
ID	0019
Class	ARCHITECTURE
Severity	1
Status	blocked
Models	—
Found	2026-09-15 audit (`e3f3107`)
Decision needed	Is `NetSolidityMeshSpec` meant to be built? Ideally checked with a Windows build.

Duplicate fixed

The doubled Test/NetSolidityMesh.cpp / Test/NetSolidityMesh.h entries in src/net/CMakeLists.txt are removed (CMake de-duplicated them, so nothing changes in the build). What remains is the question the duplicate may have been hiding.

Evidence

src/net/Test/NetSolidityMeshSpec.cpp and .h exist, and src/net/Test/NetSolidityMesh.h:6 includes the header and takes a const NetSolidityMeshSpec* in its constructor (:13). Neither file is in TEST_STRUCTURE_SRC or TEST_STRUCTURE_H (src/net/CMakeLists.txt:9-35). The doubled entry looked like a paste that was meant to add them.

Possible fix

(a) Add Test/NetSolidityMeshSpec.cpp and .h to the two lists, if the spec is meant to be built.
(b) Leave it out, if NetSolidityMesh is itself unmaintained (see 0005).

Owner question, open

is NetSolidityMeshSpec supposed to be built? The whole block is Windows-only, so this cannot be checked from a Linux build.

Test that would prove it

A Windows build of the library.

Risk

Option (a) adds a translation unit to a Windows-only build that nothing in CI covers, so it could break that build in a way this repository cannot detect.

Why not solved (fix pass 2026-09-15)

Partially solved. Commit 49918bf in fhsim_marine_elements removed the doubled Test/NetSolidityMesh.cpp and .h entries from src/net/CMakeLists.txt; CMake de-duplicated them, so the build is unchanged. The item text above was rewritten to the remaining question: Test/NetSolidityMeshSpec.{cpp,h} exists, is included by NetSolidityMesh.h:6, and is in no source list. Adding it would add a translation unit to a Windows-only block that no CI covers, so a Linux checkout cannot tell whether that build would still link. Smallest owner decision: whether NetSolidityMeshSpec is meant to be built, ideally checked with a Windows build.