test: point the Verifies trace at the symbol that exists

[kutsibalci]

The RecordProperty("Verifies", ...) record in alternating_control_block_test.cpp:95 names a symbol that does not exist:

score::mw::log::detail::wait_free_producer_queue::alternating_control_block_test::GetOppositeLinearControlBlock

Two components of that path are not namespaces:

• wait_free_producer_queue is the directory. grep -rn "namespace *wait_free_producer_queue" over the repository returns nothing.
• alternating_control_block_test is this test file's own basename. Same result — it is never declared as a namespace.

GetOppositeLinearControlBlock is declared in alternating_control_block.h:75, inside score → mw → log → detail (opened at lines 19–25, closed at 78–81). So the qualified name is:

score::mw::log::detail::GetOppositeLinearControlBlock

Why it matters here rather than being cosmetic

Verifies is the trace from a test to the unit under test, and this test is marked ASIL B. The record is traceability evidence: it builds, the test passes, and the link points at nothing. Nothing catches it either — clang-tidy sees a string literal, and the string is only ever read by whatever consumes the test report.

Form

The other twenty Verifies records in the repository, all in score/datarouter/test/ut/ut_logging/test_socketserver.cpp, are written with a leading :: and a trailing ():

RecordProperty("Verifies", "::score::platform::datarouter::SocketServer::CreateDltServer()");

so this one is written the same way rather than only having the two bogus components removed.

-    RecordProperty("Verifies",
-                   "score::mw::log::detail::wait_free_producer_queue::alternating_control_block_test::"
-                   "GetOppositeLinearControlBlock");
+    RecordProperty("Verifies", "::score::mw::log::detail::GetOppositeLinearControlBlock()");

How this was found, and what I did not touch

I extracted every RecordProperty("Verifies", ...) value in the repository — 21 sites, which matches grep -c exactly, so the extraction is complete — and resolved each named symbol against the declarations in the tree. This is the only one that does not resolve.

Something I noticed but deliberately left alone: Verifies coverage in score/mw/log/detail/wait_free_producer_queue/ is sparse — 1 record across 22 tests in 5 files, and this was that one record.

file	Verifies	tests
alternating_control_block_test.cpp	1	5
linear_control_block_test.cpp	0	2
linear_reader_test.cpp	0	1
wait_free_alternating_writer_test.cpp	0	5
wait_free_linear_writer_test.cpp	0	9

Adding the missing ones would mean deciding what each test actually verifies, which is a call for whoever owns that component, so this PR only corrects the record that is wrong. Happy to follow up if that is wanted.

Checks

• Both bogus path components confirmed absent as namespaces across all .h, .hpp and .cpp in the repository.
• The corrected path read from the enclosing namespace blocks in alternating_control_block.h, not inferred from the directory layout.
• One line of test metadata; no production code, no test logic, no build changes.
• ECA signed; the commit carries Signed-off-by.

AI disclosure

AI-assisted (Claude Code). The extraction and this description were produced with the tool, and I checked the result before opening: I confirmed the extraction found every RecordProperty("Verifies", ...) in the tree by comparing against grep -c, read the namespace nesting in the header rather than assuming it from the path, searched for both suspect components as namespace declarations and found neither, and compared against the twenty existing records to match their form.

Worth stating, because it nearly cost me this finding: my first pass validated each claimed symbol against the set of identifiers appearing anywhere in the repository — which included the claim string itself, so alternating_control_block_test "existed" and the record looked fine. A claim cannot be checked against a corpus that contains the claim. I have reviewed and understood the change and take responsibility for it.