Errors-to: owner-drums@CS.UTK.EDU
Message-Id: <199509200048.UAA03881@CS.UTK.EDU>
To: drums@CS.UTK.EDU
From: "Roger Fajman" <RAF@CU.NIH.GOV>
Date:     Tue, 19 Sep 1995  20:44:07 EDT
Subject:  Re:  Two reply commands

> From your viewpoint, the only possible solution would be always to
> force the write of a reply to specify which of the recipients
> of a multi-recipient message to send the reply to (and also
> of course allow the user to add additional recipients who did
> not get the original message at all.)

Yes, that is the solution I am suggesting.

> To have one single reply command, with one default set of recipients,
> would simply make things even worse. Thus, you are saying that
> no defaults should be allowed for the recipients to a reply.

Yes, for multirecipient messages.  For a lot of personal messages there
is only one address to select, so a default may be reasonable there.
It also emphasizes the case where there are multiple addresses to
possibly reply to.

> My experience is that this is a case where uses often make mistakes.
> Especially embarrasing is when users by mistake send personal messages
> to larger groups than intended. So this is a reason for not having
> any default recipients.

Yes, that's true.  Also, it's not embarassing, but it is still a problem
when users send replies to smaller groups than intended (i.e., a reply
that should have been sent to a list, but wasn't because the user didn't
think about what they were doing).

> However, people wants things to be simple, so would users accept
> a mail system which did not use any defaults for reply recipients?
> The choice of two commands, or two variants of a command (plus
> of course the option of specifying recipients individually)
> might be what users want?

I am using a system now that works that way (it was developed here).
I have not heard complaints about that feature.

There's a saying that says one should make something as simple as
possible, but no simpler.  It is simply not the case that there
are only two reasonable ways to reply to a message.  Selecting
addresses from a list is not hard work.  It's not as if the users
had to retype them.

I would also like to point out that solving a problem in the user
agent helps people faster than solving it by defining new headers.
With a user agent fix, once I've started to use a UA with the fix,
the problem is solved for me.  If I have to wait for the majority
of the systems on the Internet to adopt new headers, it will be
much longer before I see the benefit.  And there will be more
confusion during the transition, as things work one way for some
messages and another way for others.