Received: from localhost (daemon@localhost) by CS.UTK.EDU with SMTP (cf v2.9s-UTK) id MAA15078; Thu, 26 Sep 1996 12:50:59 -0400 Received: by CS.UTK.EDU (bulk_mailer v1.6); Thu, 26 Sep 1996 12:50:41 -0400 Received: from SEARN.SUNET.SE ([192.36.125.4]) by CS.UTK.EDU with SMTP (cf v2.9s-UTK) id MAA14985; Thu, 26 Sep 1996 12:50:27 -0400 Message-Id: <199609261650.MAA14985@CS.UTK.EDU> Received: from SEARN.SUNET.SE by SEARN.SUNET.SE (IBM VM SMTP V2R3) with BSMTP id 4888; Thu, 26 Sep 96 18:46:59 +0200 Received: from SEARN.SUNET.SE (NJE origin ERIC@SEARN) by SEARN.SUNET.SE (LMail V1.2b/1.8b) with RFC822 id 9968; Thu, 26 Sep 1996 18:47:00 +0200 Date: Thu, 26 Sep 1996 18:36:32 +0200 From: Eric Thomas Subject: Maximum userid/hostname lengths To: drums@cs.utk.edu What are the current maximum values for local-part and hostname in SMTP mail? These were defined as 64 + 64 by RFC821, but I'm sure the hostname would have been redefined in a more appropriate standard, and I'm not sure the local-part was ever changed (not in 1123 anyway). Also, are there known cases where values significantly larger than 64 are used? I'm aware of the fact that X.400 gateways routinely exceed this limit. I know what values are safe, and it's easy to implement mail software so that arbitrary lengths are supported, so it's normally a non-issue. I just have a customer who wants to size a database field so that there is a 100% guarantee that any SMTP address will fit. My take is that this is unrealistic, unless you're willing to spend the disk space on a huge value. Hostnames are unlikely to ever get significantly longer than 64 bytes, because people are going to have to be able to type them, but I can see gateways stuffing more and more information into the local-part, not to mention %-hacks, etc. Send an MCI address through a X.400 gateway and back and I wonder how many miles it takes up. Eric