On Thu, Nov 3, 2022 at 7:01 PM John Cowan <[email protected]> wrote:
> There is a fundamental problem with having an `errno` codeset that > contains numeric codes, namely that the numerics are OS-dependent. For > example, ENOENT is 1 pretty much everywhere (though on IBM Z/OS it's 3465), > but ENAMETOOLONG is 36 on Linux, 63 on BSD and MacOS, 91 on Cygwin, 38 on > Windows, etc. etc. > >> But can't the implementation account for that, i.e. supply the right mapping for the device on which it is running?
