 |
 |
"George J. Perkins" <[log in to unmask]> writes: > On Tue, 26 Apr 2005, m. kolb wrote: > >> Dennis Boone <[log in to unmask]> writes: >> >> > This morning's "entertainment" leads me to the following question: >> > >> > Why doesn't mail.msu.edu add message-id headers when it receives a >> > message without them? >> >> That doesn't make sense. The message-id is an origination mechanism. >> It is our servers job to include a "Received:" on incoming messages. >> Placing a Message-ID on a message we are not sending (but delivering) >> just muddies up the water, and would actually be a violation of the >> RFC. > > No. RFC2822, section 3.6.4 says that every message SHOULD (but is not > absolutely required to) have a Message-ID header. Its only requirement is > that it be globally unique. While it is obviously best if the originating > system puts it there, and subsequent systems keep it as long as the content > of the message remains the same (not counting additional Received: headers > and the like), there is nothing that says that a message without such a > header cannot acquire one from a later system on the path to the end user. You are correct. I checked 2821 for comments (and obviously 2822) and 2821 clearly demonstrates your point in section 6.3, and goes on to say: The less information the server has about the client, the less likely these changes are to be correct and the more caution and conservatism should be applied when considering whether or not to perform fixes and how. These changes MUST NOT be applied by an SMTP server that provides an intermediate relay function. Again, it is not a RFC violation as George has stated. ./muk -- m. kolb <[log in to unmask]>