gst@gnosys.svle.ma.us (Gary S. Trujillo) (10/18/89)
I'm going crazy hand-rolling return addresses to received mail. I'm using ELM 2.2 [PL10] (yes I know there's a patch #11, but I'm too busy to install it just now). Following is a modified version of a query I sent to a couple of folks who I thought might be able to help, but I guess they've been too busy to respond. Here's what I sent (the site name has been replaced with "foobar" to protect the innocent): Back in April, we exchanged a couple of messages regarding a problem I started to experience then, having to do with dealing with messages passing through foobar. I just installed smail, and have been motivated to take a closer look at how mail addressing works, and think I have discovered the problem I reported earlier, and have a suggested fix. To refresh your memory, here's an excerpt of our earlier correspondence: ||To:foobar!manager Tue Apr 18 12:46:56 1989 ||Date: Tue, 18 Apr 89 12:46:56 EDT ||To: foobar!manager ||Subject: problem with mail headers || ||Greetings. || ||I just wanted to bring to your attention that sometime within the past week, ||there has been a change in the format of the ">From" line in messages received ||via uucp from foobar. This change has resulted in a couple of unpleasant con- ||sequences relative to the ELM mail system. I am not sufficiently aware of ||what the RFC822 standard says regarding mail headers and their parsing to ||claim that the change is outside the spec. Nor do I know if version 2.2 of ||ELM might deal more gracefully with the change, as I have not yet installed ||that recently-released version. What I do know is that the change is going ||to be hard to accomodate unless it is dealt with or can be dealt with by ||a change in how the message header is parsed by ELM. || ||First, here's the sort of thing I used to get: || ||>From decwrl.dec.com!escd!es37!user Tue Apr 11 16:55:40 1989 remote from foobar.harvard.edu || ||Now I get: || ||>From escd!es37!user@decwrl.dec.com Fri Apr 14 20:55:28 1989 [BTW, I did just install ELM version 2.2 (PL10), and notice no difference with respect to the reported difficulty.] You replied to my earlier note: |sorry, I did not see the previous message. | |The address you site is a totaly legit 822 address ( as a mater of fact |it is the style that is suggested by most people working in the area and |is the form that all internet address are in ) | |I think that you should look into fixing ELM since it would need to |be fixed if you ( or anyone using ELM ) were ever to be on the internet. Well, the problem is not with the address, but with the absence of the "remote from ..." information in the ">From" line. After careful ex- perimentation, I have discovered that, if one is not relying on the address information contained in the "From:" line (which I am not, as I have found it to be inconsistent and unreliable), the only way the reply feature of ELM can determine the return address is by constructing a return path based on what's in the ">From" line. It does so by con- catenating the hostname given in the "remote from <hostname>" part of the line to the rest of the address. Thus, in the example I mentioned earlier: >From escd!es37!user@decwrl.dec.com Fri Apr 14 20:55:28 1989 if the line had a "remote from foobar" at the end, ELM could create a valid return address "foobar!escd!es37!user@decwrl.dec.com" easily. (Yes, I know the "escd!es37" part is non-essential, but I think it doesn't hurt, as it can be stripped and replaced by sendmail in the remote host.) Whether or not one is using smail (or sendmail, for that matter, which I have also installed), the above would permit one to use the reply command, instead of having to laboriously roll their own return address. Now there could be some way I could get sendmail to paste the name of the host a message came from into the header. In fact, I'd be rather dis- appointed in it if it couldn't. However, I haven't yet gotten into the baroque details of how to configure the thing yet. In any case, I suspect it would be fairly easy for you folks to restore the "remote from ..." information. In any case, thanks for listening. Gary -- Gary S. Trujillo gst@gnosys.svle.ma.us Somerville, Massachusetts {wjh12,spdcc,ima,cdp}!gnosys!gst -- Gary S. Trujillo gst@gnosys.svle.ma.us Somerville, Massachusetts {wjh12,spdcc,ima,cdp}!gnosys!gst
rsalz@bbn.com (Rich Salz) (10/18/89)
In <399@gnosys.svle.ma.us> gst@gnosys.svle.ma.us (Gary S. Trujillo) writes: >Well, the problem is not with the address, but with the absence of the >"remote from ..." information in the ">From" line. After careful ex- >perimentation, I have discovered that, if one is not relying on the >address information contained in the "From:" line (which I am not, as >I have found it to be inconsistent and unreliable) ... I assume that by ">From" you really mean the "From_" (_ == space) line line that is the UUCP "envelope"? Ouch. Don't count on that. Count on the From: line. Sites are much more likely to leave that alone, and have it a correct domain address, then they are the From_ line, which is usually mangled by every UUCP site along the way, but not all of them.... /r$ -- Please send comp.sources.unix-related mail to rsalz@uunet.uu.net. Use a domain-based address or give alternate paths, or you may lose out.