Re: mail-followup-to / mail-copies-to

On Thu, 26 May 2005, Bruce Lilly wrote:

> Well, if an MUA user selects "reply (only) to list", is it really 
> surprising that the reply goes only to the list mailbox? 
> Ultimately, what a respondent chooses will affect where a response 
> goes, no matter what somebody else might suggest.  That's a key 
> point.

I agree. If a respondent really wants "send to list", they should get 
that.

However, if I would suggest this "list-reply" functionality should be 
taken as "reply in a list context" rather than "reply /only/ to 
list", particularly since one popular MUA /will/ copy those who 
indicate a preference for it - using a (unfortunately non-standard) 
mail header - most definitely not "reply /only/ to list".

This MUA also ignores From/Reply-To when "reply in list context" is 
used.

> The field simply specifies "the mailbox(es) to which the author of 
> the message suggests that replies be sent".  Whether a response is 
> "public" or "private" is at least as much under the control of the 
> respondent. Now it's certainly true that the Reply-To field doesn't 
> indicate *why* the message author has made the suggestion in any 
> machine-readable form, but an author who wishes to indicate a 
> reason can use a phrase (in any suitable charset and language using 
> RFC 2047 mechanisms) in named group syntax or name-addr syntax, or 
> in a (mildly deprecated) comment.

So you're saying I should have something vaguely like:

 	(please include me in list replies) <[email protected]>

as my From address on list email? I'm not sure that'll work. I have 
actually tried something like this before. Most people dont look at 
the addresses, they just hit 'reply', 'reply-all' or 'list-reply' and 
figure their MUA will work it out. All I want is a way to help those 
other MUAs figure out that I /do/ like to get copies for /list/ 
replies.

> In your specific case, distinguishing between suggestions for 
> "private" and "public" responses won't help, because you want 
> responses to go to *both* a public list and to your personal 
> mailbox.

No, not quite.

I dont want responses to go to both a public list and my personal 
mailbox. But that's all I can achieve with Reply-To.

What I want is to be able to indicate a preference on whether I 
prefer a direct copy when an MUA decides automatically which 
mailbox(es)? to send email to when its user has asked for a "list 
context" reply.

> "Only to list" really has to mean "only to list".  If there were a 
> hypothetical way for a message originator to cause a recipient's 
> MUA to do something else for "reply only to list", matters would be 
> far worse than now (I think Keith Moore made that observation last 
> year).

I would disagree that the list-reply functionality of some MUAs means 
"reply /only/ to list". One of more popular MUA which implements 
list-reply supports this view (as it will take MFT into 
consideration).

> Without a) a carefully worded question, and b) a scientific, 
> statistically significant sample of responses, I would hesitate to 
> make such a claim.

Agreed.

Sorry, that assertion is made based only on vague anecdotal evidence 
- I should have made that clear. (I asked the question on the 'ILUG' 
list a while ago, the responses were a small sample of its 700 
subscribers, all but 1 indicated they preferred to not receive direct 
copies).

> Sometimes it's beyond the scope of what standardization can do.

Standardisation can not do it, I agree.

Standards can /allow/ for it however.

A complete lack of suitable standards can make it impossible. (And/or 
lead to ad-hoc'ery, eg MFT).

>> However there is no hope of influencing future implementations to do
>> 'the right thing' without some attempt to standardise what this
>> 'right thing' is. (To be specific, this "right thing" for me
>> currently is a way to allow me to specify my unusual preference to be
>> included in list replies).
>
> "specify [a] preference" == "suggest [where] replies be sent" == Reply-To.
>
> Whether or not a particular recipient will honor that suggestion, or
> whether his choice of MUA will permit it, are separate matters.

That is fine.

I'm not arguing for MUST, I'm arguing there should be something at 
least that allows for MAY.

> Many UAs provide multiple response options.  If a respondent 
> (habitually or otherwise) chooses something other than "where the 
> author suggested" there isn't anything you can do.

I agree.

The problem is there is no good way for me to even make the 
suggestion.

As you point out above, Reply-To would suggest a preference for both 
'public' and 'private' replies, which is /not/ quite the preference I 
/really/ want to indicate to others, or anyone else like me who 
prefers to receive direct copies on list replies and any subsequent 
replies in a thread. (And the difference is critical to me, even if i 
fail to convince anyone here of the need for extra standards work, I 
wont be using Reply-To, because of limitations/dangers in its use in 
a list context).

> One thing is certain; if you do not provide a suggestion (via the 
> established standard mechanism), you guarantee that no respondent 
> can comply with your wishes other than by accident.  In the absence 
> of such a suggestion, some respondents may be forced into 
> "reply-to-all" (which happens to be compatible with what you want), 
> whereas others may choose "reply-to-list-only", etc.

A means to provide the suggestion is all I want :)

>> I very much wish to avoid getting stuck in #2-#4.

> You can't avoid #2; the respondent's freedom of choice trumps the 
> author's suggestion.

Of course, I am not at all arguing for  "MUST include the mailbox in 
the Foo-Copy-Me-Please in any replies" behaviour.

I meant more I do not wish to try change any of the already 
standardised headers that affect the broader scope of #2-#4, or how 
MUAs present functionality to users.

I merely wish to have a /valid/ way of indicating my preference to an 
MUA, when its user asks for something along the lines of "list 
context reply". That, eg, Mutt and at least other MUA seem amienable 
to modifying their list-reply behaviour based on information in 
headers, where they will not consider Reply-To/From, seems to suggest 
there is at least some validity in wanting to be able to express such 
a preference.

> Always.  You can try to convince MUA developers to provide a set of 
> response options that includes "where the author suggested" -- 
> maybe some will, maybe some won't (that also of course applies to a 
> new proposal, but encouraging support for a standard mechanism may 
> be easier to get across).

Well, my preferred MUA is Pine. There is 0 chance of persuading the 
Pine developers to add support for such things without something that 
is to make it to standards track. Nor would I approach other MUAs 
unless there were some proposal with a good chance of achieving 
consensus. :)

> There's no registered field with that name, though there's a draft 
> that contains a proposal.  First-come, first-served. (but I don't 
> think that's the answer)

Aha.

So is there any answer? Without wishing to have you prejudice your 
stance on the question(s), would you be able to suggest which avenues 
would the most fruitful to investigate?

> The first problem is that no MUA currently recognizes that, so none 
> will copy it to responses.

Agreed.

> A second issue is what happens when a respondent intentionally 
> wishes to send a private response?  And how is the MUA supposed to 
> know the respondent's intentions?

Good questions.

I would suggest a private response should not take any 
'Foo-Copies-To' header into consideration. Reply-To it probably 
should. (But then, didn't I read that From and only From should be 
taken as being the originator's mailbox? But that's not really my 
concern here.).

> First question is precisely what semantics are you trying to 
> indicate, and how is that different from "the mailbox(es) to which 
> the author of the message suggests that replies be sent"?  I.e. 
> first carefully formulate the problem that you're trying to solve, 
> as clearly and precisely as possible -- solving the wrong problem 
> is rarely a good approach.

Agreed.

Am I right in thinking this is where I draw up a draft, eg one 
focused solely on the issue of affecting mailbox selection in "list 
context" replies? (Or would it be better to first answer these 
questions here, in list email?).

Thanks for this paragraph of your reply, most wise.

> Another point to consider is that solutions to problems do not 
> exclusively lie in the realm of "let's invent a new header field to 
> ...".

I agree.

Initially I tried whining on the mailling list concerned "Please, use 
reply-all like $DEITY intended - i like to get direct mail for 
replies and threads i'm involved in". Replies ranged from:

- Why could you possibly want a direct copy?
- So we're supposed to remember every person's preference when we
   reply?
- Your fault for using Pine, use Mutt
- Set a Mail-Followup-To header.
- Oh stop whining paul
- Why dont you just filter based on In-Reply-To and/or References and
   create the extra copy yourself?

The more I whinge, the more beer gets spilled on me 'accidently' at 
LUG meets. (And recently things have taken a turn for the worse, the 
empty glasses now also get thrown at me). :)

regards,
-- 
Paul Jakma	[email protected]	[email protected]	Key ID: 64A2FF6A
Fortune:
What happens when you cut back the jungle?  It recedes.