@@ -60,7 +69,7 @@ EOFSUFFIX='[clang-analyzer-core.NullDereference,-warnings-as-errors]'-# Whitelisted: counted under the plain check name, then suppressed.+# 1. Whitelisted: counted under the plain check name, then suppressed.rc=0printf '%s\n' "sample.c:2:12: warning: possible null dereference $SUFFIX" \| run_gate > /dev/null 2> "$FIX/whitelisted.err" || rc=$?@@ -73,8 +82,24 @@ if ! grep -q 'memory gate clean (1 argued false positive' "$FIX/whitelisted.err"show "$FIX/whitelisted.err"fi+# 2. Not whitelisted: the gate must fail and name the finding.+rc=0+printf '%s\n' "sample.c:6:12: warning: possible null dereference $SUFFIX" \+ | run_gate > /dev/null 2> "$FIX/unlisted.err" || rc=$?+if [ "$rc" -ne 1 ]; then+ fail "an unwhitelisted suffixed finding left the gate at exit $rc, expected 1" \+ "(exit 0 = the suffixed line was parsed as no finding, #2442)"+ show "$FIX/unlisted.err"+fi+if ! grep -qF 'sample.c:6: possible null dereference [clang-analyzer-core.NullDereference]' "$FIX/unlisted.err" ||+ ! grep -qF 'no whitelist entry for unlisted()' "$FIX/unlisted.err" ||+ ! grep -qF '1 finding(s) unaccounted for' "$FIX/unlisted.err"; then+ fail "the gate did not report the unwhitelisted suffixed finding under its plain check name"+ show "$FIX/unlisted.err"+fi+if [ "$failures" -ne 0 ]; thenecho "lint-mem-gate warning suffix contract: $failures assertion(s) failed" >&2exit 1fi-echo 'lint-mem-gate warning suffix contract passed'+echo 'lint-mem-gate warning suffix contract passed (whitelisted + unwhitelisted suffixed findings)'