Hector Martin

Use Chinese. Use Japanese. Use Cyrillic, Greek, Hebrew & Arabic. Use Unicode. Use it all. Choose life. Choose UX. Choose your own password.

If you think using Japanese or Chinese for passwords is reasonable, you clearly know nothing about how those two languages work.

Hint: the sequence of keypresses to type in those languages is unmemorizable and even nondeterministic, thus incompatible w/ password boxes.

Huh. I assumed that if a char is type-able, result would be independent of keystroke combos. Where can I read more about this?

Japanese is typed phonetically but that representation is ambiguous. Disambiguation is an interactive process with feedback.

And since modern IMEs are adaptive and take into account past usage, the disambiguation part is nondeterministic.

Read https://en.wikipedia.org/wiki/Japanese_input_methods#Kana_to_kanji_conversion … . In practice what ends up happening is your password is exposed until IME is done, then turns into ••••••••.

(Which was really annoying for me when KDE4 had that bug where it kept allowing the IME on password boxes; KDE5 fixed it)