Manuals, windows, auto-newlines

[Chris Torek]

In article <238 at brl-smoke.ARPA>, gwyn at brl-smoke.ARPA (Doug Gwyn) writes:
>With stream i/o networking, and suitable protocols, it should
>be possible to have the "terminal handler" on the remote end
>completely aware of the terminal state.

This is clearly the way to go.  It would solve all those little
glitches in rlogin.  Sigh.

>Occasionally we even get that here using 4.2+BSD "rlogin", so far as
>initial window size is concerned.  (I don't think window changes are
>passed along, but they could be and should be.)

They are passed along from our Suns to our Vaxen (I moved the 4.3
rlogin to the Suns).  There is, however, a bug in the 4.3 rlogin.
The server and client must do a tiny bit of negotiation to agree
to pass window sizes: the client should not do it unless the server
understands it.  This was done by passing an out of band data byte
with the high bit set.  Unfortunately, this depends on the timing
of an F_SETOWN ioctl, and works on Vaxen and not on Suns.  To fix
it requires rewriting rcmd(), or expanding it in line (or a kernel
change to start sockets owned by their creators).  I left it unfixed,
so that our Suns always pass window size changes, even if the server
does not understand them.  Sigh.
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 1415)
UUCP:	seismo!umcp-cs!chris
CSNet:	chris at umcp-cs		ARPA:	chris at mimsy.umd.edu