Mailing to several people.

Mailing to several people.

[Andy Eskilsson]

I would be interesting to be able to select a bunch of articles,
pressing something like mail, and a new mail-buffer would open, with
the authors of the selected articles on the To: line..

BTW Is this an undocumented feature or .. selecting a bunch of
articles with '#' and pressing 'R' nice feature.. I dig it.. but I
think it is buggy, any comments??

	/andy

Re: Mailing to several people.

[Steven L. Baur]

>>>>> "Andy" == Andy Eskilsson <mpt95aes@sikt.hk-r.se> writes:

    Andy> I would be interesting to be able to select a bunch of
    Andy> articles, pressing something like mail, and a new
    Andy> mail-buffer would open, with the authors of the selected
    Andy> articles on the To: line..

This should be exactly what happens when you mark a number of
articles and press `r'.

    Andy> BTW Is this an undocumented feature or .. selecting a bunch
    Andy> of articles with '#' and pressing 'R' nice feature.. I dig
    Andy> it.. but I think it is buggy, any comments??

What Andy is describing is marking several articles, then doing a
Reply with Original on all of them.  The net effect is to have one
cited article containing the contents of all the marked articles.
This isn't a good feature, it's a great feature (assuming you trim all
the attributions appropriately).

It does look buggy, but it's probably only a line or two of fixes away
from actually working, (with the new To:/Cc: collection code being put
into 0.15).

Problems I observed:
* Super Cite prompts end up displayed in the minibuffer with the
buffer changed elsewhere.
* The To: line should reflect all of the articles being collected,
(instead of just the first marked article).  Assuming this can be
fixed, `r' should then work as Andy suggests.

Documenting it requires adding somewhere that R and r respect the
prefix/process conventions.

-- 
steve@miranova.com baur

Re: Mailing to several people.

[Lars Magne Ingebrigtsen]

steve@miranova.com (Steven L. Baur) writes:

> What Andy is describing is marking several articles, then doing a
> Reply with Original on all of them.  The net effect is to have one
> cited article containing the contents of all the marked articles.
> This isn't a good feature, it's a great feature (assuming you trim all
> the attributions appropriately).

Yes, it would be nice.  The reason it's not documented is that I feel
the current way of doing it just doesn't feel right.

After fixing a couple of bugs, this is what happens in September 0.17:

1) A mail reply buffer is set up the normal way based on the first
process marked article.

2) All subsequent articles are yanked with citation lines (hopefully)
now inserted where they should be.  (This didn't work in 0.16.)

That's yucky.  But it's not self-evident what should happen here.

I think all From, To and Cc lines from the articles should be
catenated as if replying to a single article that had quite long From,
To and Cc lines:

Article A:

From: larsi
Cc: ding
To: per

Article B:

From: gnus-bug
Cc: Steve

 =>

From: larsi, gnus-bug
Cc: ding, Steve
To: Per

And then Gnus uses the normal methods to find out what headers it
should put in the reply mail.

What about References, In-Reply-To and Subject, though?  

-- 
Home is where the cat is.

Re: Mailing to several people.

[Edward J. Sabol]

Excerpts from mail: (29-Nov-95) Re: Mailing to several people. by Lars Magne Ingebrigtsen
> I think all From, To and Cc lines from the articles should be catenated as
> if replying to a single article that had quite long From, To and Cc lines:
> 
> Article A:
> 
> From: larsi
> Cc: ding
> To: per
> 
> Article B:
> 
> From: gnus-bug
> Cc: Steve
> 
>  =>
> 
> From: larsi, gnus-bug
> Cc: ding, Steve
> To: Per

Can you really have multiple addresses on the "From:" line according the
RFC-<mumble, mumble>?

> And then Gnus uses the normal methods to find out what headers it
> should put in the reply mail.

Sounds good to me.

> What about References, In-Reply-To and Subject, though?

References: Concatenate the References header from all articles
Subject: Use the Subject from the first article marked, prepended with "Re: "
	if necessary, of course.
In-Reply-To: Concatenate them all using the format
	"<messageID@hostname> (e-mail@address)", assuming it's OK with the
	relevant RFCs, of course.

Later,
Ed

Re: Mailing to several people.

[Steven L. Baur]

>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

    Lars> After fixing a couple of bugs, this is what happens in
    Lars> September 0.17:

    Lars> 1) A mail reply buffer is set up the normal way based on the
    Lars> first process marked article.

    Lars> 2) All subsequent articles are yanked with citation lines
    Lars> (hopefully) now inserted where they should be.  (This didn't
    Lars> work in 0.16.)

    Lars> That's yucky.  But it's not self-evident what should happen
    Lars> here.

    Lars> I think all From, To and Cc lines from the articles should
    Lars> be catenated as if replying to a single article that had
    Lars> quite long From, To and Cc lines:

    Lars> Article A:

    Lars> From: larsi Cc: ding To: per

    Lars> Article B:

    Lars> From: gnus-bug Cc: Steve

    Lars>  =>

    Lars> From: larsi, gnus-bug Cc: ding, Steve To: Per

    Lars> And then Gnus uses the normal methods to find out what
    Lars> headers it should put in the reply mail.

    Lars> What about References, In-Reply-To and Subject, though?

All existing references should be catenated.  Add all the collected
message-ids to them.  Since this ties previously unconnected threads
together it will make for some interesting later threading decisions.

In-Reply-To, take the first article collected.

Subject, take the subject from the first article collected.  You could
perhaps add a configurable string to it like ``, etc.'' or ``-reply'' :-).

-- 
steve@miranova.com baur

Re: Mailing to several people.

[Steven L. Baur]

>>>>> "Edward" == Edward J Sabol <sabol@thuban.gsfc.nasa.gov> writes:

    Edward> Can you really have multiple addresses on the "From:" line
    Edward> according the RFC-<mumble, mumble>?

Yes.  This was discussed in the procmail mailing list after a recent
spamming that contained multiple addresses in the From: line.

The proponents were quite vigorous in their support.

-- 
steve@miranova.com baur

Re: Mailing to several people.

[Felix Lee]

> All existing references should be catenated.  Add all the collected

urr.  doesn't this cause problems if someone's threading assumes that
References: are in ancestor-to-descendent order?
--

Re: Mailing to several people.

[Lars Magne Ingebrigtsen]

"Steven L. Baur" <steve@miranova.com> writes:

> All existing references should be catenated.  Add all the collected
> message-ids to them.  Since this ties previously unconnected threads
> together it will make for some interesting later threading decisions.

I think this breaks all just about all RFCs ever written, which is a
plus.  :-)  

Actually, it doesn't break RFC822, which doesn't specify anything
about the References header except that it should "refer" to the mails
you reply to, so it would be ok.  However, I don't think any other
mail agent does anything like this, and there are practical
ramifications that I'd rather not think too much about.

Uhmn, I just don't know.  I'm leaning towards dropping the References
line completely and creating a Subject line without any "Re: ", with
all the different subjects separated by a " / ".  So one would start a
new thread by replying to several mails at once.

This won't break anything, but, on the other hand, it might not be
what one would want...

Either that, or just use the References from one of the messages --
the first, for instance.

-- 
Home is where the cat is.