Path: utzoo!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!cs.utexas.edu!yale!mintaka!bloom-beacon!BRL.MIL!phil
From: phil@BRL.MIL (Phil Dykstra)
Newsgroups: comp.windows.x
Subject: Re:  Remote xterm strategy
Message-ID: <9001081442.aa01513@SPARK.BRL.MIL>
Date: 8 Jan 90 19:42:37 GMT
Sender: daemon@athena.mit.edu (Mr Background)
Organization: The Internet
Lines: 17

> Rlogind uses select(2) to keep
> track of both the master side of the pty and the network connection
> instead of having separate reader and writer processes as is used by
> rlogin.  I don't know why rlogin doesn't do this also.  It's quite
> possible that there's no reason other than history.

For what it's worth, I once implemented a single process rlogin client
via select.  The sad fact was that it ended up slower (i.e. more sys
time) than the two process version!  Select turns out to be a pretty
expensive syscall - more expensive than the single sided sleep/wakeups
of the two process version.  Studying the kernel code convinced me that
gprof wasn't lying, so we never installed the new one.  Sigh.

[This was 4.3BSD on a Vax 780 about two years ago.  You mileage may
vary, especially based on your machine's context switch overhead.]

- Phil