Closed
Description
I need to revisit run_make_support::fs::create_symlink
because it's a wrong abstraction: there's a good reason why symlink_file
/symlink_dir
are different operations under std::os::windows::fs
.
Furthermore, I need to double-check our symlink handling in rmake.rs tests. It's ok if symlinks are removed via fs::remove_dir_all
because that has special handling for symlinks on Windows, but need to be extra careful if symlinks are attempted to be removed with fs::{remove_file, remove_dir}
if the test can be run on Windows.