Closed Bug 759018 Opened 12 years ago Closed 11 years ago

Reply-All behavior not right, doesn't include self in To: list

Categories

(Thunderbird :: Message Compose Window, defect)

12 Branch
x86
Windows XP
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED INVALID

People

(Reporter: frank, Unassigned)

Details

User Agent: Mozilla/5.0 (Windows NT 5.1; rv:12.0) Gecko/20100101 Firefox/12.0
Build ID: 20120420145725

Steps to reproduce:

Excerpted from "https://getsatisfaction.com/mozilla_messaging/topics/thunderbird_doesnt_include_me_in_reply_all":

I understand this has been an issue for years, some people want reply-all not to include themselves. However, anyone engaging in more than one conversation needs to look back at a thread of discussion. If reply-all doesn't include me, then the only way to get a full sense of the back and forth in the discussion is to have some cross-folder threading: matching E-mails in my inbox (or whatever box) with E-mails in my Sent folder <-- completely impractical and ridiculous, especially across multiple reading devices.

The "always include Bcc" doesn't work because others don't see me on the list of recipients.



Actual results:

Reply all button not visible (because there is only one recipient other than me). or Reply all doesn't include me.


Expected results:


In short, reply-all should:

(1) copy all the To: recipients from the old message and put them in the To: recipients in the new message; copy all the Cc: recipients to the CC: line in the new message

(2) include me on the To: line ***at the end of the list*** (so messages look like they are being sent to someone else, not just me)

For the people that don't want themselves included, then there might be a configuration option "reply-all should exclude me", which would be in Tools -> Options -> Composition.

Finally, these options need to be in the GUI not some random config editor option (which aren't documented), which change from release to release, and have no way of peeling out my customizations to copy-paste to another user or another instance of TB.

And the GUI needs to be consistent across messages. The buttons seem to change whether there is one person in the E-mail, some address that looks list-like, and so on. Maybe it's OK to grey things out, but don't change the buttons. For example, I ran down a wrong path trying to Customize by adding reply-all to my buttons, but it would work. Later I discovered that I couldn't Customize via *THAT* particulate E-mail because there was only one recipient.
There is also a fairly recent option to "auto cc"  like the old "auto bcc". That should work for you, right?
Thank you for the response, but that doesn't work.  For certain (human) communication protocols, the Cc: line means something lesser than the To: line.  Eudora used to get this right in reply-all: all the people on the To: line remained on the To: line, all the people on the Cc: line remained on the Cc: line.  So if I responded to a message where I was on the To: line, I'd be added to the To: line in reply-all ... and I'd be on the Cc: line for messages where I was on the Cc: line.

Although the Eudora behavior is the best fit for corporate communications, if there were an "auto to" feature, I could live with that.  But "auto cc" makes no sense because it DEMOTES my status in the communications.  In other words, if you are going to recognize that there are two equally functional ways of sending E-mail (To: vs. Cc:), then you need to respect that human communications treat those distinctions with importance and "auto cc" does not recognize the necessary human communications protocol.
But the user case you cite is keeping threading per folder intact. You'd still be in From also so I don't see how demoting would be an issue. You're still the sender!
(In reply to Frank Farance from comment #0)
> If reply-all doesn't include me, then the only way to get a full sense of
> the back and forth in the discussion is to have some cross-folder threading:
> matching E-mails in my inbox (or whatever box) with E-mails in my Sent
> folder <-- completely impractical and ridiculous, especially across multiple
> reading devices.

Does the "Open message in Conversation" not do what you want? Across multiple reading devices would only work correctly with IMAP with sent messages on server and then the Conversation view should also do what you want.
(In reply to Magnus Melin from comment #3)
> But the user case you cite is keeping threading per folder intact. You'd
> still be in From also so I don't see how demoting would be an issue. You're
> still the sender!

Right, but I don't get a copy of the mail because it is never sent back to me.  Furthermore, I'm not on the To: line and I need to be on the To: or the Cc: line.

There are E-mail filtering/sorting/searching programs that look at the To: and Cc: line and match/sort based upon that.  I don't want messages sent by *me* lost in those filtering programs because I wouldn't be on the To: or Cc: line.
(In reply to Frank Farance from comment #5)
> Right, but I don't get a copy of the mail because it is never sent back to me.

If you set up auto-cc, why wouldn't it get sent back to you?

> Furthermore, I'm not on the To: line and I need to be on the To: or the
> Cc: line.

Well, you would be in in the Cc line if you do set up auto-cc.
(In reply to Richard Marti [:paenglab] from comment #4)
> Does the "Open message in Conversation" not do what you want? Across
> multiple reading devices would only work correctly with IMAP with sent
> messages on server and then the Conversation view should also do what you
> want.

No this doesn't work.  For example, in my Blackberry I wouldn't see the complete set of messages, nor would I see the complete set of messages on my webmail (SquirrelMail), nor my ipad.  <-- This is because you are attempting to fix the problem in the E-mail client viewing, when you should be fixing the problem in the content (composed by the E-mail client).

And having a Sent message doesn't count because it doesn't have the SMTP header messages that can only come from actually sending the message.  For example, if I need to forward a message again because I sent it already but the person did not receive it, I would forward the message with the original SMTP headers ... only possibly if you're on the To:/Cc: line.

Your approach doesn't work when reading messages on multiple devices with multiple different behaviors.  Just being on the From: line does not guarantee that a reply-all message is sent to me.  In fact, a careful reading of RFC 2822 shows that the From: address might not actually receive the E-mail (e.g., when Sender: is used), so it is not necessarily a recipient of the original E-mail (which means this Conversations stuff doesn't work).   In RFC 2822, it says:

        When a message is a reply to another message, the mailboxes of the authors of the original message (the mailboxes in the "From:" field) or mailboxes specified in the "Reply-To:" field (if it exists) MAY appear in the "To:" field of the reply since these would normally be the primary recipients of the reply.  If a reply is sent to a message that has destination fields, it is often desirable to send a copy of the reply to all of the recipients of the message, in addition to the author.

In my experience sending E-mail in the past 35 years, it is the case that systems behave differently and not being on the To: line means you don't get a reply from someone's message.  As you can see from the above excerpt, the word MAY is used which means a conforming implementation is permitted to NOT follow that provision, i.e., a conforming implementation is permitted to NOT include the address in the From:/Reply-To: line as part of a reply message.

Furthermore, the second sentence provides clear guidance that is consistent with my request: "If a reply is sent to a message that has destination fields, it is often desirable to send a copy of the reply to all of the recipients of the message, in addition to the author", i.e., a copy is also sent "in addition to the author" and this whole idea of Conversations spanning Send and various other folders is completely misguided.

I note that RFC 2822 makes the semantic distinction between To: (primary recipient) and Cc: (other recipient) and Thunderbird should preserve this distinction (i.e., the Eudora behavior).

Thus, an option for sending back to self (without necessarily demoting to Cc) should be included in the GUI interface, maybe calling to "RFC 2822 Reply to Self To:/Cc:" behavior, i.e., the Eudora method.
(In reply to Magnus Melin from comment #6)
> (In reply to Frank Farance from comment #5)
> > Right, but I don't get a copy of the mail because it is never sent back to me.
> 
> If you set up auto-cc, why wouldn't it get sent back to you?
> 
> > Furthermore, I'm not on the To: line and I need to be on the To: or the
> > Cc: line.
> 
> Well, you would be in in the Cc line if you do set up auto-cc.

Yes, with auto-cc I'm on the Cc: line, but I'm demoted to Cc: status.  If you are not going to do it right (the Eudora method, as described in RFC 2822, see above), then at least provide a "auto-to" ... most people responding are primary recipients.
(In reply to Frank Farance from comment #8)
> Yes, with auto-cc I'm on the Cc: line, but I'm demoted to Cc: status.  If
> you are not going to do it right (the Eudora method, as described in RFC
> 2822, see above), then at least provide a "auto-to" ... most people
> responding are primary recipients.

And if you are going to provide "auto-to" then it should be the *last* recipient in the list, not the first recipient, so when I look at messages sorted by Recipient in the Sent folder (or my recipients sort their messages sorted by Receiver), I should not be the first recipient.  I know "auto-bcc" does this wrong.

The goal here is: to hit Reply-All (whether one recipient or many) and continue the discussion with all the recipients ... I SHOULD NOT have to go to the envelope portion of EVERY MESSAGE I SEND (as I do now) because TB has the wrong behavior on To:/Cc: (and Bcc: with auto-bcc).
(In reply to Frank Farance from comment #7)
> And having a Sent message doesn't count because it doesn't have the SMTP
> header messages that can only come from actually sending the message.  For
> example, if I need to forward a message again because I sent it already but
> the person did not receive it, I would forward the message with the original
> SMTP headers ... only possibly if you're on the To:/Cc: line.

This is simply not true. The headers that matter is there in the Sent folder. The only things added by the SMTP servers are Received and other fields that have no relevance for any user.
 
> Your approach doesn't work when reading messages on multiple devices with
> multiple different behaviors.  Just being on the From: line does not
> guarantee that a reply-all message is sent to me.  In fact, a careful
> reading of RFC 2822 shows that the From: address might not actually receive
> the E-mail (e.g., when Sender: is used), so it is not necessarily a

Not relevant, as *you* are controlling Sender (and in a typical setup it's not used), same thing with Reply-To. If you're in From you can be sure that recipients using reply or reply to all will include you too, claiming otherwise is just nonsense imo. (Well, for mailing lists there are other mechanisms...) 

> I note that RFC 2822 makes the semantic distinction between To: (primary
> recipient) and Cc: (other recipient) and Thunderbird should preserve this
> distinction (i.e., the Eudora behavior).

Well, if we're talking semantics Cc is much more semantically correct so than using To to yourself. To is the primary recipient/s, which you are not.
(In reply to Magnus Melin from comment #10)
> (In reply to Frank Farance from comment #7)
> > And having a Sent message doesn't count because it doesn't have the SMTP
> > header messages that can only come from actually sending the message.  For
> > example, if I need to forward a message again because I sent it already but
> > the person did not receive it, I would forward the message with the original
> > SMTP headers ... only possibly if you're on the To:/Cc: line.
> 
> This is simply not true. The headers that matter is there in the Sent
> folder. The only things added by the SMTP servers are Received and other
> fields that have no relevance for any user.

Basic E-mail 101: Received lines tell you who handled the mail, which can be tied to (say) sendmail logs and further validated.

Those Received are exactly the headers I'm looking for that validate that I've sent the message.  Yes, that it why it needs to be sent.

> > Your approach doesn't work when reading messages on multiple devices with
> > multiple different behaviors.  Just being on the From: line does not
> > guarantee that a reply-all message is sent to me.  In fact, a careful
> > reading of RFC 2822 shows that the From: address might not actually receive
> > the E-mail (e.g., when Sender: is used), so it is not necessarily a
> 
> Not relevant, as *you* are controlling Sender (and in a typical setup it's
> not used), same thing with Reply-To. If you're in From you can be sure that
> recipients using reply or reply to all will include you too, claiming
> otherwise is just nonsense imo. (Well, for mailing lists there are other
> mechanisms...) 

I'm not the Sender, I'm the From.  Someone (or something) else is the Sender.  Maybe you haven't worked with web servers and PHP backends that send E-mails.

> > I note that RFC 2822 makes the semantic distinction between To: (primary
> > recipient) and Cc: (other recipient) and Thunderbird should preserve this
> > distinction (i.e., the Eudora behavior).
> 
> Well, if we're talking semantics Cc is much more semantically correct so
> than using To to yourself. To is the primary recipient/s, which you are not.

You clearly haven't spend more than 2 seconds thinking about this because if you had spent 5 seconds thinking about your suggestion (or if you had experience in corporate E-mail), you'd see why your solution makes no sense.

Let's say I am having a discussion with 3 people (Moe, Larry, Curly).  But I want to include Moe's boss (a Cc: to Abbott) and Larry's boss (a Cc: to Costello).  The four of us (me, Moe, Larry, Curly) are the primary recipients of the messages (To:) and Abbott and Costello are the non-primary recipients (Cc:) of the messages.  Being the sender of the message does not preclude me of being one of the primary recipients of the discussion.

So if I send a message to Moe, I am now on the Cc: line.  If Moe E-mails Larry, Larry E-mails Curly, and Curly E-mails me, then EVERYONE IS ON THE CC line <-- WHICH MAKES NO SENSE (according to your algorithm).

Thus, you misunderstand Cc.  I am still considered one of the primary recipients of the discussion, whether I am sending (From) or receiving (To), which means I am put on the To: line.  If one of my employees wanted to CC their boss (me), it would mean in that discussion I was a non-primary recipient (and I'd be continually Cc'd, even on messages I send).  In other words, the Eudora behavior is the right behavior.

Maybe you can pass this bug onto someone else to work on ... from your comments above, it sounds like your not familiar with E-mail headers or multi-user/corporate E-mail discussions ... I'm not going to convince you if you don't have the requisite experience.  Thanks.
> So if I send a message to Moe, I am now on the Cc: line.  

You're also in From, so replies are To you.
Let's recap: you now understand why a message must actually be sent to the E-mail server and using the Sent folder is No Good.

> > So if I send a message to Moe, I am now on the Cc: line.  

> You're also in From, so replies are To you.

Not true.  Here it is with your suggestion "if we're talking semantics Cc is much more semantically correct so than using To to yourself. To is the primary recipient/s, which you are not":

#1:
From: frank
To: moe, larry, curly
Cc: abbott, costello, frank

#2:
From: moe
To: larry, curly
Cc: abbott, costello, frank, moe

#3:
From: larry
To: curly
Cc: abbott, costello, frank, moe, larry

#4:
From: curly
To:
Cc: abbott, costello, frank, moe, larry, curly


Now the way I had suggested would produce the following:

#1:
From: frank
To: moe, larry, curly, frank
Cc: abbott, costello

#2:
From: moe
To: larry, curly, frank, moe
Cc: abbott, costello

#3:
From: larry
To: curly, frank, moe, larry
Cc: abbott, costello

From: curly
To: frank, moe, larry, curly
Cc: abbott, costello


Thus, your way produces everyone on the Cc line (as I said), which is no good. And the Eudora way preserves all the primary and non-primary recipient roles (as I said).
I don't know what to tell you - the scenarios that you say simply doesn't happen (since *you* are in From!).

You would get

From: frank
To: moe, larry, curly
Cc: abbott, costello, frank

--- moe replies-all---

From: moe
To: frank
Cc: moe, larry, curly, abbott, costello


Unrelated: That To/CC isn't preserved for reply-all I do consider a bug (bug 250187, which i'm planning to fix), though it's common behavior in many mail clients and webmails.
Not true.  When Moe replies to me and sees that I'm on the Cc list (*regardless* that I'm the From:), he will put me on the Cc list.

The way you describe above in TB is equally wrong (as described in bug 250187).

Really, if you had experience in corporate E-mail communications and you had carefully read RFC 2822 (that provides distinction of To: and Cc:, which is how it is used in corporate E-mail), then we wouldn't be arguing about this.

However, you keep arguing that:

From: moe
To: frank
Cc: moe, larry, curly, abbott, costello

makes sense (which demotes larry and curly), which it does not, as described in bug 250187.

The fact the some mail clients and webmails get it wrong is caused by those people having a lack of understanding about corporate E-mail or not carefully reading RFC 2822 which preserves such a distinction.
(In reply to Frank Farance from comment #15)
> Not true.  When Moe replies to me and sees that I'm on the Cc list
> (*regardless* that I'm the From:), he will put me on the Cc list.

Manually people can do anything, but that would just be moe being stupid. I really doubt *any* software would do that automatically.

> makes sense (which demotes larry and curly), which it does not, as described
> in bug 250187.
> 
> The fact the some mail clients and webmails get it wrong is caused by those
> people having a lack of understanding about corporate E-mail or not
> carefully reading RFC 2822 which preserves such a distinction.

Like i said before i don't think it's good behavior, however it's not what this bug is about. The current behavior is explicitly a MAY in rfc 2822 though.
AFAIKT, this bug is invalid. I've explained why in the many, many comments above. You can do what you need using auto-cc.
Status: UNCONFIRMED → RESOLVED
Closed: 11 years ago
Resolution: --- → INVALID
AutoCc doesn't work because it can (and usually does) demote my status in the E-mail discussion.  AutoTo (if it existed) might work most of the time, but the real issue is including all the recipients at the SAME status (if I was CC'd before, then I'm Cc: now on the reply; if I was To'd before, then I'm To: now on the reply).

Since you're unfamiliar with corporate communications and, thus, don't have experience with this use-case, but you're familiar with the code: could you do me a favor and save me some time by pointing me to the code module where this logic occurs, and I'll fork my own TB source and build it myself?  Really, it is that important.
You need to log in before you can comment on or make changes to this bug.