Re: rtm and uucp


alice!dmr@ucbvax.Berkeley.EDU
9 Nov 88 08:52:32 GMT


References: <1445@anasaz.UUCP> <772@mailrus.cc.umich.edu>

Pursuant to the responses of Honeyman and Mitchell to the worries
of Moore and Nagle:

Robert Morris (rtm, Morris Minor, the little enchilada) spent two
summers, several years ago, in our group at Bell Labs. During
the first, his major accomplishment was a complete rewrite of
the uucp and accompanying software. As Peter noted, his version
was considerably more secure than previous versions, and some
of his insights influenced HoneyDanBer uucp. We ran it on our machines
for nearly a year thereafter, but dropped it in favor of HDB,
mainly because HDB was rapidly gaining favor within AT&T, and Robert's
version had no superiority sufficient for us to push it or keep
it going in the absence of its author. I believe it was
free of intentional trapdoors, unlike sendmail.
In any event, the code is long gone except from backup tapes.

The second summer, his major product was a streams implementation
of TCP/IP that is still the basis of the Eighth/Ninth edition
version of that module. It has since been reworked considerably,
mainly to remove the vestiges of the socket mechanisms (he started
from the Berkeley code), but again, we have never found any evidence
of funny business that wasn't in what he started with.

None of the work he did is in any product, and he didn't have
any opportunity to tamper with the master source code--
that is really quite far away from Research.

                Dennis Ritchie



This archive was generated by hypermail 2.0b3 on Thu Mar 09 2000 - 14:44:30 GMT