I would be thrilled if 99.9% of my input files were UTF-8, but I'm not the one writing them. SJIS remains in widespread use in Japanese companies and tons of legacy codebases do not work with UTF-8. If you're lucky there's a config flag for UTF-16.