Open RossComputerGuy opened 4 months ago
There are a number of signed/unsigned conversions throughout the audit package. A number of these come from the i686 days. At this point, I no longer test 32 bit and would be afraid to make changes that may impact i686. While it's no longer a priority, I still don't want to knowingly break it.
The error around rawmemchr is somewhat concerning. It is inside an ifdef that configure tested to say it exists for the installed C library. For glibc, this also requires passing _GNU_SOURCE as a CFLAG which Makefile.am does. I suspect the ifdef was actually created for people using musl. Maybe config.h is not 100% right as I think we've met at the libcap-ng repo.
Zig has a C compiler which wraps around clang and sets all warnings as errors. libaudit builds correctly for glibc but fails with musl.