f3007e8624
GCC doesn't have a specific warning flag for this warning, so we can't add a GCC case to the existing Clang suppression in the header. Hovever, the issue can be fixed by including the moc file into tst_moc.cpp, so that the compiler sees the definition of the inline signal provided by the said moc file. This means we can also drop the Clang warning suppressions. To prevent duplicate definition linker errors, remove the header file from JSON_FILES, and the forced separate compilation that entails. I'm pretty sure that some cmake magic could preserve the JSON test coverage, but a) the file doesn't emit anything special in its json (certainly nothing about the inline'ness of the signals) and b) the presence of mySlot() and mySignal() as slot and signal on these classes is already tested by singleFunctionKeywordSignalAndSlot(). Amends the start of the public history, but not picking back far, because a) a prequel commit to this commit wasn't picked further, either, and b) this has been such a long-standing issue that it's clear no-one cares and so it's just not worth the risk, however small that may be, to backport. On GCC, the only remaining warnings thrown by tst_moc are now moc-generated ones. Task-number: QTBUG-119042 Change-Id: Icde1eee6c97bd48d72e15461b6ec80c50c044eaa Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io> Reviewed-by: Thiago Macieira <thiago.macieira@intel.com> (cherry picked from commit b1b3af02c0fb8084d2eb984b1f7ae68f1861f9a4) Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>