[e2e] FTP data port

[Bob Braden]

  *> 
  *> Why the connection was opened by the server is probably because both ends 
  *> needed socket numbers to be chosen and verified, for "security" reasons, 
  *> and the logical end for the SOCK command to be originated from was the client.
  *> 

Dave,

I recently reviewed the early FTP specs, and I was very much involved
in the FTP spec in 1973-1974.  John Day's explanation of how TIPs worked was
right on, and it is an adequate (and the correct) explanation for the
choice of parity of connections.  It allowed a user on host X to cause
the server on host Y (normally a Tenex ;-)) to send a print file to
a printer port on TIP Z.

Somewhere back there you will find comments about these issues,
probably in an RFC by Alex McKenzie.  (When I have time, I will pull it
out.)

Bob

  *> Since I was only a spectator at the time (I got involved in the Internet 
  *> project, and only kibitzed on ARPANET stuff), I may be wrong.
  *>