\input texinfo                  @c -*-texinfo-*-

@setfilename ../../info/message.info
@settitle Message Manual
@include docstyle.texi
@synindex fn cp
@synindex vr cp
@synindex pg cp
@copying
This file documents Message, the Emacs message composition mode.

Copyright @copyright{} 1996--2020 Free Software Foundation, Inc.

@quotation
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with no
Invariant Sections, with the Front-Cover Texts being ``A GNU Manual'',
and with the Back-Cover Texts as in (a) below.  A copy of the license
is included in the section entitled ``GNU Free Documentation License''.

(a) The FSF's Back-Cover Text is: ``You have the freedom to copy and
modify this GNU manual.''
@end quotation
@end copying

@dircategory Emacs network features
@direntry
* Message: (message).           Mail and news composition mode that
                                  goes with Gnus.
@end direntry
@iftex
@finalout
@end iftex

@titlepage
@ifset WEBHACKDEVEL
@title Message Manual (DEVELOPMENT VERSION)
@end ifset
@ifclear WEBHACKDEVEL
@title Message Manual
@end ifclear

@author by Lars Magne Ingebrigtsen
@page

@vskip 0pt plus 1filll
@insertcopying
@end titlepage

@summarycontents
@contents

@node Top
@top Message

@ifnottex
@insertcopying
@end ifnottex

All message composition from Gnus (both mail and news) takes place in
Message mode buffers.

@menu
* Interface::         Setting up message buffers.
* Commands::          Commands you can execute in message mode buffers.
* Variables::         Customizing the message buffers.
* Appendices::        More technical things.
* GNU Free Documentation License:: The license for this documentation.
* Index::             Variable, function and concept index.
* Key Index::         List of Message mode keys.
@end menu

@c Adjust ../Makefile.in if you change the following lines:
Message is distributed with Gnus.  The Gnus distribution
@c
corresponding to this manual is Gnus v5.13


@node Interface
@chapter Interface

When a program (or a person) wants to respond to a message---reply,
follow up, forward, cancel---the program (or person) should just put
point in the buffer where the message is and call the required command.
@code{Message} will then pop up a new @code{message} mode buffer with
appropriate headers filled out, and the user can edit the message before
sending it.

@menu
* New Mail Message::     Editing a brand new mail message.
* New News Message::     Editing a brand new news message.
* Reply::                Replying via mail.
* Wide Reply::           Responding to all people via mail.
* Followup::             Following up via news.
* Canceling News::       Canceling a news article.
* Superseding::          Superseding a message.
* Forwarding::           Forwarding a message via news or mail.
* Resending::            Resending a mail message.
* Bouncing::             Bouncing a mail message.
* Mailing Lists::        Send mail to mailing lists.
@end menu

You can customize the Message Mode tool bar, see @kbd{M-x
customize-apropos @key{RET} message-tool-bar}.  This feature is only available
in Emacs.

@node New Mail Message
@section New Mail Message

@findex message-mail
The @code{message-mail} command pops up a new message buffer.

Two optional parameters are accepted: The first will be used as the
@code{To} header and the second as the @code{Subject} header.  If these
are @code{nil}, those two headers will be empty.


@node New News Message
@section New News Message

@findex message-news
The @code{message-news} command pops up a new message buffer.

This function accepts two optional parameters.  The first will be used
as the @code{Newsgroups} header and the second as the @code{Subject}
header.  If these are @code{nil}, those two headers will be empty.


@node Reply
@section Reply

@findex message-reply
The @code{message-reply} function pops up a message buffer that's a
reply to the message in the current buffer.

@vindex message-reply-to-function
Message uses the normal methods to determine where replies are to go
(@pxref{Responses}), but you can change the behavior to suit your needs
by fiddling with the @code{message-reply-to-function} variable.

If you want the replies to go to the @code{Sender} instead of the
@code{From}, you could do something like this:

@lisp
(setq message-reply-to-function
      (lambda ()
       (cond ((equal (mail-fetch-field "from") "somebody")
               (list (cons 'To (mail-fetch-field "sender"))))
             (t
              nil))))
@end lisp

This function will be called narrowed to the head of the article that is
being replied to.

As you can see, this function should return a list.  In this case, it
returns @code{((To . "Whom"))} if it has an opinion as to what the To
header should be.  If it does not, it should just return @code{nil}, and
the normal methods for determining the To header will be used.

Each list element should be a cons, where the @sc{car} should be the
name of a header (e.g., @code{CC}) and the @sc{cdr} should be the header
value (e.g., @samp{larsi@@ifi.uio.no}).  All these headers will be
inserted into the head of the outgoing mail.


@node Wide Reply
@section Wide Reply

@findex message-wide-reply
The @code{message-wide-reply} pops up a message buffer that's a wide
reply to the message in the current buffer.  A @dfn{wide reply} is a
reply that goes out to all people listed in the @code{To}, @code{From}
(or @code{Reply-To}) and @code{CC} headers.

@vindex message-wide-reply-to-function
Message uses the normal methods to determine where wide replies are to go,
but you can change the behavior to suit your needs by fiddling with the
@code{message-wide-reply-to-function}.  It is used in the same way as
@code{message-reply-to-function} (@pxref{Reply}).

@vindex message-dont-reply-to-names
Addresses that match the @code{message-dont-reply-to-names} regular
expression (or list of regular expressions or a predicate function)
will be removed from the @code{CC} header.  A value of @code{nil} means
to exclude only your email address.

@vindex message-prune-recipient-rules
@code{message-prune-recipient-rules} is used to prune the addresses
used when doing a wide reply.  It's meant to be used to remove
duplicate addresses and the like.  It's a list of lists, where the
first element is a regexp to match the address to trigger the rule,
and the second is a regexp that will be expanded based on the first,
to match addresses to be pruned.

It's complicated to explain, but it's easy to use.

For instance, if you get an email from @samp{foo@@example.org}, but
@samp{foo@@zot.example.org} is also in the @code{CC} list, then your
wide reply will go out to both these addresses, since they are unique.

To avoid this, do something like the following:

@lisp
(setq message-prune-recipient-rules
      '(("^\\([^@@]+\\)@@\\(.*\\)" "\\1@@.*[.]\\2")))
@end lisp

If, for instance, you want all wide replies that involve messages from
@samp{cvs@@example.org} to go to that address, and nowhere else (i.e.,
remove all other recipients if @samp{cvs@@example.org} is in the
recipient list:

@lisp
(setq message-prune-recipient-rules
      '(("cvs@@example.org" ".")))
@end lisp

@vindex message-wide-reply-confirm-recipients
If @code{message-wide-reply-confirm-recipients} is non-@code{nil} you
will be asked to confirm that you want to reply to multiple
recipients.  The default is @code{nil}.

@node Followup
@section Followup

@findex message-followup
The @code{message-followup} command pops up a message buffer that's a
followup to the message in the current buffer.

@vindex message-followup-to-function
Message uses the normal methods to determine where followups are to go,
but you can change the behavior to suit your needs by fiddling with the
@code{message-followup-to-function}.  It is used in the same way as
@code{message-reply-to-function} (@pxref{Reply}).

@vindex message-use-followup-to
The @code{message-use-followup-to} variable says what to do about
@code{Followup-To} headers.  If it is @code{use}, always use the value.
If it is @code{ask} (which is the default), ask whether to use the
value.  If it is @code{t}, use the value unless it is @samp{poster}.  If
it is @code{nil}, don't use the value.


@node Canceling News
@section Canceling News

@findex message-cancel-news
The @code{message-cancel-news} command cancels the article in the
current buffer.

@vindex message-cancel-message
The value of @code{message-cancel-message} is inserted in the body of
the cancel message.  The default is @samp{I am canceling my own
article.}.

@cindex Cancel Locks
@vindex message-insert-canlock
@cindex canlock
When Message posts news messages, it inserts @code{Cancel-Lock}
headers by default.  This is a cryptographic header that ensures that
only you can cancel your own messages, which is nice.  The downside
is that if you lose your @file{.emacs} file (which is where Gnus
stores the secret cancel lock password (which is generated
automatically the first time you use this feature)), you won't be
able to cancel your message.  If you want to manage a password yourself,
you can put something like the following in your @file{~/.gnus.el} file:

@lisp
(setq canlock-password "geheimnis"
      canlock-password-for-verify canlock-password)
@end lisp

Whether to insert the header or not is controlled by the
@code{message-insert-canlock} variable.

Not many news servers respect the @code{Cancel-Lock} header yet, but
this is expected to change in the future.


@node Superseding
@section Superseding

@findex message-supersede
The @code{message-supersede} command pops up a message buffer that will
supersede the message in the current buffer.

@vindex message-ignored-supersedes-headers
Headers matching the @code{message-ignored-supersedes-headers} are
removed before popping up the new message buffer.  The default is@*
@samp{^Path:\\|^Date\\|^NNTP-Posting-Host:\\|^Xref:\\|^Lines:\\|@*
^Received:\\|^X-From-Line:\\|^X-Trace:\\|^X-Complaints-To:\\|@*
Return-Path:\\|^Supersedes:\\|^NNTP-Posting-Date:\\|^X-Trace:\\|@*
^X-Complaints-To:\\|^Cancel-Lock:\\|^Cancel-Key:\\|^X-Hashcash:\\|@*
^X-Payment:\\|^Approved:}.



@node Forwarding
@section Forwarding

@findex message-forward
The @code{message-forward} command pops up a message buffer to forward
the message in the current buffer.  If given a prefix, forward using
news.

@table @code
@item message-forward-ignored-headers
@vindex message-forward-ignored-headers
In non-@code{nil}, all headers that match this regexp will be deleted
when forwarding a message.

@item message-forward-included-headers
@vindex message-forward-included-headers
In non-@code{nil}, only headers that match this regexp will be kept
when forwarding a message.  This can also be a list of regexps.

@item message-make-forward-subject-function
@vindex message-make-forward-subject-function
A list of functions that are called to generate a subject header for
forwarded messages.  The subject generated by the previous function is
passed into each successive function.

The provided functions are:

@table @code
@item message-forward-subject-author-subject
@findex message-forward-subject-author-subject
Source of article (author or newsgroup), in brackets followed by the
subject.

@item message-forward-subject-fwd
Subject of article with @samp{Fwd:} prepended to it.
@end table

@item message-wash-forwarded-subjects
@vindex message-wash-forwarded-subjects
If this variable is @code{t}, the subjects of forwarded messages have
the evidence of previous forwards (such as @samp{Fwd:}, @samp{Re:},
@samp{(fwd)}) removed before the new subject is
constructed.  The default value is @code{nil}.

@item message-forward-as-mime
@vindex message-forward-as-mime
If this variable is @code{t}, forwarded messages are included as
inline @acronym{MIME} RFC822 parts.  If it's @code{nil} (the default),
forwarded messages will just be copied inline to the new message, like
previous, non @acronym{MIME}-savvy versions of Gnus would do.

@item message-forward-before-signature
@vindex message-forward-before-signature
If non-@code{nil}, put forwarded message before signature, else after.

@end table


@node Resending
@section Resending

@findex message-resend
The @code{message-resend} command will prompt the user for an address
and resend the message in the current buffer to that address.

@vindex message-ignored-resent-headers
Headers that match the @code{message-ignored-resent-headers} regexp will
be removed before sending the message.


@node Bouncing
@section Bouncing

@findex message-bounce
The @code{message-bounce} command will, if the current buffer contains a
bounced mail message, pop up a message buffer stripped of the bounce
information.  A @dfn{bounced message} is typically a mail you've sent
out that has been returned by some @code{mailer-daemon} as
undeliverable.

@vindex message-ignored-bounced-headers
Headers that match the @code{message-ignored-bounced-headers} regexp
will be removed before popping up the buffer.  The default is
@samp{^\\(Received\\|Return-Path\\|Delivered-To\\):}.


@node Mailing Lists
@section Mailing Lists

@cindex Mail-Followup-To
Sometimes while posting to mailing lists, the poster needs to direct
followups to the post to specific places.  The Mail-Followup-To (MFT)
was created to enable just this.  Three example scenarios where this is
useful:

@itemize @bullet
@item
A mailing list poster can use MFT to express that responses should be
sent to just the list, and not the poster as well.  This will happen
if the poster is already subscribed to the list.

@item
A mailing list poster can use MFT to express that responses should be
sent to the list and the poster as well.  This will happen if the poster
is not subscribed to the list.

@item
If a message is posted to several mailing lists, MFT may also be used
to direct the following discussion to one list only, because
discussions that are spread over several lists tend to be fragmented
and very difficult to follow.

@end itemize

Gnus honors the MFT header in other's messages (i.e., while following
up to someone else's post) and also provides support for generating
sensible MFT headers for outgoing messages as well.

@c @menu
@c * Honoring an MFT post::        What to do when one already exists
@c * Composing with a MFT header:: Creating one from scratch.
@c @end menu

@c @node Composing with a MFT header
@subsection  Composing a correct MFT header automagically

The first step in getting Gnus to automagically generate a MFT header
in posts you make is to give Gnus a list of the mailing lists
addresses you are subscribed to.  You can do this in more than one
way.  The following variables would come in handy.

@table @code

@vindex message-subscribed-addresses
@item message-subscribed-addresses
This should be a list of addresses the user is subscribed to.  Its
default value is @code{nil}.  Example:
@lisp
(setq message-subscribed-addresses
      '("ding@@gnus.org" "bing@@noose.org"))
@end lisp

@vindex message-subscribed-regexps
@item message-subscribed-regexps
This should be a list of regexps denoting the addresses of mailing
lists subscribed to.  Default value is @code{nil}.  Example: If you
want to achieve the same result as above:
@lisp
(setq message-subscribed-regexps
      '("\\(ding@@gnus\\)\\|\\(bing@@noose\\)\\.org")
@end lisp

@vindex message-subscribed-address-functions
@item message-subscribed-address-functions
This can be a list of functions to be called (one at a time!!) to
determine the value of MFT headers.  It is advisable that these
functions not take any arguments.  Default value is @code{nil}.

There is a pre-defined function in Gnus that is a good candidate for
this variable.  @code{gnus-find-subscribed-addresses} is a function
that returns a list of addresses corresponding to the groups that have
the @code{subscribed} (@pxref{Group Parameters, ,Group Parameters,
gnus, The Gnus Manual}) group parameter set to a non-@code{nil} value.
This is how you would do it.

@lisp
(setq message-subscribed-address-functions
      '(gnus-find-subscribed-addresses))
@end lisp

@vindex message-subscribed-address-file
@item message-subscribed-address-file
You might be one organized human freak and have a list of addresses of
all subscribed mailing lists in a separate file!  Then you can just
set this variable to the name of the file and life would be good.

@end table

You can use one or more of the above variables.  All their values are
``added'' in some way that works :-)

Now you are all set.  Just start composing a message as you normally do.
And just send it; as always.  Just before the message is sent out, Gnus'
MFT generation thingy kicks in and checks if the message already has a
MFT field.  If there is one, it is left alone.  (Except if it's empty;
in that case, the field is removed and is not replaced with an
automatically generated one.  This lets you disable MFT generation on a
per-message basis.)  If there is none, then the list of recipient
addresses (in the To: and CC: headers) is checked to see if one of them
is a list address you are subscribed to.  If none of them is a list
address, then no MFT is generated; otherwise, a MFT is added to the
other headers and set to the value of all addresses in To: and CC:

@kindex C-c C-f C-a
@findex message-generate-unsubscribed-mail-followup-to
@kindex C-c C-f C-m
@findex message-goto-mail-followup-to
Hm.  ``So'', you ask, ``what if I send an email to a list I am not
subscribed to?  I want my MFT to say that I want an extra copy.''  (This
is supposed to be interpreted by others the same way as if there were no
MFT, but you can use an explicit MFT to override someone else's
to-address group parameter.)  The function
@code{message-generate-unsubscribed-mail-followup-to} might come in
handy.  It is bound to @kbd{C-c C-f C-a} by default.  In any case, you
can insert a MFT of your own choice; @kbd{C-c C-f C-m}
(@code{message-goto-mail-followup-to}) will help you get started.

@c @node Honoring an MFT post
@subsection Honoring an MFT post

@vindex message-use-mail-followup-to
When you followup to a post on a mailing list, and the post has a MFT
header, Gnus' action will depend on the value of the variable
@code{message-use-mail-followup-to}.  This variable can be one of:

@table @code
@item use
 Always honor MFTs.  The To: and CC: headers in your followup will be
 derived from the MFT header of the original post.  This is the default.

@item nil
 Always dishonor MFTs (just ignore the darned thing)

@item ask
Gnus will prompt you for an action.

@end table

It is considered good netiquette to honor MFT, as it is assumed the
fellow who posted a message knows where the followups need to go
better than you do.

@node Commands
@chapter Commands

@menu
* Buffer Entry::        Commands after entering a Message buffer.
* Header Commands::     Commands for moving headers or changing headers.
* Movement::            Moving around in message buffers.
* Insertion::           Inserting things into message buffers.
* MIME::                @acronym{MIME} considerations.
* IDNA::                Non-@acronym{ASCII} domain name considerations.
* Security::            Signing and encrypting messages.
* Various Commands::    Various things.
* Sending::             Actually sending the message.
* Mail Aliases::        How to use mail aliases.
* Spelling::            Having Emacs check your spelling.
@end menu


@node Buffer Entry
@section Buffer Entry
@cindex undo
@kindex C-_

You most often end up in a Message buffer when responding to some other
message of some sort.  Message does lots of handling of quoted text, and
may remove signatures, reformat the text, or the like---depending on
which used settings you're using.  Message usually gets things right,
but sometimes it stumbles.  To help the user unwind these stumblings,
Message sets the undo boundary before each major automatic action it
takes.  If you press the undo key (usually located at @kbd{C-_}) a few
times, you will get back the un-edited message you're responding to.


@node Header Commands
@section Header Commands

@subsection Commands for moving to headers

These following commands move to the header in question.  If it doesn't
exist, it will be inserted.

@table @kbd

@item C-c ?
@kindex C-c ?
@findex describe-mode
Describe the message mode.

@item C-c C-f C-t
@kindex C-c C-f C-t
@findex message-goto-to
Go to the @code{To} header (@code{message-goto-to}).

@item C-c C-f C-o
@kindex C-c C-f C-o
@findex message-goto-from
Go to the @code{From} header (@code{message-goto-from}).  (The ``o''
in the key binding is for Originator.)

@item C-c C-f C-b
@kindex C-c C-f C-b
@findex message-goto-bcc
Go to the @code{BCC} header (@code{message-goto-bcc}).

@item C-c C-f C-w
@kindex C-c C-f C-w
@findex message-goto-fcc
Go to the @code{FCC} header (@code{message-goto-fcc}).

@item C-c C-f C-c
@kindex C-c C-f C-c
@findex message-goto-cc
Go to the @code{CC} header (@code{message-goto-cc}).

@item C-c C-f C-s
@kindex C-c C-f C-s
@findex message-goto-subject
Go to the @code{Subject} header (@code{message-goto-subject}).

@item C-c C-f C-r
@kindex C-c C-f C-r
@findex message-goto-reply-to
Go to the @code{Reply-To} header (@code{message-goto-reply-to}).

@item C-c C-f C-n
@kindex C-c C-f C-n
@findex message-goto-newsgroups
Go to the @code{Newsgroups} header (@code{message-goto-newsgroups}).

@item C-c C-f C-d
@kindex C-c C-f C-d
@findex message-goto-distribution
Go to the @code{Distribution} header (@code{message-goto-distribution}).

@item C-c C-f C-f
@kindex C-c C-f C-f
@findex message-goto-followup-to
Go to the @code{Followup-To} header (@code{message-goto-followup-to}).

@item C-c C-f C-k
@kindex C-c C-f C-k
@findex message-goto-keywords
Go to the @code{Keywords} header (@code{message-goto-keywords}).

@item C-c C-f C-u
@kindex C-c C-f C-u
@findex message-goto-summary
Go to the @code{Summary} header (@code{message-goto-summary}).

@item C-c C-f C-i
@kindex C-c C-f C-i
@findex message-insert-or-toggle-importance
This inserts the @samp{Importance:} header with a value of
@samp{high}.  This header is used to signal the importance of the
message to the receiver.  If the header is already present in the
buffer, it cycles between the three valid values according to RFC
1376: @samp{low}, @samp{normal} and @samp{high}.

@item C-c C-f C-a
@kindex C-c C-f C-a
@findex message-generate-unsubscribed-mail-followup-to
Insert a reasonable @samp{Mail-Followup-To:} header
(@pxref{Mailing Lists}) in a post to an
unsubscribed list.  When making original posts to a mailing list you are
not subscribed to, you have to type in a @samp{Mail-Followup-To:} header
by hand.  The contents, usually, are the addresses of the list and your
own address.  This function inserts such a header automatically.  It
fetches the contents of the @samp{To:} header in the current mail
buffer, and appends the current @code{user-mail-address}.

If the optional argument @code{include-cc} is non-@code{nil}, the
addresses in the @samp{CC:} header are also put into the
@samp{Mail-Followup-To:} header.

@end table

@subsection  Commands to change headers

@table @kbd

@item C-c C-o
@kindex C-c C-o
@findex message-sort-headers
@vindex message-header-format-alist
Sort headers according to @code{message-header-format-alist}
(@code{message-sort-headers}).

@item C-c C-t
@kindex C-c C-t
@findex message-insert-to
Insert a @code{To} header that contains the @code{Reply-To} or
@code{From} header of the message you're following up
(@code{message-insert-to}).

@item C-c C-n
@kindex C-c C-n
@findex message-insert-newsgroups
Insert a @code{Newsgroups} header that reflects the @code{Followup-To}
or @code{Newsgroups} header of the article you're replying to
(@code{message-insert-newsgroups}).

@item C-c C-l
@kindex C-c C-l
@findex message-to-list-only
Send a message to the list only.  Remove all addresses but the list
address from @code{To:} and @code{CC:} headers.

@item C-c M-n
@kindex C-c M-n
@findex message-insert-disposition-notification-to
Insert a request for a disposition
notification.  (@code{message-insert-disposition-notification-to}).
This means that if the recipient supports RFC 2298 she might send you a
notification that she received the message.

@item M-x message-insert-importance-high
@findex message-insert-importance-high
@cindex Importance
Insert an @samp{Importance} header with a value of @samp{high},
deleting headers if necessary.

@item M-x message-insert-importance-low
@findex message-insert-importance-low
@cindex Importance
Insert an @samp{Importance} header with a value of @samp{low}, deleting
headers if necessary.

@item C-c C-f s
@kindex C-c C-f s
@findex message-change-subject
@cindex Subject
Change the current @samp{Subject} header.  Ask for new @samp{Subject}
header and append @samp{(was: <Old Subject>)}.  The old subject can be
stripped on replying, see @code{message-subject-trailing-was-query}
(@pxref{Message Headers}).

@item C-c C-f x
@kindex C-c C-f x
@findex message-cross-post-followup-to
@vindex message-cross-post-default
@vindex message-cross-post-note-function
@cindex X-Post
@cindex cross-post
Set up the @samp{FollowUp-To} header with a target newsgroup for a
cross-post, add that target newsgroup to the @samp{Newsgroups} header if
it is not a member of @samp{Newsgroups}, and insert a note in the body.
If @code{message-cross-post-default} is @code{nil} or if this command is
called with a prefix-argument, only the @samp{FollowUp-To} header will
be set but the target newsgroup will not be added to the
@samp{Newsgroups} header.  The function to insert a note is controlled
by the @code{message-cross-post-note-function} variable.

@item C-c C-f t
@kindex C-c C-f t
@findex message-reduce-to-to-cc
Replace contents of @samp{To} header with contents of @samp{CC}
header (or the @samp{BCC} header, if there is no @samp{CC} header).

@item C-c C-f w
@kindex C-c C-f w
@findex message-insert-wide-reply
Insert @samp{To} and @samp{CC} headers as if you were doing a wide
reply even if the message was not made for a wide reply first.

@item C-c C-f a
@kindex C-c C-f a
@findex message-add-archive-header
@vindex message-archive-header
@vindex message-archive-note
@cindex X-No-Archive
Insert @samp{X-No-Archive: Yes} in the header and a note in the body.
The header and the note can be customized using
@code{message-archive-header} and @code{message-archive-note}.  When
called with a prefix argument, ask for a text to insert.  If you don't
want the note in the body, set @code{message-archive-note} to
@code{nil}.

@end table


@node Movement
@section Movement

@table @kbd
@item C-c C-b
@kindex C-c C-b
@findex message-goto-body
Move to the beginning of the body of the message
(@code{message-goto-body}).

@item C-c C-i
@kindex C-c C-i
@findex message-goto-signature
Move to the signature of the message (@code{message-goto-signature}).

@item C-a
@kindex C-a
@findex message-beginning-of-line
@vindex message-beginning-of-line
If at beginning of header value, go to beginning of line, else go to
beginning of header value.  (The header value comes after the header
name and the colon.)  This behavior can be disabled by toggling
the variable @code{message-beginning-of-line}.

@end table


@node Insertion
@section Insertion

@table @kbd

@item C-c C-y
@kindex C-c C-y
@findex message-yank-original
Yank the message that's being replied to into the message buffer
(@code{message-yank-original}).

@item C-c C-M-y
@kindex C-c C-M-y
@findex message-yank-buffer
Prompt for a buffer name and yank the contents of that buffer into the
message buffer (@code{message-yank-buffer}).

@item C-c C-q
@kindex C-c C-q
@findex message-fill-yanked-message
Fill the yanked message (@code{message-fill-yanked-message}).  Warning:
Can severely mess up the yanked text if its quoting conventions are
strange.  You'll quickly get a feel for when it's safe, though.  Anyway,
just remember that @kbd{C-x u} (@code{undo}) is available and you'll be
all right.

@item C-c C-w
@kindex C-c C-w
@findex message-insert-signature
Insert a signature at the end of the buffer
(@code{message-insert-signature}).

@item C-c M-h
@kindex C-c M-h
@findex message-insert-headers
Insert the message headers (@code{message-insert-headers}).

@item C-c M-m
@kindex C-c M-m
@findex message-mark-inserted-region
Mark some region in the current article with enclosing tags.  See
@code{message-mark-insert-begin} and @code{message-mark-insert-end}.
When called with a prefix argument, use slrn style verbatim marks
(@samp{#v+} and @samp{#v-}).

@item C-c M-f
@kindex C-c M-f
@findex message-mark-insert-file
Insert a file in the current article with enclosing tags.
See @code{message-mark-insert-begin} and @code{message-mark-insert-end}.
When called with a prefix argument, use slrn style verbatim marks
(@samp{#v+} and @samp{#v-}).

@end table


@node MIME
@section MIME
@cindex MML
@cindex MIME
@cindex multipart
@cindex attachment

Message is a @acronym{MIME}-compliant posting agent.  The user generally
doesn't have to do anything to make the @acronym{MIME} happen---Message will
automatically add the @code{Content-Type} and
@code{Content-Transfer-Encoding} headers.

@findex mml-attach-file
@kindex C-c C-a
The most typical thing users want to use the multipart things in
@acronym{MIME} for is to add ``attachments'' to mail they send out.
This can be done with the @kbd{C-c C-a} command (@kbd{M-x mml-attach-file}),
which will prompt for a file name and a @acronym{MIME} type.

@vindex mml-dnd-protocol-alist
@vindex mml-dnd-attach-options
If your Emacs supports drag and drop, you can also drop the file in the
Message buffer.  The variable @code{mml-dnd-protocol-alist} specifies
what kind of action is done when you drop a file into the Message
buffer.  The variable @code{mml-dnd-attach-options} controls which
@acronym{MIME} options you want to specify when dropping a file.  If it
is a list, valid members are @code{type}, @code{description} and
@code{disposition}.  @code{disposition} implies @code{type}.  If it is
@code{nil}, don't ask for options.  If it is @code{t}, ask the user
whether or not to specify options.

You can also create arbitrarily complex multiparts using the @acronym{MML}
language (@pxref{Composing, , Composing, emacs-mime, The Emacs MIME
Manual}).

@node IDNA
@section IDNA
@cindex IDNA
@cindex internationalized domain names
@cindex non-ascii domain names

@acronym{IDNA} is a standard way to encode non-@acronym{ASCII} domain
names into a readable @acronym{ASCII} string.  The details can be
found in RFC 3490.

Message is a @acronym{IDNA}-compliant posting agent.  The user
generally doesn't have to do anything to make the @acronym{IDNA}
happen---Message will encode non-@acronym{ASCII} domain names in @code{From},
@code{To}, and @code{CC} headers automatically.

Until @acronym{IDNA} becomes more well known, Message queries you
whether @acronym{IDNA} encoding of the domain name really should
occur.  Some users might not be aware that domain names can contain
non-@acronym{ASCII} now, so this gives them a safety net if they accidentally
typed a non-@acronym{ASCII} domain name.

@vindex message-use-idna
The @code{message-use-idna} variable control whether @acronym{IDNA} is
used.  If the variable is @code{nil} no @acronym{IDNA} encoding will
ever happen, if it is set to the symbol @code{ask} the user will be
queried, and if set to @code{t} (which is the default if @acronym{IDNA}
is fully available) @acronym{IDNA} encoding happens automatically.

@findex message-idna-to-ascii-rhs
If you want to experiment with the @acronym{IDNA} encoding, you can
invoke @kbd{M-x message-idna-to-ascii-rhs @key{RET}} in the message buffer
to have the non-@acronym{ASCII} domain names encoded while you edit
the message.

Note that you must have @uref{https://www.gnu.org/software/libidn/, GNU
Libidn} installed in order to use this functionality.

@node Security
@section Security
@cindex Security
@cindex S/MIME
@cindex PGP
@cindex PGP/MIME
@cindex sign
@cindex encrypt
@cindex secure

By default, e-mails are transmitted without any protection around the
Internet, which implies that they can be read and changed by lots of
different parties.  In particular, they are analyzed under bulk
surveillance, which violates basic human rights.  To defend those
rights, digital self-defense is necessary (in addition to legal
changes), and encryption and digital signatures are powerful
techniques for self-defense.  In essence, encryption ensures that
only the intended recipient will be able to read a message, while
digital signatures make sure that modifications to messages can be
detected by the recipient.

Nowadays, there are two major incompatible e-mail encryption
standards, namely @acronym{OpenPGP} and @acronym{S/MIME}.  Both of
these standards are implemented by the @uref{https://www.gnupg.org/,
GNU Privacy Guard (GnuPG)}, which needs to be installed as external
software in addition to GNU Emacs.  Before you can start to encrypt,
decrypt, and sign messages, you need to create a so-called key-pair,
which consists of a private key and a public key.  Your @emph{public} key
(also known as @emph{certificate}, in particular with @acronym{S/MIME}), is
used by others (a) to encrypt messages intended for you and (b) to verify
digital signatures created by you.  In contrast, you use your @emph{private}
key (a) to decrypt messages and (b) to sign messages.  (You may want to
think of your public key as an open safe that you offer to others such
that they can deposit messages and lock the door, while your private
key corresponds to the opening combination for the safe.)

Thus, you need to perform the following steps for e-mail encryption,
typically outside Emacs.  See, for example, the
@uref{https://www.gnupg.org/gph/en/manual.html, The GNU Privacy
Handbook} for details covering the standard @acronym{OpenPGP} with
@acronym{GnuPG}.
@enumerate
@item
Install GnuPG.
@item
Create a key-pair for your own e-mail address.
@item
Distribute your public key, e.g., via upload to key servers.
@item
Import the public keys for the recipients to which you want to send
encrypted e-mails.
@end enumerate

Whether to use the standard @acronym{OpenPGP} or @acronym{S/MIME} is
beyond the scope of this documentation.  Actually, you can use one
standard for one set of recipients and the other standard for
different recipients (depending their preferences or capabilities).

In case you are not familiar with all those acronyms: The standard
@acronym{OpenPGP} is also called @acronym{PGP} (Pretty Good Privacy).
The command line tools offered by @acronym{GnuPG} for
@acronym{OpenPGP} are called @command{gpg} and @command{gpg2}, while
the one for @acronym{S/MIME} is called @command{gpgsm}.  An
alternative, but discouraged, tool for @acronym{S/MIME} is
@command{openssl}.  To make matters worse, e-mail messages can be
formed in two different ways with @acronym{OpenPGP}, namely
@acronym{PGP} (RFC 1991/4880) and @acronym{PGP/MIME} (RFC 2015/3156).

The good news, however, is the following: In GNU Emacs, Message
supports all those variants, comes with reasonable defaults that can
be customized according to your needs, and invokes the proper command
line tools behind the scenes for encryption, decryption, as well as
creation and verification of digital signatures.

Message uses the @acronym{MML} language for the creation of signed
and/or encrypted messages as explained in the following.


@menu
* Signing and encryption::      Signing and encrypting commands.
* Using S/MIME::                Using S/MIME
* Using OpenPGP::               Using OpenPGP
* Passphrase caching::          How to cache passphrases
* PGP Compatibility::           Compatibility with older implementations
* Encrypt-to-self::             Reading your own encrypted messages
* BCC Warning::                 Do not use encryption with BCC headers
@end menu

@node Signing and encryption
@subsection Signing and encrypting commands

Instructing @acronym{MML} to perform security operations on a
@acronym{MIME} part is done using the @kbd{C-c C-m s} key map for
signing and the @kbd{C-c C-m c} key map for encryption, as follows.
@table @kbd

@item C-c C-m s s
@kindex C-c C-m s s
@findex mml-secure-message-sign-smime

Digitally sign current message using @acronym{S/MIME}.

@item C-c C-m s o
@kindex C-c C-m s o
@findex mml-secure-message-sign-pgp

Digitally sign current message using @acronym{PGP}.

@item C-c C-m s p
@kindex C-c C-m s p
@findex mml-secure-message-sign-pgpmime

Digitally sign current message using @acronym{PGP/MIME}.

@item C-c C-m c s
@kindex C-c C-m c s
@findex mml-secure-message-encrypt-smime

Digitally encrypt current message using @acronym{S/MIME}.

@item C-c C-m c o
@kindex C-c C-m c o
@findex mml-secure-message-encrypt-pgp

Digitally encrypt current message using @acronym{PGP}.

@item C-c C-m c p
@kindex C-c C-m c p
@findex mml-secure-message-encrypt-pgpmime

Digitally encrypt current message using @acronym{PGP/MIME}.

@item C-c C-m C-n
@kindex C-c C-m C-n
@findex mml-unsecure-message
Remove security related @acronym{MML} tags from message.

@end table

These commands do not immediately sign or encrypt the message, they
merely insert the proper @acronym{MML} secure tag to instruct the
@acronym{MML} engine to perform that operation when the message is
actually sent.  They may perform other operations too, such as locating
and retrieving a @acronym{S/MIME} certificate of the person you wish to
send encrypted mail to.  When the mml parsing engine converts your
@acronym{MML} into a properly encoded @acronym{MIME} message, the secure
tag will be replaced with either a part or a multipart tag.  If your
message contains other mml parts, a multipart tag will be used; if no
other parts are present in your message a single part tag will be used.
This way, message mode will do the Right Thing (TM) with
signed/encrypted multipart messages.

Since signing and especially encryption often is used when sensitive
information is sent, you may want to have some way to ensure that your
mail is actually signed or encrypted.  After invoking the above
sign/encrypt commands, it is possible to preview the raw article by
using @kbd{C-u C-c @key{RET} P} (@code{mml-preview}).  Then you can
verify that your long rant about what your ex-significant other or
whomever actually did with that funny looking person at that strange
party the other night, actually will be sent encrypted.

@emph{Note!}  Neither @acronym{PGP/MIME} nor @acronym{S/MIME} encrypt/signs
RFC822 headers.  They only operate on the @acronym{MIME} object.  Keep this
in mind before sending mail with a sensitive Subject line.

By default, when encrypting a message, Gnus will use the
``signencrypt'' mode, which means the message is both signed and
encrypted.  If you would like to disable this for a particular
message, give the @code{mml-secure-message-encrypt-*} command a prefix
argument, e.g., @kbd{C-u C-c C-m c p}.

Actually using the security commands above is not very difficult.  At
least not compared with making sure all involved programs talk with each
other properly.  Thus, we now describe what external libraries or
programs are required to make things work, and some small general hints.

@node Using S/MIME
@subsection Using S/MIME

@acronym{S/MIME} requires an external implementation, such as
@uref{https://www.gnupg.org/, GNU Privacy Guard} or
@uref{https://www.openssl.org/, OpenSSL}.  The default Emacs interface
to the S/MIME implementation is EasyPG (@pxref{Top,,EasyPG Assistant
User's Manual, epa, EasyPG Assistant User's Manual}), which has been
included in Emacs since version 23 and which relies on the command
line tool @command{gpgsm} provided by @acronym{GnuPG}.  That tool
implements certificate management, including certificate revocation
and expiry, while such tasks need to be performed manually, if OpenSSL
is used.

The choice between EasyPG and OpenSSL is controlled by the variable
@code{mml-smime-use}, which needs to be set to the value @code{epg}
for EasyPG@.  Depending on your version of Emacs that value may be the
default; if not, you can either customize that variable or place the
following line in your @file{.emacs} file (that line needs to be
placed above other code related to message/gnus/encryption):

@lisp
(require 'epg)
@end lisp

Moreover, you may want to customize the variables
@code{mml-default-encrypt-method} and
@code{mml-default-sign-method} to the string @code{"smime"}.

That's all if you want to use S/MIME with EasyPG, and that's the
recommended way of using S/MIME with Message.

If you think about using OpenSSL instead of EasyPG, please read the
BUGS section in the manual for the @command{smime} command coming with
OpenSSL first.  If you still want to use OpenSSL, the following
applies.

@emph{Note!}  The remainder of this section assumes you have a basic
familiarity with modern cryptography, @acronym{S/MIME}, various PKCS
standards, OpenSSL and so on.

The @acronym{S/MIME} support in Message (and @acronym{MML}) can use
OpenSSL@.  OpenSSL performs the actual @acronym{S/MIME} sign/encrypt
operations.  OpenSSL can be found at @uref{https://www.openssl.org/}.
OpenSSL 0.9.6 and later should work.  Version 0.9.5a cannot extract mail
addresses from certificates, and it insert a spurious CR character into
@acronym{MIME} separators so you may wish to avoid it if you would like
to avoid being regarded as someone who send strange mail.  (Although by
sending @acronym{S/MIME} messages you've probably already lost that
contest.)

To be able to send encrypted mail, a personal certificate is not
required.  Message (@acronym{MML}) need a certificate for the person to whom you
wish to communicate with though.  You're asked for this when you type
@kbd{C-c C-m c s}.  Currently there are two ways to retrieve this
certificate, from a local file or from DNS@.  If you chose a local
file, it need to contain a X.509 certificate in @acronym{PEM} format.
If you chose DNS, you're asked for the domain name where the
certificate is stored, the default is a good guess.  To my belief,
Message (@acronym{MML}) is the first mail agent in the world to support
retrieving @acronym{S/MIME} certificates from DNS, so you're not
likely to find very many certificates out there.  At least there
should be one, stored at the domain @code{simon.josefsson.org}.  LDAP
is a more popular method of distributing certificates, support for it
is planned.  (Meanwhile, you can use @code{ldapsearch} from the
command line to retrieve a certificate into a file and use it.)

As for signing messages, OpenSSL can't perform signing operations
without some kind of configuration.  Especially, you need to tell it
where your private key and your certificate is stored.  @acronym{MML}
uses an Emacs interface to OpenSSL, aptly named @code{smime.el}, and it
contain a @code{custom} group used for this configuration.  So, try
@kbd{M-x customize-group @key{RET} smime @key{RET}} and look around.

Currently there is no support for talking to a CA (or RA) to create
your own certificate.  None is planned either.  You need to do this
manually with OpenSSL or using some other program.  I used Netscape
and got a free @acronym{S/MIME} certificate from one of the big CA's on the
net.  Netscape is able to export your private key and certificate in
PKCS #12 format.  Use OpenSSL to convert this into a plain X.509
certificate in PEM format as follows.

@example
$ openssl pkcs12 -in ns.p12 -clcerts -nodes > key+cert.pem
@end example

The @file{key+cert.pem} file should be pointed to from the
@code{smime-keys} variable.  You should now be able to send signed mail.

@emph{Note!}  Your private key is now stored unencrypted in the file,
so take care in handling it.  Storing encrypted keys on the disk are
supported, and Gnus will ask you for a passphrase before invoking
OpenSSL@.  Read the OpenSSL documentation for how to achieve this.  If
you use unencrypted keys (e.g., if they are on a secure storage, or if
you are on a secure single user machine) simply press @code{RET} at
the passphrase prompt.

@node Using OpenPGP
@subsection Using OpenPGP

Use of OpenPGP requires an external software, such
as @uref{https://www.gnupg.org/, GNU Privacy Guard}.  Pre-OpenPGP
implementations such as PGP 2.x and PGP 5.x are also supported.  The
default Emacs interface to the PGP implementation is EasyPG
(@pxref{Top,,EasyPG Assistant User's Manual, epa, EasyPG Assistant
User's Manual}), but PGG (@pxref{Top, ,PGG, pgg, PGG Manual}) and
Mailcrypt are also supported.  @xref{PGP Compatibility}.

As stated earlier, messages encrypted with OpenPGP can be formatted
according to two different standards, namely @acronym{PGP} or
@acronym{PGP/MIME}.  The variables
@code{mml-default-encrypt-method} and
@code{mml-default-sign-method} determine which variant to prefer,
@acronym{PGP/MIME} by default.

@node Passphrase caching
@subsection Passphrase caching

@cindex @command{gpg-agent}
Message with EasyPG internally calls GnuPG (the @command{gpg} or
@command{gpgsm} command) to perform
data encryption, and in certain cases (decrypting or signing for
example), @command{gpg}/@command{gpgsm} requires user's passphrase.
Currently the recommended way to supply your passphrase is to use the
@command{gpg-agent} program.

In particular, the @command{gpg-agent} program supports passphrase
caching so that you do not need to enter your passphrase for every
decryption/sign operation.  @xref{Agent Options, , , gnupg, Using the
GNU Privacy Guard}.

How to use @command{gpg-agent} in Emacs depends on your version of
GnuPG@.  With GnuPG version 2.1, @command{gpg-agent} is started
automatically if necessary.  With older versions you may need to run
the following command from the shell before starting Emacs.

@example
eval `gpg-agent --daemon`
@end example

This will invoke @command{gpg-agent} and set the environment variable
@code{GPG_AGENT_INFO} to allow @command{gpg} to communicate with it.
It might be good idea to put this command in your @file{.xsession} or
@file{.bash_profile}.  @xref{Invoking GPG-AGENT, , , gnupg, Using the
GNU Privacy Guard}.

Once your @command{gpg-agent} is set up, it will ask you for a
passphrase as needed for @command{gpg}.  Under the X Window System,
you will see a new passphrase input dialog appear.  The dialog is
provided by PIN Entry (the @command{pinentry} command), reasonably
recent versions of which can also cooperate with Emacs on a text
console.  If that does not work, you may need to put a passphrase into
gpg-agent's cache beforehand.  The following command does the trick.

@example
gpg --use-agent --sign < /dev/null > /dev/null
@end example

@node PGP Compatibility
@subsection Compatibility with older implementations

@vindex gpg-temp-directory
Note, if you are using the @code{gpg.el} you must make sure that the
directory specified by @code{gpg-temp-directory} have permissions
0700.

Creating your own key is described in detail in the documentation of
your PGP implementation, so we refer to it.

If you have imported your old PGP 2.x key into GnuPG, and want to send
signed and encrypted messages to your fellow PGP 2.x users, you'll
discover that the receiver cannot understand what you send. One
solution is to use PGP 2.x instead (e.g., if you use @code{pgg}, set
@code{pgg-default-scheme} to @code{pgp}). You could also convince your
fellow PGP 2.x users to convert to GnuPG@.
@vindex mml-signencrypt-style-alist
As a final workaround, you can make the sign and encryption work in
two steps; separately sign, then encrypt a message.  If you would like
to change this behavior you can customize the
@code{mml-signencrypt-style-alist} variable.  For example:

@lisp
(setq mml-signencrypt-style-alist '(("smime" separate)
                                    ("pgp" separate)
                                    ("pgpauto" separate)
                                    ("pgpmime" separate)))
@end lisp

This causes to sign and encrypt in two passes, thus generating a
message that can be understood by PGP version 2.

(Refer to @uref{https://www.gnupg.org/gph/en/pgp2x.html} for more
information about the problem.)

@node Encrypt-to-self
@subsection Encrypt-to-self

By default, messages are encrypted to all recipients (@code{To},
@code{CC}, @code{BCC} headers).  Thus, you will not be able to decrypt
your own messages.  To make sure that messages are also encrypted to
your own key(s), several alternative solutions exist:
@enumerate
@item
Use the @code{encrypt-to} option in the file @file{gpg.conf} (for
OpenPGP) or @file{gpgsm.conf} (for @acronym{S/MIME} with EasyPG).
@xref{Invoking GPG, , , gnupg, Using the GNU Privacy Guard}, or
@xref{Invoking GPGSM, , , gnupg, Using the GNU Privacy Guard}.
@item
Include your own e-mail address (for which you created a key-pair)
among the recipients.
@item
Customize the variable @code{mml-secure-openpgp-encrypt-to-self} (for
OpenPGP) or @code{mml-secure-smime-encrypt-to-self} (for
@acronym{S/MIME} with EasyPG).
@end enumerate

@node BCC Warning
@subsection BCC Warning

The @code{BCC} header is meant to hide recipients of messages.
However, when encrypted messages are used, the e-mail addresses of all
@code{BCC}-headers are given away to all recipients without
warning, which is a bug.
@vindex mml-secure-safe-bcc-list
But now Message got to warn if @code{BCC} recipients are found in an
encrypted message when you are just about to send it.  If you are sure
those @code{BCC} addresses are safe to expose, set the
@code{mml-secure-safe-bcc-list} variable, that is a list of e-mail
addresses.  See
@uref{https://debbugs.gnu.org/cgi/bugreport.cgi?bug=18718}.


@node Various Commands
@section Various Commands

@table @kbd

@item C-c C-r
@kindex C-c C-r
@findex message-caesar-buffer-body
Caesar rotate (aka. rot13) the current message
(@code{message-caesar-buffer-body}).  If narrowing is in effect, just
rotate the visible portion of the buffer.  A numerical prefix says how
many places to rotate the text.  The default is 13.

@item C-c C-e
@kindex C-c C-e
@findex message-elide-region
@vindex message-elide-ellipsis
Elide the text between point and mark (@code{message-elide-region}).
The text is killed and replaced with the contents of the variable
@code{message-elide-ellipsis}.  The default value is to use an ellipsis
(@samp{[...]}).

This is a format-spec string, and you can use @samp{%l} to say how
many lines were removed, and @samp{%c} to say how many characters were
removed.

@item C-c M-k
@kindex C-c M-k
@findex message-kill-address
Kill the address under point.

@item C-c C-z
@kindex C-c C-z
@findex message-kill-to-signature
Kill all the text up to the signature, or if that's missing, up to the
end of the message (@code{message-kill-to-signature}).

@item C-c C-v
@kindex C-c C-v
@findex message-delete-not-region
Delete all text in the body of the message that is outside the region
(@code{message-delete-not-region}).

@item M-@key{RET}
@kindex M-RET
@findex message-newline-and-reformat
Insert four newlines, and then reformat if inside quoted text.

Here's an example:

@example
> This is some quoted text.  And here's more quoted text.
@end example

If point is before @samp{And} and you press @kbd{M-@key{RET}}, you'll get:

@example
> This is some quoted text.

*

> And here's more quoted text.
@end example

@samp{*} says where point will be placed.

@item C-c M-r
@kindex C-c M-r
@findex message-rename-buffer
Rename the buffer (@code{message-rename-buffer}).  If given a prefix,
prompt for a new buffer name.

@item @key{TAB}
@kindex TAB
@findex message-tab
@vindex message-tab-body-function
If @code{message-tab-body-function} is non-@code{nil}, execute the
function it specifies.  Otherwise use the function bound to @key{TAB} in
@code{text-mode-map} or @code{global-map}.

@end table


@node Sending
@section Sending

@table @kbd
@item C-c C-c
@kindex C-c C-c
@findex message-send-and-exit
Send the message and bury the current buffer
(@code{message-send-and-exit}).

@item C-c C-s
@kindex C-c C-s
@findex message-send
Send the message (@code{message-send}).

@item C-c C-d
@kindex C-c C-d
@findex message-dont-send
Bury the message buffer and exit (@code{message-dont-send}).

@item C-c C-k
@kindex C-c C-k
@findex message-kill-buffer
Kill the message buffer and exit (@code{message-kill-buffer}).

@end table



@node Mail Aliases
@section Mail Aliases
@cindex mail aliases
@cindex aliases
@cindex completion
@cindex ecomplete

@vindex message-mail-alias-type
The @code{message-mail-alias-type} variable controls what type of mail
alias expansion to use.  Currently two forms are supported:
@code{mailabbrev} and @code{ecomplete}.  If this variable is
@code{nil}, no mail alias expansion will be performed.

@code{mailabbrev} works by parsing the @file{/etc/mailrc} and
@file{~/.mailrc} files.  These files look like:

@example
alias lmi "Lars Magne Ingebrigtsen <larsi@@ifi.uio.no>"
alias ding "ding@@ifi.uio.no (ding mailing list)"
@end example

After adding lines like this to your @file{~/.mailrc} file, you should
be able to just write @samp{lmi} in the @code{To} or @code{CC} (and so
on) headers and press @kbd{SPC} to expand the alias.

No expansion will be performed upon sending of the message---all
expansions have to be done explicitly.

If you're using @code{ecomplete}, all addresses from @code{To} and
@code{CC} headers will automatically be put into the
@file{~/.ecompleterc} file.  When you enter text in the @code{To} and
@code{CC} headers, @code{ecomplete} will check out the values stored
there and ``electrically'' say what completions are possible.  To
choose one of these completions, use the @kbd{M-n} command to move
down to the list.  Use @kbd{@key{DOWN}} or @kbd{M-n} and
@kbd{@key{UP}} or @kbd{M-p} to move down and up the list, and
@kbd{@key{RET}} to choose a completion.

The @code{ecomplete-sort-predicate} variable controls how
@code{ecomplete} matches are sorted.

@node Spelling
@section Spelling
@cindex spelling
@findex ispell-message

There are two popular ways to have Emacs spell-check your messages:
@code{ispell} and @code{flyspell}.  @code{ispell} is the older and
probably more popular package.  You typically first write the message,
and then run the entire thing through @code{ispell} and fix all the
typos.  To have this happen automatically when you send a message, put
something like the following in your @file{.emacs} file:

@lisp
(add-hook 'message-send-hook 'ispell-message)
@end lisp

@vindex ispell-message-dictionary-alist
If you're in the habit of writing in different languages, this can be
controlled by the @code{ispell-message-dictionary-alist} variable:

@lisp
(setq ispell-message-dictionary-alist
      '(("^Newsgroups:.*\\bde\\." . "deutsch8")
        (".*" . "default")))
@end lisp

@code{ispell} depends on having the external @samp{ispell} command
installed.

The other popular method is using @code{flyspell}.  This package checks
your spelling while you're writing, and marks any mis-spelled words in
various ways.

To use @code{flyspell}, put something like the following in your
@file{.emacs} file:

@lisp
(defun my-message-setup-routine ()
  (flyspell-mode 1))
(add-hook 'message-setup-hook 'my-message-setup-routine)
@end lisp

@code{flyspell} depends on having the external @samp{ispell} command
installed.


@node Variables
@chapter Variables

@menu
* Message Headers::             General message header stuff.
* Mail Headers::                Customizing mail headers.
* Mail Variables::              Other mail variables.
* News Headers::                Customizing news headers.
* News Variables::              Other news variables.
* Insertion Variables::         Customizing how things are inserted.
* Various Message Variables::   Other message variables.
* Sending Variables::           Variables for sending.
* Message Buffers::             How Message names its buffers.
* Message Actions::             Actions to be performed when exiting.
@end menu


@node Message Headers
@section Message Headers

Message is quite aggressive on the message generation front.  It has to
be---it's a combined news and mail agent.  To be able to send combined
messages, it has to generate all headers itself (instead of letting the
mail/news system do it) to ensure that mail and news copies of messages
look sufficiently similar.

@table @code

@item message-generate-headers-first
@vindex message-generate-headers-first
If @code{t}, generate all required headers before starting to
compose the message.  This can also be a list of headers to generate:

@lisp
(setq message-generate-headers-first
      '(References))
@end lisp

@vindex message-required-headers
The variables @code{message-required-headers},
@code{message-required-mail-headers} and
@code{message-required-news-headers} specify which headers are
required.

Note that some headers will be removed and re-generated before posting,
because of the variable @code{message-deletable-headers} (see below).

@item message-draft-headers
@vindex message-draft-headers
When running Message from Gnus, the message buffers are associated
with a draft group.  @code{message-draft-headers} says which headers
should be generated when a draft is written to the draft group.

@item message-from-style
@vindex message-from-style
Specifies how @code{From} headers should look.  There are four valid
values:

@table @code
@item nil
Just the address---@samp{king@@grassland.com}.

@item parens
@samp{king@@grassland.com (Elvis Parsley)}.

@item angles
@samp{Elvis Parsley <king@@grassland.com>}.

@item default
Look like @code{angles} if that doesn't require quoting, and
@code{parens} if it does.  If even @code{parens} requires quoting, use
@code{angles} anyway.

@end table

@item message-deletable-headers
@vindex message-deletable-headers
Headers in this list that were previously generated by Message will be
deleted before posting.  Let's say you post an article.  Then you decide
to post it again to some other group, you naughty boy, so you jump back
to the @file{*post-buf*} buffer, edit the @code{Newsgroups} line, and
ship it off again.  By default, this variable makes sure that the old
generated @code{Message-ID} is deleted, and a new one generated.  If
this isn't done, the entire empire would probably crumble, anarchy would
prevail, and cats would start walking on two legs and rule the world.
Allegedly.

@item message-default-headers
@vindex message-default-headers
Header lines to be inserted in outgoing messages before you edit the
message, so you can edit or delete their lines. If set to a string, it
is directly inserted. If set to a function, it is called and its
result is inserted.

@item message-subject-re-regexp
@vindex message-subject-re-regexp
@cindex Aw
@cindex Sv
@cindex Re
Responses to messages have subjects that start with @samp{Re: }.  This
is @emph{not} an abbreviation of the English word ``response'', but is
Latin, and means ``in response to''.  Some illiterate nincompoops have
failed to grasp this fact, and have ``internationalized'' their software
to use abominations like @samp{Aw: } (``antwort'') or @samp{Sv: }
(``svar'') instead, which is meaningless and evil.  However, you may
have to deal with users that use these evil tools, in which case you may
set this variable to a regexp that matches these prefixes.  Myself, I
just throw away non-compliant mail.

Here's an example of a value to deal with these headers when
responding to a message:

@lisp
(setq message-subject-re-regexp
      (concat
       "^[ \t]*"
         "\\("
           "\\("
             "[Aa][Nn][Tt][Ww]\\.?\\|"     ; antw
             "[Aa][Ww]\\|"                 ; aw
             "[Ff][Ww][Dd]?\\|"            ; fwd
             "[Oo][Dd][Pp]\\|"             ; odp
             "[Rr][Ee]\\|"                 ; re
             "[Rr][\311\351][Ff]\\.?\\|"   ; ref
             "[Ss][Vv]"                    ; sv
           "\\)"
           "\\(\\[[0-9]*\\]\\)"
           "*:[ \t]*"
         "\\)"
       "*[ \t]*"
       ))
@end lisp

@item message-subject-trailing-was-query
@vindex message-subject-trailing-was-query
@vindex message-subject-trailing-was-ask-regexp
@vindex message-subject-trailing-was-regexp
Controls what to do with trailing @samp{(was: <old subject>)} in subject
lines.  If @code{nil}, leave the subject unchanged.  If it is the symbol
@code{ask}, query the user what to do.  In this case, the subject is
matched against @code{message-subject-trailing-was-ask-regexp}.  If
@code{message-subject-trailing-was-query} is @code{t}, always strip the
trailing old subject.  In this case,
@code{message-subject-trailing-was-regexp} is used.

@item message-alternative-emails
@vindex message-alternative-emails
Regexp or predicate function matching alternative email addresses.
The first address in the To, CC or From headers of the original
article matching this variable is used as the From field of outgoing
messages, replacing the default From value.

For example, if you have two secondary email addresses john@@home.net
and john.doe@@work.com and want to use them in the From field when
composing a reply to a message addressed to one of them, you could set
this variable like this:

@lisp
(setq message-alternative-emails
      (regexp-opt '("john@@home.net" "john.doe@@work.com")))
@end lisp

This variable has precedence over posting styles and anything that runs
off @code{message-setup-hook}.

@item message-allow-no-recipients
@vindex message-allow-no-recipients
Specifies what to do when there are no recipients other than
@code{Gcc} or @code{FCC}.  If it is @code{always}, the posting is
allowed.  If it is @code{never}, the posting is not allowed.  If it is
@code{ask} (the default), you are prompted.

@item message-hidden-headers
@vindex message-hidden-headers
A regexp, a list of regexps, or a list where the first element is
@code{not} and the rest are regexps.  It says which headers to keep
hidden when composing a message.

@lisp
(setq message-hidden-headers
      '(not "From" "Subject" "To" "CC" "Newsgroups"))
@end lisp

Headers are hidden using narrowing, you can use @kbd{M-x widen} to
expose them in the buffer.

@item message-header-synonyms
@vindex message-header-synonyms
A list of lists of header synonyms.  E.g., if this list contains a
member list with elements @code{CC} and @code{To}, then
@code{message-carefully-insert-headers} will not insert a @code{To}
header when the message is already @code{CC}ed to the recipient.

@end table


@node Mail Headers
@section Mail Headers

@table @code
@item message-required-mail-headers
@vindex message-required-mail-headers
@xref{News Headers}, for the syntax of this variable.  It is
@code{(From Subject Date (optional . In-Reply-To) Message-ID
(optional . User-Agent))} by default.

@item message-ignored-mail-headers
@vindex message-ignored-mail-headers
Regexp of headers to be removed before mailing.  The default is@*
@samp{^[GF]cc:\\|^Resent-FCC:\\|^Xref:\\|^X-Draft-From:\\|@*
^X-Gnus-Agent-Meta-Information:}.

@item message-default-mail-headers
@vindex message-default-mail-headers
This string is inserted at the end of the headers in all message
buffers that are initialized as mail.

@item message-generate-hashcash
@vindex message-generate-hashcash
Variable that indicates whether @samp{X-Hashcash} headers
should be computed for the message.  @xref{Hashcash, ,Hashcash,gnus,
The Gnus Manual}.  If @code{opportunistic}, only generate the headers
when it doesn't lead to the user having to wait.

@end table


@node Mail Variables
@section Mail Variables

@table @code
@item message-send-mail-function
@vindex message-send-mail-function
@findex message-send-mail-function
@findex message-send-mail-with-sendmail
@findex message-send-mail-with-mh
@findex message-send-mail-with-qmail
@findex message-smtpmail-send-it
@findex smtpmail-send-it
@findex feedmail-send-it
@findex message-send-mail-with-mailclient
Function used to send the current buffer as mail.  The default is
@code{message-send-mail-with-sendmail}, or @code{smtpmail-send-it}
according to the system.  Other valid values include
@code{message-send-mail-with-mailclient},
@code{message-send-mail-with-mh}, @code{message-send-mail-with-qmail},
@code{message-smtpmail-send-it} and @code{feedmail-send-it}.

The function
@code{message-send-mail-with-sendmail} pipes your article to the
@code{sendmail} binary for further queuing and sending.  When your local
system is not configured for sending mail using @code{sendmail}, and you
have access to a remote @acronym{SMTP} server, you can set
@code{message-send-mail-function} to @code{smtpmail-send-it} and make
sure to setup the @code{smtpmail} package correctly.  An example:

@lisp
(setq message-send-mail-function 'smtpmail-send-it
      smtpmail-default-smtp-server "YOUR SMTP HOST")
@end lisp

To the thing similar to this, there is
@code{message-smtpmail-send-it}.  It is useful if your @acronym{ISP}
requires the @acronym{POP}-before-@acronym{SMTP} authentication.
@xref{POP before SMTP, , POP before SMTP, gnus, The Gnus Manual}.

@cindex X-Message-SMTP-Method
If you have a complex @acronym{SMTP} setup, and want some messages to
go via one mail server, and other messages to go through another, you
can use the @samp{X-Message-SMTP-Method} header.  These are the
supported values:

@table @samp
@item smtpmail

@example
X-Message-SMTP-Method: smtp smtp.fsf.org 587
@end example

This will send the message via @samp{smtp.fsf.org}, using port 587.

@example
X-Message-SMTP-Method: smtp smtp.fsf.org 587 other-user
@end example

This is the same as the above, but uses @samp{other-user} as the user
name when authenticating.  This is handy if you have several
@acronym{SMTP} accounts on the same server.

@item sendmail

@example
X-Message-SMTP-Method: sendmail
@end example

This will send the message via the locally installed sendmail/exim/etc
installation.

@end table

@item message-mh-deletable-headers
@vindex message-mh-deletable-headers
Most versions of MH doesn't like being fed messages that contain the
headers in this variable.  If this variable is non-@code{nil} (which is
the default), these headers will be removed before mailing when sending
messages via MH@.  Set it to @code{nil} if your MH can handle these
headers.

@item message-qmail-inject-program
@vindex message-qmail-inject-program
@cindex qmail
Location of the qmail-inject program.

@item message-qmail-inject-args
@vindex message-qmail-inject-args
Arguments passed to qmail-inject programs.
This should be a list of strings, one string for each argument.  It
may also be a function.

E.g., if you wish to set the envelope sender address so that bounces
go to the right place or to deal with listserv's usage of that address, you
might set this variable to @code{'("-f" "you@@some.where")}.

@item message-sendmail-f-is-evil
@vindex message-sendmail-f-is-evil
@cindex sendmail
Non-@code{nil} means don't add @samp{-f username} to the sendmail
command line.  Doing so would be even more evil than leaving it out.

@item message-sendmail-envelope-from
@vindex message-sendmail-envelope-from
When @code{message-sendmail-f-is-evil} is @code{nil}, this specifies
the address to use in the @acronym{SMTP} envelope.  If it is
@code{nil}, use @code{user-mail-address}.  If it is the symbol
@code{header}, use the @samp{From} header of the message.

@item message-mailer-swallows-blank-line
@vindex message-mailer-swallows-blank-line
Set this to non-@code{nil} if the system's mailer runs the header and
body together.  (This problem exists on SunOS 4 when sendmail is run
in remote mode.)  The value should be an expression to test whether
the problem will actually occur.

@item message-send-mail-partially-limit
@vindex message-send-mail-partially-limit
@cindex split large message
The limitation of messages sent as message/partial.  The lower bound
of message size in characters, beyond which the message should be sent
in several parts.  If it is @code{nil} (which is the default), the
size is unlimited.

@end table


@node News Headers
@section News Headers

@vindex message-required-news-headers
@code{message-required-news-headers} a list of header symbols.  These
headers will either be automatically generated, or, if that's
impossible, they will be prompted for.  The following symbols are valid:

@table @code

@item From
@cindex From
@findex user-full-name
@findex user-mail-address
This required header will be filled out with the result of the
@code{message-make-from} function, which depends on the
@code{message-from-style}, @code{user-full-name},
@code{user-mail-address} variables.

@item Subject
@cindex Subject
This required header will be prompted for if not present already.

@item Newsgroups
@cindex Newsgroups
This required header says which newsgroups the article is to be posted
to.  If it isn't present already, it will be prompted for.

@item Organization
@cindex organization
@vindex message-user-organization
@vindex message-user-organization-file
This optional header will be filled out depending on the
@code{message-user-organization} variable.
@code{message-user-organization-file} will be used if this variable is
@code{t}.  This variable can also be a string (in which case this string
will be used), or it can be a function (which will be called with no
parameters and should return a string to be used).

@item Lines
@cindex Lines
This optional header will be computed by Message.

@item Message-ID
@cindex Message-ID
@vindex message-user-fqdn
@vindex mail-host-address
@vindex user-mail-address
@findex system-name
@cindex Sun
@cindex i-did-not-set--mail-host-address--so-tickle-me
This required header will be generated by Message.  A unique ID will be
created based on the date, time, user name (for the local part) and the
domain part.  For the domain part, message will look (in this order) at
@code{message-user-fqdn}, @code{system-name}, @code{mail-host-address}
and @code{message-user-mail-address} (i.e., @code{user-mail-address})
until a probably valid fully qualified domain name (FQDN) was found.

@item User-Agent
@cindex User-Agent
This optional header will be filled out according to the
@code{message-newsreader} local variable.

@item In-Reply-To
This optional header is filled out using the @code{Date} and @code{From}
header of the article being replied to.

@item Expires
@cindex Expires
@vindex message-expires
This extremely optional header will be inserted according to the
@code{message-expires} variable.  It is highly deprecated and shouldn't
be used unless you know what you're doing.

@item Distribution
@cindex Distribution
@vindex message-distribution-function
This optional header is filled out according to the
@code{message-distribution-function} variable.  It is a deprecated and
much misunderstood header.

@item Path
@cindex path
@vindex message-user-path
This extremely optional header should probably never be used.
However, some @emph{very} old servers require that this header is
present.  @code{message-user-path} further controls how this
@code{Path} header is to look.  If it is @code{nil}, use the server name
as the leaf node.  If it is a string, use the string.  If it is neither
a string nor @code{nil}, use the user name only.  However, it is highly
unlikely that you should need to fiddle with this variable at all.
@end table

@cindex Mime-Version
In addition, you can enter conses into this list.  The @sc{car} of this cons
should be a symbol.  This symbol's name is the name of the header, and
the @sc{cdr} can either be a string to be entered verbatim as the value of
this header, or it can be a function to be called.  This function should
take no arguments, and return a string to be inserted.  For
instance, if you want to insert @code{Mime-Version: 1.0}, you should
enter @code{(Mime-Version . "1.0")} into the list.

If the list contains a cons where the @sc{car} of the cons is
@code{optional}, the @sc{cdr} of this cons will only be inserted if it is
non-@code{nil}.

If you want to delete an entry from this list, the following Lisp
snippet might be useful.  Adjust accordingly if you want to remove
another element.

@lisp
(setq message-required-news-headers
      (delq 'Message-ID message-required-news-headers))
@end lisp

Other variables for customizing outgoing news articles:

@table @code

@item message-syntax-checks
@vindex message-syntax-checks
Controls what syntax checks should not be performed on outgoing posts.
To disable checking of long signatures, for instance, add

@lisp
(signature . disabled)
@end lisp

to this list.

Valid checks are:

@table @code
@item approved
@cindex approved
Check whether the article has an @code{Approved} header, which is
something only moderators should include.
@item continuation-headers
Check whether there are continuation header lines that don't begin with
whitespace.
@item control-chars
Check for invalid characters.
@item empty
Check whether the article is empty.
@item existing-newsgroups
Check whether the newsgroups mentioned in the @code{Newsgroups} and
@code{Followup-To} headers exist.
@item from
Check whether the @code{From} header seems nice.
@item illegible-text
Check whether there is any non-printable character in the body.
@item invisible-text
Check whether there is any invisible text in the buffer.
@item long-header-lines
Check for too long header lines.
@item long-lines
@cindex long lines
Check for too long lines in the body.
@item message-id
Check whether the @code{Message-ID} looks syntactically ok.
@item multiple-headers
Check for the existence of multiple equal headers.
@item new-text
Check whether there is any new text in the messages.
@item newsgroups
Check whether the @code{Newsgroups} header exists and is not empty.
@item quoting-style
Check whether text follows last quoted portion.
@item repeated-newsgroups
Check whether the @code{Newsgroups} and @code{Followup-To} headers
contains repeated group names.
@item reply-to
Check whether the @code{Reply-To} header looks ok.
@item sender
@cindex Sender
Insert a new @code{Sender} header if the @code{From} header looks odd.
@item sendsys
@cindex sendsys
Check for the existence of version and sendsys commands.
@item shoot
Check whether the domain part of the @code{Message-ID} header looks ok.
@item shorten-followup-to
Check whether to add a @code{Followup-To} header to shorten the number
of groups to post to.
@item signature
Check the length of the signature.
@item size
Check for excessive size.
@item subject
Check whether the @code{Subject} header exists and is not empty.
@item subject-cmsg
Check the subject for commands.
@item valid-newsgroups
Check whether the @code{Newsgroups} and @code{Followup-To} headers
are valid syntactically.
@end table

All these conditions are checked by default, except for @code{sender}
for which the check is disabled by default if
@code{message-insert-canlock} is non-@code{nil} (@pxref{Canceling News}).

@item message-ignored-news-headers
@vindex message-ignored-news-headers
Regexp of headers to be removed before posting.  The default is@*
@samp{^NNTP-Posting-Host:\\|^Xref:\\|^[BGF]cc:\\|^Resent-FCC:\\|@*
^X-Draft-From:\\|^X-Gnus-Agent-Meta-Information:}.

@item message-default-news-headers
@vindex message-default-news-headers
This string is inserted at the end of the headers in all message
buffers that are initialized as news.

@end table


@node News Variables
@section News Variables

@table @code
@item message-send-news-function
@vindex message-send-news-function
Function used to send the current buffer as news.  The default is
@code{message-send-news}.

@item message-post-method
@vindex message-post-method
Gnusish @dfn{select method} (see the Gnus manual for details) used for
posting a prepared news message.

@end table


@node Insertion Variables
@section Insertion Variables

@table @code
@item message-cite-style
@vindex message-cite-style
The overall style to be used when replying to messages. This controls
things like where the reply should be put relative to the original,
how the citation is formatted, where the signature goes, etc.

Value is either @code{nil} (no variable overrides) or a let-style list
of pairs @code{(VARIABLE VALUE)} to override default values.

See @code{gnus-posting-styles} to set this variable for specific
groups. Presets to impersonate popular mail agents are available in the
@code{message-cite-style-*} variables.

@item message-cite-reply-position
@vindex message-cite-reply-position
Where the reply should be positioned. Available styles are
@code{traditional} to reply inline, @code{above} for top-posting, and
@code{below} for bottom-posting

@item message-ignored-cited-headers
@vindex message-ignored-cited-headers
All headers that match this regexp will be removed from yanked
messages.  The default is @samp{.}, which means that all headers will be
removed.

@item message-cite-prefix-regexp
@vindex message-cite-prefix-regexp
Regexp matching the longest possible citation prefix on a line.

@item message-citation-line-function
@vindex message-citation-line-function
@cindex attribution line
Function called to insert the citation line.  The default is
@code{message-insert-citation-line}, which will lead to citation lines
that look like:

@example
Hallvard B Furuseth <h.b.furuseth@@usit.uio.no> writes:
@end example

@c FIXME: Add 'message-insert-formatted-citation-line' and
@c 'message-citation-line-format'.

Point will be at the beginning of the body of the message when this
function is called.

Note that Gnus provides a feature where clicking on @samp{writes:} hides the
cited text.  If you change the citation line too much, readers of your
messages will have to adjust their Gnus, too.  See the variable
@code{gnus-cite-attribution-suffix}.  @xref{Article Highlighting, ,
Article Highlighting, gnus, The Gnus Manual}, for details.

@item message-yank-prefix
@vindex message-yank-prefix
@cindex yanking
@cindex quoting
When you are replying to or following up an article, you normally want
to quote the person you are answering.  Inserting quoted text is done by
@dfn{yanking}, and each line you yank will have
@code{message-yank-prefix} prepended to it (except for quoted lines
which use @code{message-yank-cited-prefix} and empty lines which use
@code{message-yank-empty-prefix}).  The default is @samp{> }.

@item message-yank-cited-prefix
@vindex message-yank-cited-prefix
@cindex yanking
@cindex cited
@cindex quoting
When yanking text from an article which contains already cited text,
each line will be prefixed with the contents of this variable.  The
default is @samp{>}.  See also @code{message-yank-prefix}.

@item message-yank-empty-prefix
@vindex message-yank-empty-prefix
@cindex yanking
@cindex quoting
When yanking text from an article, each empty line will be prefixed with
the contents of this variable.  The default is @samp{>}.  You can set
this variable to an empty string to split the cited text into paragraphs
automatically.  See also @code{message-yank-prefix}.

@item message-indentation-spaces
@vindex message-indentation-spaces
Number of spaces to indent yanked messages.

@item message-cite-function
@vindex message-cite-function
@findex message-cite-original
@findex message-cite-original-without-signature
Function for citing an original message.  The default is
@code{message-cite-original}, which simply inserts the original message
and prepends @samp{> } to each line.
@code{message-cite-original-without-signature} does the same, but elides
the signature.

@item message-indent-citation-function
@vindex message-indent-citation-function
Function for modifying a citation just inserted in the mail buffer.
This can also be a list of functions.  Each function can find the
citation between @code{(point)} and @code{(mark t)}.  And each function
should leave point and mark around the citation text as modified.

@item message-mark-insert-begin
@vindex message-mark-insert-begin
String to mark the beginning of some inserted text.

@item message-mark-insert-end
@vindex message-mark-insert-end
String to mark the end of some inserted text.

@item message-signature
@vindex message-signature
String to be inserted at the end of the message buffer.  If @code{t}
(which is the default), the @code{message-signature-file} file will be
inserted instead.  If a function, the result from the function will be
used instead.  If a form, the result from the form will be used instead.
If this variable is @code{nil}, no signature will be inserted at all.

@item message-signature-file
@vindex message-signature-file
File containing the signature to be inserted at the end of the buffer.
If a path is specified, the value of
@code{message-signature-directory} is ignored, even if set.
The default is @file{~/.signature}.

@item message-signature-directory
@vindex message-signature-directory
Name of directory containing signature files.  Comes in handy if you
have many such files, handled via Gnus posting styles for instance.
If @code{nil} (the default), @code{message-signature-file} is expected
to specify the directory if needed.


@item message-signature-insert-empty-line
@vindex message-signature-insert-empty-line
If @code{t} (the default value) an empty line is inserted before the
signature separator.

@end table

Note that RFC1036bis says that a signature should be preceded by the three
characters @samp{-- } on a line by themselves.  This is to make it
easier for the recipient to automatically recognize and process the
signature.  So don't remove those characters, even though you might feel
that they ruin your beautiful design, like, totally.

Also note that no signature should be more than four lines long.
Including @acronym{ASCII} graphics is an efficient way to get
everybody to believe that you are silly and have nothing important to
say.


@node Various Message Variables
@section Various Message Variables

@table @code
@item message-default-charset
@vindex message-default-charset
@cindex charset
Symbol naming a @acronym{MIME} charset.  Non-@acronym{ASCII} characters
in messages are assumed to be encoded using this charset.  The default
is @code{iso-8859-1} on non-@sc{mule} Emacsen; otherwise @code{nil},
which means ask the user.  (This variable is used only on non-@sc{mule}
Emacsen.)  @xref{Charset Translation, , Charset Translation, emacs-mime,
Emacs MIME Manual}, for details on the @sc{mule}-to-@acronym{MIME}
translation process.

@item message-fill-column
@vindex message-fill-column
@cindex auto-fill
Local value for the column beyond which automatic line-wrapping should
happen for message buffers.  If non-@code{nil} (the default), also turn on
auto-fill in message buffers.

@item message-signature-separator
@vindex message-signature-separator
Regexp matching the signature separator.  It is @samp{^-- *$} by
default.

@item mail-header-separator
@vindex mail-header-separator
String used to separate the headers from the body.  It is @samp{--text
follows this line--} by default.

@item message-directory
@vindex message-directory
Directory used by many mailish things.  The default is @file{~/Mail/}.
All other mail file variables are derived from @code{message-directory}.

@item message-auto-save-directory
@vindex message-auto-save-directory
Directory where Message auto-saves buffers if Gnus isn't running.  If
@code{nil}, Message won't auto-save.  The default is @file{~/Mail/drafts/}.

@item message-signature-setup-hook
@vindex message-signature-setup-hook
Hook run when initializing the message buffer.  It is run after the
headers have been inserted but before the signature has been inserted.

@item message-setup-hook
@vindex message-setup-hook
Hook run as the last thing when the message buffer has been initialized,
but before yanked text is inserted.

@item message-header-setup-hook
@vindex message-header-setup-hook
Hook called narrowed to the headers after initializing the headers.

For instance, if you're running Gnus and wish to insert a
@samp{Mail-Copies-To} header in all your news articles and all messages
you send to mailing lists, you could do something like the following:

@lisp
(defun my-message-header-setup-hook ()
  (let ((group (or gnus-newsgroup-name "")))
    (when (or (message-fetch-field "newsgroups")
              (gnus-group-find-parameter group 'to-address)
              (gnus-group-find-parameter group 'to-list))
      (insert "Mail-Copies-To: never\n"))))

(add-hook 'message-header-setup-hook
          'my-message-header-setup-hook)
@end lisp

@item message-send-hook
@vindex message-send-hook
Hook run before sending messages.

If you want to add certain headers before sending, you can use the
@code{message-add-header} function in this hook.  For instance:
@findex message-add-header

@lisp
(add-hook 'message-send-hook 'my-message-add-content)
(defun my-message-add-content ()
  (message-add-header "X-In-No-Sense: Nonsense")
  (message-add-header "X-Whatever: no"))
@end lisp

This function won't add the header if the header is already present.

@item message-send-mail-hook
@vindex message-send-mail-hook
Hook run before sending mail messages.  This hook is run very late:
just before the message is actually sent as mail.

@item message-send-news-hook
@vindex message-send-news-hook
Hook run before sending news messages.  This hook is run very late:
just before the message is actually sent as news.

@item message-sent-hook
@vindex message-sent-hook
Hook run after sending messages.

@item message-cancel-hook
@vindex message-cancel-hook
Hook run when canceling news articles.

@item message-mode-syntax-table
@vindex message-mode-syntax-table
Syntax table used in message mode buffers.

@item message-cite-articles-with-x-no-archive
@vindex message-cite-articles-with-x-no-archive
If non-@code{nil}, don't strip quoted text from articles that have
@samp{X-No-Archive} set.  Even if this variable isn't set, you can
undo the stripping by hitting the @code{undo} keystroke.

@item message-strip-special-text-properties
@vindex message-strip-special-text-properties
Emacs has a number of special text properties which can break message
composing in various ways.  If this option is set, message will strip
these properties from the message composition buffer.  However, some
packages requires these properties to be present in order to work.  If
you use one of these packages, turn this option off, and hope the
message composition doesn't break too bad.

@item message-send-method-alist
@vindex message-send-method-alist
@findex message-mail-p
@findex message-news-p
@findex message-send-via-mail
@findex message-send-via-news
Alist of ways to send outgoing messages.  Each element has the form:

@lisp
(@var{type} @var{predicate} @var{function})
@end lisp

@table @var
@item type
A symbol that names the method.

@item predicate
A function called without any parameters to determine whether the
message is a message of type @var{type}.  The function will be called in
the buffer where the message is.

@item function
A function to be called if @var{predicate} returns non-@code{nil}.
@var{function} is called with one parameter---the prefix.
@end table

The default is:

@lisp
((news message-news-p message-send-via-news)
 (mail message-mail-p message-send-via-mail))
@end lisp

The @code{message-news-p} function returns non-@code{nil} if the message
looks like news, and the @code{message-send-via-news} function sends the
message according to the @code{message-send-news-function} variable
(@pxref{News Variables}).  The @code{message-mail-p} function returns
non-@code{nil} if the message looks like mail, and the
@code{message-send-via-mail} function sends the message according to the
@code{message-send-mail-function} variable (@pxref{Mail Variables}).

All the elements in this alist will be tried in order, so a message
containing both a valid @samp{Newsgroups} header and a valid @samp{To}
header, for example, will be sent as news, and then as mail.
@end table



@node Sending Variables
@section Sending Variables

@table @code

@item message-fcc-handler-function
@vindex message-fcc-handler-function
A function called to save outgoing articles.  This function will be
called with the name of the file to store the article in.  The default
function is @code{message-output} which saves in Unix mailbox format.

@item message-courtesy-message
@vindex message-courtesy-message
When sending combined messages, this string is inserted at the start of
the mailed copy.  If the string contains the format spec @samp{%s}, the
newsgroups the article has been posted to will be inserted there.  If
this variable is @code{nil}, no such courtesy message will be added.
The default value is @samp{"The following message is a courtesy copy of
an article\\nthat has been posted to %s as well.\\n\\n"}.

@item message-fcc-externalize-attachments
@vindex message-fcc-externalize-attachments
If @code{nil}, attach files as normal parts in FCC copies; if it is
non-@code{nil}, attach local files as external parts.

@item message-interactive
@vindex message-interactive
If non-@code{nil} wait for and display errors when sending a message;
if @code{nil} let the mailer mail back a message to report errors.

@item message-confirm-send
@vindex message-confirm-send
When non-@code{nil}, Gnus will ask for confirmation when sending a
message.

@end table


@node Message Buffers
@section Message Buffers

Message will generate new buffers with unique buffer names when you
request a message buffer.  When you send the message, the buffer isn't
normally killed off.  Its name is changed and a certain number of old
message buffers are kept alive.

@table @code
@item message-generate-new-buffers
@vindex message-generate-new-buffers
Controls whether to create a new message buffer to compose a message.
Valid values include:

@table @code
@item nil
Generate the buffer name in the Message way (e.g., *mail*, *news*, *mail
to whom*, *news on group*, etc.)@: and continue editing in the existing
buffer of that name.  If there is no such buffer, it will be newly
created.

@item unique
@item t
Create the new buffer with the name generated in the Message way.

@item unsent
Similar to @code{unique} but the buffer name begins with "*unsent ".

@item standard
Similar to @code{nil} but the buffer name is simpler like *mail
message*.
@end table
@table @var
@item function
If this is a function, call that function with three parameters: The
type, the To address and the group name (any of these may be
@code{nil}).  The function should return the new buffer name.
@end table

The default value is @code{unsent}.

@item message-max-buffers
@vindex message-max-buffers
This variable says how many old message buffers to keep.  If there are
more message buffers than this, the oldest buffer will be killed.  The
default is 10.  If this variable is @code{nil}, no old message buffers
will ever be killed.

@item message-send-rename-function
@vindex message-send-rename-function
After sending a message, the buffer is renamed from, for instance,
@samp{*reply to Lars*} to @samp{*sent reply to Lars*}.  If you don't
like this, set this variable to a function that renames the buffer in a
manner you like.  If you don't want to rename the buffer at all, you can
say:

@lisp
(setq message-send-rename-function 'ignore)
@end lisp

@item message-kill-buffer-on-exit
@findex message-kill-buffer-on-exit
If non-@code{nil}, kill the buffer immediately on exit.

@end table


@node Message Actions
@section Message Actions

When Message is being used from a news/mail reader, the reader is likely
to want to perform some task after the message has been sent.  Perhaps
return to the previous window configuration or mark an article as
replied.

@vindex message-kill-actions
@vindex message-postpone-actions
@vindex message-exit-actions
@vindex message-send-actions
The user may exit from the message buffer in various ways.  The most
common is @kbd{C-c C-c}, which sends the message and exits.  Other
possibilities are @kbd{C-c C-s} which just sends the message, @kbd{C-c
C-d} which postpones the message editing and buries the message buffer,
and @kbd{C-c C-k} which kills the message buffer.  Each of these actions
have lists associated with them that contains actions to be executed:
@code{message-send-actions}, @code{message-exit-actions},
@code{message-postpone-actions}, and @code{message-kill-actions}.

Message provides a function to interface with these lists:
@code{message-add-action}.  The first parameter is the action to be
added, and the rest of the arguments are which lists to add this action
to.  Here's an example from Gnus:

@lisp
  (message-add-action
   `(set-window-configuration ,(current-window-configuration))
   'exit 'postpone 'kill)
@end lisp

This restores the Gnus window configuration when the message buffer is
killed, postponed or exited.

An @dfn{action} can be either: a normal function, or a list where the
@sc{car} is a function and the @sc{cdr} is the list of arguments, or
a form to be @code{eval}ed.


@node Appendices
@chapter Appendices

@menu
* Responses::          Standard rules for determining where responses go.
@end menu


@node Responses
@section Responses

To determine where a message is to go, the following algorithm is used
by default.

@table @dfn
@item reply
A @dfn{reply} is when you want to respond @emph{just} to the person who
sent the message via mail.  There will only be one recipient.  To
determine who the recipient will be, the following headers are
consulted, in turn:

@table @code
@item Reply-To

@item From
@end table


@item wide reply
A @dfn{wide reply} is a mail response that includes @emph{all} entities
mentioned in the message you are responding to.  All mailboxes from the
following headers will be concatenated to form the outgoing
@code{To}/@code{CC} headers:

@table @code
@item From
(unless there's a @code{Reply-To}, in which case that is used instead).

@item CC

@item To
@end table

If a @code{Mail-Copies-To} header is present, it will also be included
in the list of mailboxes.  If this header is @samp{never}, that means
that the @code{From} (or @code{Reply-To}) mailbox will be suppressed.


@item followup
A @dfn{followup} is a response sent via news.  The following headers
(listed in order of precedence) determine where the response is to be
sent:

@table @code

@item Followup-To

@item Newsgroups

@end table

If a @code{Mail-Copies-To} header is present, it will be used as the
basis of the new @code{CC} header, except if this header is
@samp{never}.

@end table


@node GNU Free Documentation License
@chapter GNU Free Documentation License
@include doclicense.texi

@node Index
@chapter Index
@printindex cp

@node Key Index
@chapter Key Index
@printindex ky

@bye

@c End: