You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently rustc has to pass the actual value of the --target argument to cc::windows_registry::find_tool, making it one of the only places in rustc where the name of the target matters rather than the contents of the target. This also for example means that if you take the target spec of x86_64-pc-windows-msvc and put it in a file called aarch64-foo.json, rustc will incorrectly try to find a linker for aarch64. Or more realistically if you name it foo.json, rustc will not know which target to find a linker for even though the target spec itself contains the architecture to compile for.
The text was updated successfully, but these errors were encountered:
Maybe we should split that part into a separate crate?
I feel like it is not exactly part of cc, just put it in cc for convenience, and sometimes we might want to change its API.
If we are going to put it in a separate crate though, we probably want to do the same for other parts of cc-rs as well, which will also improve compile-time.
Currently rustc has to pass the actual value of the
--target
argument tocc::windows_registry::find_tool
, making it one of the only places in rustc where the name of the target matters rather than the contents of the target. This also for example means that if you take the target spec of x86_64-pc-windows-msvc and put it in a file called aarch64-foo.json, rustc will incorrectly try to find a linker for aarch64. Or more realistically if you name it foo.json, rustc will not know which target to find a linker for even though the target spec itself contains the architecture to compile for.The text was updated successfully, but these errors were encountered: