From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.5-pre1 (2020-06-20) on
	ip-172-31-74-118.ec2.internal
X-Spam-Level: 
X-Spam-Status: No, score=0.5 required=3.0 tests=BAYES_05,TO_NO_BRKTS_PCNT
	autolearn=no autolearn_force=no version=3.4.5-pre1
Date: 26 Jul 93 18:49:26 GMT
From: rational.com!bonnie!rlk@uunet.uu.net  (Robert Kitzberger)
Subject: Re: TCP/IP in Ada
Message-ID: <rlk.743712566@bonnie>
List-Id: <comp.lang.ada>

labtek@cs.yale.edu (Tom Griest) writes:

>   \ASR\DDN   - written in 1985 by E-Systems
>   \ASR\DDN2  - written in 1987 by Unisys

Also note that the DDN code only supports UDP/IP, and the DDN2 code
only supports TCP/IP.  Neither supports both TCP and UDP.  Adding UDP
is a straightforward bit of work.  Adding TCP to DDN, on the other
hand, is ill-advised, since TCP is many times more complex than UDP).
Nonetheless, the DDN2 code has a much cleaner architecture, and is
easier to plug in device drivers and socket layers.  Unfortunately, as
you point out, the DDN2 code uses tasking too liberally, to the tune
of 10+ context switches per packet transmission or reception.  If your
compiler supports passive tasking, then you can passivate some of
these (your mileage may vary).  With quite a bit of tweaking,
passivating, and optimization, my socket-layer performance was 10% of
my device-driver layer performance.

Caveat emptor.

	.Bob.
--
Bob Kitzberger                          Internet:   rlk@rational.com
Rational, Grass Valley, CA              CompuServe: 70743,1550
"Man is a rational animal who always loses his temper when he is called
 upon to act in accordance with the dictates of reason." -- Oscar Wilde