Path: utzoo!mnetor!uunet!husc6!rutgers!ucla-cs!zen!ucbvax!XX.LCS.MIT.EDU!ROMKEY
From: ROMKEY@XX.LCS.MIT.EDU (John Romkey)
Newsgroups: comp.protocols.tcp-ip
Subject: Re: Need TCP/IP on RSTS/E
Message-ID: <12364531098.11.ROMKEY@XX.LCS.MIT.EDU>
Date: 6 Jan 88 22:37:53 GMT
References: <306612.880104.JBVB@AI.AI.MIT.EDU>
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 26

> All fine, except that PC-IP is 1) dependent on an 18 ticks/sec clock in a
> lot of places 2) very 808x-dependent in the tasking, timer and display
> modules and 3) has a built-in hardware driver which would need a complete
> re-write for another interface/architecture.  The problem with RSTS is

It's not *that* much of a problem. The 18 ticks/second clock is parametrized
in a constant in a header file so that all references to ticks can
be done as seconds * (ticks/second). It's not really a problem through
most of the code.

The 8086 dependencies are simply that there's some assembly code. In
the tasking package, you need to rewrite the code which does a context
switch. The timer code isn't 8086 dependent, and anyone who might be
porting PC/IP to another system would probably through out the entire
display package, anyway.

Any hardware drivers would have to be rewritten from scratch, anyway, since
it's a different set of hardware.

PC/IP origially ran on a PDP-11. If porting it back to a PDP-11 turns out
to be difficult, it shouldn't be because of architectural problems
or because of 8086 dependencies. PC/IP was derived from the PDP-11
V6 Unix code Mike Patton mentioned in a later message. I don't know if you
can still find someone at MIT who can or will give you that code.
				- john romkey
-------