some terminals (Rxvt?) can display simultaneously Latin-1 and UTF-8.
Something has to provide the mode-switch between UTF-8 and Latin-1.
I meant *without* mode switch. Something that would print (fake):
| $ printf "\0351 \0303\0251\n" # 1 Latin-1 e acute, and 1 UTF-8
| é é # 2 e acute glyphs
I never saw that with my eyes, only got a report about rxvt-unicode.
I imagine this could work by interpreting input as UTF-8, but on each
erroneous byte reinterpret it as being Latin-1. I can also very well
imagine that it's not rock solid, as some sequences of Latin-1 chars may
happen to appear as a valid UTF-8 character. But this trick can probably
give good results in practice, on say a Latin-1 French text. Wild guess,
never saw, should not talk.