datta@vacs.uwp.wisc.edu (David Datta) (03/10/91)
Now that we have all determined that the Errors-To: is ignored in many cases and that modifying the headers so bounced mail isn't forwarded to the list, is considered "a bad thing" what should we do about it now? I run 2 mailing lists and I have been changing the headers, recipient's MUA be damned. After our system was brought to it's knees due to stupid mailers bouncing mail to the list, I started changing the headers. I know this is rather harsh but the reason for having the list is that the sender wants his mail forwarded to the list, the recipients don't want to read mailer-daemon mega-bounces, we don't want our machine shoveling daemon bounces back and forth. The simplest and only solution I could find is to start munging the headers. It makes it difficult to reply only to the sender but I feel that this solution is the best of two "bad things". So, instead of discussing "you are doing an evil", how do we go about fixing the problem? Does someone have something that both WORKS and is GOOD? -- -Dave datta@vacs.uwp.edu [....uwm!uwpvacs!datta] Are you a lemming? Lemmings jump off cliffs every four years. There are no five year old lemmings. Unless they've learned to think for themselves.
david@dhw68k.cts.com (David H. Wolfskill) (03/16/91)
In article <10095@uwm.edu> datta@vacs.uwp.wisc.edu (David Datta) writes: >Now that we have all determined that the Errors-To: is ignored in many >cases and that modifying the headers so bounced mail isn't forwarded to >the list, is considered "a bad thing" what should we do about it now? It's not clear to me that header-modifications to prevent bounce-o-grams from entering mailing lists are necessarily a "bad thing". >It makes it difficult to reply only to the sender but I feel that this >solution is the best of two "bad things". That seems to depend on the MUA(s) in question. I use Berkeley Mail; it is capable of making the distinction -- the way I modify the headers. >So, instead of discussing "you are doing an evil", how do we go about >fixing the problem? Does someone have something that both WORKS and >is GOOD? This is not guaranteed to be universally approved; indeed, I had an extended conversation with a participant in this discussion who expressed some misgivings about what I do. However, it seems to work for me -- I just wish the other mailing lists I'm on were run similarly (so I could easily reply either to the entire list or to the originator of a message -- without editing the To: line by hand). First, this machine is only connected via UUCP. Thus, submissions for mailing lists that arrive here are likely to have several Received: headers in them by the time they get here. If I turn around and send such a submission out to the list, intermediate sites that run sendmail will check on the total number of Received: headers -- so if a site > 8 hops away sends something to the list (and assuming for the moment symmetric paths), that person will probably never receive the "mailing list" version of what was sent. This is because the total number of Received: headers will be > 17 -- and sendmail (by default) seems to act as if > 17 Received: headers indicates a mail loop. Therefore, I strip all Received: headers from submissions for mailing lists (that are based here) that get to this site. (After all, if the note got here, that's not a problem; if it didn't, I don't get a chance to strip the Received: headers anyway.) Next (to actually address your concern), I also throw away any pre-existing Reply-To: header, and replace it with one that refers back to the mailing list. The From: header is left strictly alone; Sender: and Errors-To: headers are placed in the message, pointing to the -request address (so I get the bounces). Lastly, a "Precedence: bulk" header is inserted. Upon receipt of such a message (from a mailing list), the UCB Mail "reply" command sends a reply to the entire list; the "Reply" command (note case distinction) sends a reply to the originator of the message. Naturally, I have a combination of specific /usr/lib/alias entries, a shell script, and a program (FSA, actually) to do this for me. I've sent copies of this stuff to various folks in the past; if there's sufficient interest, I suppose I could toss it over to comp.sources.misc. david -- David H. Wolfskill uucp: ...{spsd,zardoz,felix}!dhw68k!david InterNet: david@dhw68k.cts.com CompuServe: >internet:david@dhw68k.cts.com