davec@tektronix.UUCP (06/25/83)
Recently a couple of net messages were sent out by sites complaining
that they were being called up by the site "burl" which they had no
known uucp connection with.
A little later, Larry Auton ("burl" system administrator?) sent out
an explanation saying that he was calling those sites. The information
to make the call came from the site "ihnp4"; apparently it broadcasts
out either all or part of its L.sys file to other AT&T sites on a
somewhat regular basis. His mailer looks ahead in paths of mail passing
through its system, and if it finds a shorter path, sends the mail
through the shorter path.
There are a couple of problems with this procedure:
1. When Gary Murakami (sp?) of ihnp4 originally solicited uucp connection
information from non AT&T sites no mention was made of the fact that such
information might be broadcast out to other systems. There are a number of
people that would be very uncomfortable about such a procedure (system
security; potential load on dial-up lines; etc).
2. Using broadcast connection information to send mail creates one way
mail paths where the path that the message took to reach its destination
cannot be used to make a reply.
Dave Clemans
Tektronixsilver@csu-cs.UUCP (06/29/83)
It is a breach of etiquette (yeah, I say so) to call a system without their knowledge, just because you know how. It also creates the only possible (?) kind of one-way uucp connection because the other system, not knowing you, cannot even queue files to be sent your way.
pn@amd70.UUCP (06/29/83)
Gary did ask if one's L.sys info could be redistributed. Possibly most people that asked for a connection forgot to specify, and so their info did get sent out. I applaud the idea of optimizing mail routes, so much of the mail which goes through my site is poorly routed. The creation of one-way mail paths is a problem, though.
lda@burl.UUCP (07/01/83)
I have corrected my error in sending mail to sites who do not have a return path. Martin Levy specifically stated in his README that mail was *not* to be sent to sites without return paths. I used a program to generate my routing file which was *supposed* to insure two way paths. I have verified (manually) all the paths now myself. I will not allow that sort of thing to happen again. My apologies for the inconvenience.
gjm@ihnp4.UUCP (07/01/83)
This is in rebuttal to the previous articles about the burl controversy. (1) While ihnp4 does distribute L.sys information help to centralize the administration of uucp within the Bell System, it does NOT distribute information without the explicit permission of the non-Bell parties involved. (2) I send out a form letter containing information for ihnp4. This letter includes an explicit request for permission or denial on redistribution of information. I appreciate amd70!pn's support in pointing this out. (3) Information for systems tektronix, linus, and uw-beaver is kept private to ihnp4. I purposely try to make a point of privacy or redistribution, especially to avoid problems and accusations like some of the previous articles. I also maintain a copy of all of my correspondence and can look up conversations in question. (4) Please be careful about the statements made about people on the net. The statements insinuating that ihnp4 breaks confidence are WRONG, but the damage has already been done. I suggest that the party in question be contacted first before any statements are made to the general public (i.e. I consider myself to be a professional, and I'm upset about some of the previous statements). (5) In the next distribution to Bell sites, I broadcast a note emphasizing proper etiquette in contacting any one way neighbors. Gary Murakami Bell Labs - IH ihnp4!gjm P.S. I'm sorry that I couldn't respond earlier, I was in New Jersy until today.
jim@uw-beaver.UUCP (07/01/83)
Gary at ihnp4 does indeed ask first before distributing L.sys info.
Here is an excerpt from his connection info message:
Please specify if your L.sys info can be sent to other Bell System
sites, or if it should be kept private on ihnp4.
I suspect that a lot of people read right past that and don't realize
what it says. I know I did. If you have ever had any dealings with
Gary, you know that he is dedicated and professional about his work. I
know everyone on the net is sick and tired of apologies, but let me
offer mine to Gary if I have helped to tarnish his reputation.
Let me also thank the folks at ihnp4 for volunteering to be a uucp
central clearing house. It's a job I wouldn't want, and I think it has
helped all of us more than it has helped them.smk@linus.UUCP (Steven M. Kramer) (07/02/83)
The ol' foot in mouth. I apologize. I spoke too soon. I helped the
damage to occur, so maybe this will help reverse it (although things
never seem to work that easy). I was a victim of spreading panic.
Whoops.
--
--steve kramer
{allegra,genrad,ihnp4,utzoo,philabs,uw-beaver}!linus!smk (UUCP)
linus!smk@mitre-bedford (ARPA)pn@amd70.UUCP (07/02/83)
Re uw-beaver!jim's appreciation of ihnp4 as a uucp clearing house, I'd like to say "amen". Thank you, Gary. Phil Ngai
bhayes@sri-unix.UUCP (07/08/83)
#R:csu-cs:-229000:sri-unix:11900004:000:229 sri-unix!bhayes Jul 4 16:44:00 1983 Optimizing mail routes seems like a good idea, but a hook should be left in to let someone ensure that a message is taking some path. This would allow testing for and debugging of connections with ease. -Barry sri-unix!bhayes