[H-GEN] software runs the world

Russell Stuart russell-humbug at stuart.id.au
Sat Mar 19 19:13:05 EST 2005


On Sun, 2005-03-20 at 00:05, Greg Black wrote:
> There are no addresses to re-write here.  We're talking about
> adding a Reply-To header on email that you send, not about
> changing anything on email that you receive.

No, actually in my case I will be re-writing Reply-To in my
outgoing messages, if it is set.  The reason is I use a 
different return address for each list for spam control.
I often forget to set the dammed thing.  If I do it 
automatically in exim, I don't have to remember.

As for the use of the word "mung" - I used it wrongly.  Unlike
you I am not as precise in my use of the language as I should
be.  I was repeating the term as it was used in the "Reply-To
munging considered harmful" meme, which is usefully brought up
when this topic is discussed.  As you point out, except in my
unusual case, Reply-To's are usually added only if they don't
exist.

> > I did think about how I could re-write the reply to address
> > with my existing sendmail / Red Hat setup, but I could not 
> > think of way that only required the tiny amount of work 
> > you say it does.
> 
> I made no statement about doing that, since we weren't talking
> about that (although it is of course trivial[1], but wrong, to
> do).

Yes, it is for you and I.  Not always for others.

> You should not even be considering re-writing any Reply-To
> address.  If incoming email has a Reply-To header, then you're
> supposed to leave it alone -- it's there so your MUA can
> construct a To header correctly (and that's the reason why
> mailing lists aren't supposed to touch it either).  What we're
> talking about here is adding your own Reply-To header to email
> that you compose.

Well, I 1/2 agree.  Definitely, when someone such as yourself
has deliberately sets the Reply-To it should be respected -
absolutely.  The problem is - how do I know if the presence
of Reply-To is a reliable indicator of what the sender intended?
On this list if they set it, then it probably is.  But if it
isn't set what do I do?  I recall remarking a while ago that
you were one of the very few who actually bothered to set it.
That is still true, I think - Sandra being the only other
instance I can think of.

There are a lot of procmail scripts flying around that assume
if you didn't set it, you simply forgot and intended all replies
to go back to the list.  What is you opinion of that?

My opinion is that it is the right assumption most of the time.
When you get a message from this list that doesn't have the 
Reply-To the most likely thing is the person either is at best
indifferent to whether the reply goes back to the list or not.
That might be wrong, of course - as things stand they might
know how the list works and intended the reply to go back to
them only.  However, I can't know that, so the best thing to
do is to send the reply to the list - which is what the
procmail script does.

This is a fine solution where procmail is an option.  But for
a lot of people procmail isn't an option - as I said in my
reply to Rob.  To me it appears the best thing to do is to
add a Reply-To in mailing list software to incoming messages,
allowing senders to turn the feature off if they want total
control over Reply-To.  This would I would have to make no
assumptions as to what they intended if Reply-To isn't set.

> > Of course its is entirely possible you 
> > know more about Unix and programming than I do, but I have 
> > to balance that thought against the fact that you have not 
> > done it, despite programming being second nature to you.
> 
> I said it's not worth it and it's not in my opinion.  You
> certainly need to understand this stuff better than you do if
> you're going to try to automate it.

Another wild conclusion, Greg.  Again, you have no idea how
much I know about email, smtp, or the rfc's governing those
things.  I wish you would stop making these leaps of faith,
at least where they concern my abilities.






More information about the General mailing list