Daniel Macks

-Werror is not usually a best-practice in shipped release versions. It leaves you at the mercy of diagnostics from compilers and external headers supplied by other vendors.

folly/memcpy.S is the only compiler that does not pass both (only passes the CXX_STD one). That makes sense, since it's not a C++ source, so cmake's native flag isn't used,...

See also #1740, which at least implements a flag to override the CMakeLists.txt one.

In the autotools toolchain, soversion is via libtool's ```-version-info``` flag (see [full docs](https://www.gnu.org/software/libtool/manual/html_node/Updating-version-info.html)). Its parameter-set is "c:r:a", and the soname becomes "c-a". In the libarchive-3.6.2 release, it's ```-version-info 19:2:6```: 19-6=13,...

Seeing as how #1751 (merged Jul 31, 2022) adds ```-fvisibility=hidden``` and apparently there are a bunch of internal tokens/pragmas present in the sources that control visibility of various symbols, is...

Maybe I didn't explain the algorithm clearly. --with-pc-flag sets the _default_, not the mandatory only. PKG_CONFIG_LIBDIR still replaces it at runtime and PKG_CONFIG_PATH still prepends to take precedence. That's because...

Looks fine now. README.txt needs to be updated to match (the "Package paths" section still describes the transiently-implemented way).

Actually no, it doesn't work. 'setup.py build_py --with-pc-path=[path]' correctly stashes [path] in install_config.py, but 'setup.py install' nukes that setting (installs install_config.py containing "pc_path = None"). Looks like 'setup.py install' doesn't...

(sorry for the stream-of-consciousness/debugging session here!) I can use --skip-build in the 'setup.py install' to avoid breaking the formerly-correct install_config.py, but then install itself fails because setup.py build_py only builds...

Apple's compiler toolchain does have visibility control for symbols, just not versioning of that. The linker can take a file listing the symbols that should be exported (and therefore hide...