<<< Date Index >>>     <<< Thread Index >>>

Re: when replying, use envelope recipient



On Tue, Dec 19, 2006 at 12:45:42AM -0800, William Yardley wrote:
> Really, no one has really answered his ACTUAL question, which is to use
> the envelope-recipient of a message. I don't know if that's what he
> actually meant, since he seemed to be happy with $reverse_name...

It was myopia.  I tried it, probably on a mailing list response
where I was CC'd personally, and it did what I wanted.  Subsequent trails
led to my mlist messages going to moderator limbo.  At first I thought
I'd accidentally disabled the desired functionality while editing and
mangling my muttrc files, but I finally realized that this was, in fact,
not sufficient after all.

> However, there's no way to do what he asked with mutt AFAIK, and it
> would be hard to accomplish, since some MTAs don't include the envelope
> recipient, or don't include it in all cases, and when they do include
> it, it's not in a standard form (Sendmail uses X-Apparently-To, Postfix
> uses X-Original-To, some just include it as part of the Received
> headers, etc.). So since he's asking about something that mutt doesn't
> do...

Well, if you folks can give me advice on the best way to do it using
an external helper script, I'll write that and distribute it, and if
I can get the logic working sufficiently well I might even be willing
to hack the mutt...

> I was thinking that it would be kind of cool if mutt had such a feature
> (you give mutt a header field name to check for the message's
> envelope-recipient, and mutt checks it against reverse name).

What would the purpose of this check against reverse_name be?
I'm not following your logic here.  The way I was envisioning it
was:

1) If use_envelope_sender is true and we know the envelope sender, we use
   that.
2) If not, it falls back to reverse_name, I suppose.

You know, the docs don't say _where_ reverse_name comes from... just that it
comes from "where you received the message".  I wonder what algorithm this
implies?  It sounds like what I wanted, but isn't.

> This would
> be handy for situations where you use a unique address for stuff (e.g.,
> mailing lists) where your name isn't necessarily in the To or Cc fields.

> But I think it would be hard to make such a feature work well, plus I
> doubt there's someone who would want to write a patch, so I never put in
> a feature request.

I don't think it has to work perfectly in every case to be useful.
If it doesn't work for them, people won't use it.

I'm fortunate somewhat that I am the admin on boxes where I want to use
this.  I really like the idea of being able to backtrace where a spammer
got my email address... or how someone got my email addy when emailing
me.  I saw one guy whose system created a very long base64 alias for
every outgoing email, which he could revoke if it ever made it onto a
spammer's list.  Better yet, he could use it as a spamtrap address...

So basically I'm using postfix, so the header is always Delivered-To.
If I need to worry about duplicate or confusing headers, I'll filter
them out, or use the first line that matches (which will be the most
recent MTA, mine, I suspect), or do something I haven't thought of yet.
It shouldn't be all that difficult.

So... I know how to write the filter that extracts this... how best
to hook it into mutt's various reply/bounce/resend/forward routines?
-- 
A: No.
Q: Should I include quotations after my reply?
<URL:http://www.subspacefield.org/~travis/> -><-

Attachment: pgpATzH32kfTj.pgp
Description: PGP signature