Michael Förderer
Michael Förderer
Ok. The build path is in `/tmp`, is it possible that the directory is different between build and execution and/or cleaned up after executing the build? The gcno files are...
- The homebrew flags do not matter. If the gcda files are generated the files are instrumented. The gcno files must be somewhere.
But you still use `--gcov-object-directory=build`.
Sorry, I overlooked this. But you're still running the command in `working-directory: ${{ steps.test.outputs.buildpath }}`.
After playing a little bit with your branch I saw the message `Stderr of gcov was >>var/tmp/sunshine-20250505-31413-l83d5j/build/tests/CMakeFiles/test_sunshine.dir/__/generated-src/linux-dmabuf-unstable-v1.c.gcno:version 'B42*', prefer 'B33*'` for ubuntu. This is if GCOVR uses another version than...
The message is not from gcovr, it's from gcov itself. It seems that the `../` in the source path isn't handled correct in gcov itself. Can you provide the `build/tests/CMakeFiles/test_sunshine.dir/__/src/audio.cpp.gcda`?
By default they are created in the same directory. If you use the environment variables of gcc to change the directory of the gcda file during runtime of the executable...
@wqshr12345 Can this be closed?
The option could be reused but I'm not sure if I like the idea of changing the content of a report file by adding the "wrong" data to the elements....
> However, I'm really not a fan of the `--decisions` feature. Unless GCC 14+ is used (where we can take condition-coverage data directly from the gcov JSON), the decision analyzer...