dan@ptolemy.arc.nasa.gov (Daniel J. Dick) (02/01/90)
I tried compiling sendmail.5.61 and finally ended up going back to the old version of sendmail.mx that was on our machines for the following reasons: 1--One of our file servers was previously designated to be our mail deliverer, and messages sent out from our diskless workstations were set up to go out as if they were coming from that file server instead of the individual workstations so that a common mail area could be shared and all of the workstations would not have to be running a sendmail daemon. After installing sendmail.5.61, the headers suddenly changed so that mail was going out as if it were coming from the workstations causing replies from the outside world to fail. This happened without making any changes to any sendmail.cf files. 2--The time zones changed from PST to -0800 in the header fields. I did not know if this was proper, but it was different, and that made me hesitant. 3--I could not get sendmail.5.61 to compile properly under gcc, even though it compiled under cc. My knowledge is limited here so I would be very appreciative of anybody who might be able to help me here. -- Daniel J. Dick (dan@ptolemy.arc.nasa.gov) Systems Manager NASA Ames Research Center
moore@betelgeuse.cs.utk.EDU (Keith Moore) (02/02/90)
In article <3877@bayes.ptolemy.arc.nasa.gov>, dan@ptolemy.arc.nasa.gov (Daniel J. Dick) writes: > I tried compiling sendmail.5.61 and finally ended up going back to the old > version of sendmail.mx that was on our machines for the following reasons: > > 1--One of our file servers was previously designated to be our mail deliverer, > and messages sent out from our diskless workstations were set up to go out > as if they were coming from that file server instead of the individual > workstations so that a common mail area could be shared and all of the > workstations would not have to be running a sendmail daemon. > > After installing sendmail.5.61, the headers suddenly changed so that mail > was going out as if it were coming from the workstations causing replies > from the outside world to fail. This happened without making any changes > to any sendmail.cf files. This is not the fault of Berkeley sendmail. I suspect the only reason it worked in the first place is because Sun hacked their sendmail to be "smart" about mail clusters and to rewrite outgoing mail appropriately. It needs to be pointed out that different OS vendors have changed sendmail in significant ways, often breaking things (and adding security holes) in the process. Rarely are these changes documented, and if they are, there is no list of ``differences between our sendmail and Berkeley sendmail.'' I'm sure they feel as if they are doing us a favor. This is the main reason I run Berkeley sendmail on all of our machines -- not because it always works better, but because that way I can use the same configuration file for every machine (except the DECnet gateways). > 2--The time zones changed from PST to -0800 in the header fields. I did not > know if this was proper, but it was different, and that made me hesitant. This behavior was recommended by RFC 1123. The problem is there is little or no standardization for many timezone names outside the US. Numeric timezones are completely according to the standards. > 3--I could not get sendmail.5.61 to compile properly under gcc, even though > it compiled under cc. > It's not clear why you chose sendmail.mx over sendmail 5.61 based on the ability to compile with gcc. sendmail.mx was certainly compiled with cc rather than gcc. Keith Moore Internet: moore@cs.utk.edu University of Tenn. CS Dept. BITNET: moore@utkvx 107 Ayres Hall, UT Campus UT Decnet: utkcs::moore Knoxville Tennessee 37996-1301 Telephone: +1 615 974 0822
towfiq@interlan.Interlan.COM (Mark Towfigh) (02/06/90)
In article <3877@bayes.ptolemy.arc.nasa.gov> dan@ptolemy.arc.nasa.gov (Daniel J. Dick) writes: I tried compiling sendmail.5.61 and finally ended up going back to the old version of sendmail.mx that was on our machines for the following reasons: 3--I could not get sendmail.5.61 to compile properly under gcc, even though it compiled under cc. When you say it didn't "compile properly", do you mean the object files were not created, or that you could not link the executable, or that the executable did not run well? Did you compile with the -traditional option? Hope this helps, Mark -- Mark Towfigh, Racal Interlan, Inc. towfiq@interlan.Interlan.COM "The Earth is but One Country, and Mankind its Citizens" -- Baha'u'llah