Re: junk characters from movement keys in vi in xterm via ssh

["James K. Lowden" <jklowden%schemamania.org@localhost>]

David Laight wrote:
> 
> The 'problem' happens because the keyboard characters don't get sent
> in one TCP message - and the delay within vi/curses is shorter than
> the nagle delay (or TCP ack time) for sending the subsequent bytes.

Why would an application sensitive to such a thing not defeat the Nagle
Algorithm with TCP_NODELAY?  

> Look at the 'vi' option 'escapetime'.

I can't understand what use that could be with TCP_NODELAY.  Why would the
terminal not always send the escape sequence as a single (small) packet? 
Or is it a workaround from former days when TCP_NODELAY was not
implemented?  

--jkl