Re: 23.1; emacsclient -t leaves terminal in an odd state if daemon not started

[Chong Yidong]

Anyone have any idea about Bug#5095?  I can reproduce this:

emacs --daemon
emacsclient -t
[up]               =>     "M-o A is undefined"

Subsequent function keystrokes are handled correctly.

So it appears that the input-decode-map is not set up properly for the
very first keystroke on a terminal frame created by the daemon.  But
I've no clue why.