mh.rf

<B>Digital的Unix操作系统VAX 4.2源码</B>

.\" .(f
.\" \** Note that in this case,
.\" \fIsend\fR consults the profile entry for whatever name it was invoked as,
.\" such as \fIsendf\fR.
.\" .)f
.pp
By using \fIpush\fR, the user can free the shell to do other things,
because it appears to the shell that the \fIMH\fR command has finished.
As a result the shell will immediately prompt for another command,
despite the fact that the command is really still running.
Note that if the user indicates that annotations are to be performed
(with `\-annotate' to \fIdist\fR, \fIforw\fR, or \fIrepl\fR),
the annotations will be performed after the message has been
successfully sent.
This action will appear to occur asynchronously.
Obviously, if one of the messages that is to be annotated is
removed before the draft has been successfully sent,
then when \fIMH\fR tries to make the annotations,
it won't be able to do so.
In previous versions of \fIMH\fR,
this resulted in an error message mysteriously appearing on the user's
terminal.
In \fImh.5\fR and later versions,
in this special circumstance, no error will be generated.
.pp
If send is \fIpush\fR\0'd,
then the `\-forward' switch is examined if a failure notice is generated.
If given,
then the draft is forwarded with the failure notice sent to the user.
This allows rapid \fIburst\fR\0'ing of the failure notice to retrieve the
unsent draft.

.UH "Options at What now? Level"
.pp
By default,
the message composition programs call a program called \fIwhatnow\fR before
the initial draft composition.
The \fIMH\fR user can specify any program for this.
Following is some information about the default \*(lqWhat now?\*(rq level.
More detailed information can be found in the \fIwhatnow\fR\0(1) manual entry.
.pp
When using the \fIcomp\fR, \fIdist\fR, \fIforw\fR, and \fIrepl\fR commands at
\*(lqWhat now?\*(rq level,
the \fIedit\fR, \fIlist\fR, \fIheaders\fR, \fIrefile\fR,
and (for the \fIdist\fR and \fIrepl\fR commands) the \fIdisplay\fR options
will pass on any additional arguments given them to whatever program they
invoke.
.pp
In \fImh.1\fR (the original RAND \fIMH\fR\0)
and \fImh.2\fR (the first UCI version of \fIMH\fR\0),
\fIMH\fR used a complicated heuristic to determine if the draft should be
deleted or preserved after an unsuccessful edit.
In \fImh.3\fR,
\fIMH\fR was changed to preserve the draft always,
since \fIcomp\fR, et. al.,
could usually look at a draft, apply another set of heuristics,
and decide if it was important or not.
With the notion of a \fIdraft folder\fR, 
in which one by default gets a `new' message draft,
the edit deletion/preservation algorithm was re-implemented,
to keep the draft folder from being cluttered with aborted edits.
.pp
Also,
note that by default,
if the draft cannot be successfully sent,
these commands return to \*(lqWhat now?\*(rq level.
But, when \fIpush\fR is used, this does not happen (obviously).
Hence,
if these commands were expected to annotate any messages,
this will have to be done by hand, later on, with the \fIanno\fR command.
.pp
Finally, if the `\-delete' switch is not given to the \fIquit\fR option,
then these commands will inform the user of the name of the unsent draft file.

.UH "Digests"
.pp
The \fIforw\fR command has the beginnings of a digestifying facility,
with the `\-digest\ list', `\-issue\ number', and `\-volume\ number' switches.

If \fIforw\fR is given \*(lqlist\*(rq to the `\-digest' switch
as the name of the discussion group,
and the `\-issue\ number' switch is not given,
then \fIforw\fR looks for an entry in the user's \fIMH\fR context called
\*(lq\fIdigest\fR\-issue\-list\*(rq and increments its value to use as the
issue number.
Similarly,
if the `\-volume\ number' switch is not given,
then \fIforw\fR looks for \*(lq\fIdigest\fR\-volume\-list\*(rq
(but does not increment its value) to use as the volume number.

Having calculated the name of the digest and the volume and issue numbers,
\fIforw\fR will now process the components file using the same format string
mechanism used by \fIrepl\fR.
The current `%'\-escapes are:

.nf
.ta \w'escape  'u +\w'integer  'u
\fIescape\fR	\fItype\fR	\fIsubstitution\fR
digest	string	digest name
msg	integer	issue number
cur	integer	volume number
.re
.fi

In addition, to capture the current date,
any of the escapes valid for \fIdp\fR\0(8) are also valid for \fIforw\fR.

The default components file used by \fIforw\fR when in digest mode is:

.nf
.in +.5i
.ne 10
.eo
.so @(MHETCPATH)/digestcomps
.ec
.in -.5i
.fi

Hence, when the `\-digest' switch is present,
the first step taken by \fIforw\fR is to expand the format strings in the
component file.
The next step is to compose the draft using
the standard digest encapsulation algorithm
(even putting an \*(lqEnd of list Digest\*(rq trailer in the draft).
Once the draft is composed by \fIforw\fR,
\fIforw\fR writes out the volume and issue profile entries for the digest,
and then invokes the editor.

Naturally, when composing the draft,
\fIforw\fR will honor the `\-filter\ filterfile' switch,
which is given to \fImhl\fR to filter each message being forwarded prior to
encapsulation in the draft.
A good filter file to use, which is called \fImhl.digest\fR, is:

.nf
.in +.5i
.ne 10
.eo
.so @(MHETCPATH)/mhl.digest
.ec
.in -.5i
.fi

.uh "FOLDER HANDLING"
.pp
There are two interesting facilities for manipulating folders:
relative folder addressing,
which allows a user to shorten the typing of long folder names;
and
the folder\-stack,
which permits a user to keep a stack of current folders.

.UH "Relative Folder Addressing"
.pp
By default, when `+folder' is given,
and the folder name is not absolute
(does not start with \fB/\fR, \fB\&./\fR, or \fB\&.\&./\fR),
then the UNIX pathname of the folder is interpreted relative to the user's
\fIMH\fR directory.
Although this mechanism works fine for top\-level folders and their immediate
sub\-folders,
once the depth of the sub\-folder tree grows, it becomes rather unwieldly:

.ti +.5i
scan\0+mh/mh.4/draft/flames

is a lot of typing.
\fIMH\fR can't do anything if the current folder was \*(lq+inbox\*(rq,
but if the current folder was, say, \*(lq+mh/mh.4/draft\*(rq,
\fIMH\fR has a short\-hand notation to reference a sub\-folder of the
current folder.
Using the `@folder' notation,
the \fIMH\fR user can direct any \fIMH\fR program which expects a `+folder'
argument to look for the folder relative to the current folder instead of the
user's \fIMH\fR directory.
Hence, if the current folder \fIwas\fR \*(lq+mh/mh.4/draft\*(rq,
then

.ti +.5i
scan\0@flames

would do the trick handily.
In addition, if the current folder \fIwas\fR \*(lq+mh/mh.4/draft\*(rq,

.ti +.5i
scan\0@../pick

would scan the folder \*(lq+mh/mh.4/pick\*(rq,
since, in the UNIX fashion,
it references the folder \*(lqpick\*(rq which is a sub\-folder of
the folder that is the parent of the current folder.
Since most advanced \fIMH\fR users seem to exhibit a large degree of locality
in referencing folders when they process mail,
this convention should receive a wide range of uses.

.UH "The Folder\-Stack"
.pp
The \fIfolder\-stack\fR mechanism in \fIMH\fR gives the \fIMH\fR user a
facility similar to the \fICShell\fR\0's directory\-stack.
Simply put,

.ti +.5i
folder\0\-push\0+foo

makes \*(lqfoo\*(rq the current folder,
saving the folder that was previously the current folder on the
\fIfolder\-stack\fR.
As expected,

.ti +.5i
folder\0\-pop

takes the top of the \fIfolder\-stack\fR and makes it the current folder.
Each of these switches lists the \fIfolder\-stack\fR when they execute.
It is simple to write a \fIpushf\fR command as a shell script.
It's one line:

.ti +.5i
exec\0folder\0\-push\0$@

Probably a better way is to link \fIfolder\fR to the $HOME/bin/ directory under
the name of \fIpushf\fR and then add the entry

.ti +.5i
pushf:\0\-push

to the \&.mh\(ruprofile.
.pp
The manual page for \fIfolder\fR discusses the analogy between the
\fICShell\fR directory stack commands and the switches in \fIfolder\fR which
manipulate the \fIfolder\-stack\fR.
The \fIfolder\fR command uses the context entry `Folder\-Stack:' to keep
track of the folders in the user's stack of folders.
\"
\"              On to the Appendices
\"
.fo ''-%-''
.he ''''
.(x
.sp
Appendix
.)x _
.de $c                          \" Major Heading printer
.ce
Appendix \\n+(ch
.sp 2p
.ce
.b "\\s12\\$1\\s0"              \" 12 Point Bold Header
.(x
\ \ \ \\n(ch.\\ \\ \\$2
.)x
.sp 45p				\" 45 points or about 1/2 inch
..
.++ A
.bp
.$c "COMMAND SUMMARY" "Command Summary"
.po -.50i
.so mh-chart.me
.po +.50i
.if t \{
.ll 32P
.lt 32P
\}
.bp
.$c "MESSAGE NAME BNF" "Message Name BNF"

.nf
.in 1i
.ta \w'signed-number 'u +\w':=  'u +\w'user-defined-sequence  'u
msgs	:=	msgspec	|
		msgs msgspec

msgspec	:=	msg	|
		msg-range	|
		msg-sequence	|
		user-defined-sequence

msg	:=	msg-name	|
		<number>

msg-name	:=	\*(lqfirst\*(rq	|
		\*(lqlast\*(rq	|
		\*(lqcur\*(rq	|
		\*(lq\&.\*(rq	|
		\*(lqnext\*(rq	|
		\*(lqprev\*(rq

msg-range	:=	msg\*(lq-\*(rqmsg	|
		\*(lqall\*(rq

msg-sequence	:=	msg\*(lq:\*(rqsigned-number

signed-number	:=	\*(lq+\*(rq<number>	|
		\*(lq-\*(rq<number>	|
		<number>
.re
.fi
.sp
.lp
Where <number> is a decimal number greater than zero.
.lp
Msg-range specifies all of the messages in the given range
and must not be empty.
.lp
Msg-sequence specifies up to <number> of messages, beginning
with \*(lqmsg\*(rq (in the case of first, cur, next, or <number>),
or ending with \*(lqmsg\*(rq (in the case of prev or last).
+<number> forces \*(lqstarting with msg\*(rq, and \-<number> forces
\*(lqending with number\*(rq.
In all cases, \*(lqmsg\*(rq must exist.
.lp
User\-defined sequences are defined and manipulated with the \fIpick\fR
and \fImark\fR commands.
.in 0
.bp
.ce
.b \\s12REFERENCES\\s0
.(x
.sp
REFERENCES
.)x
.sp 3
.in .4i
.ti 0
1.  Crocker, D. H., J. J. Vittal, K. T. Pogran, and D. A. Henderson, Jr.,
\*(lqStandard for the Format of ARPA Network Text Messages,\*(rq
\fIRFC733\fR,
November 1977.

.ti 0
2.  Thompson, K., and D. M. Ritchie, \*(lqThe UNIX Time-sharing System,\*(rq
\fICommunications of the ACM\fR, Vol. 17, July 1974, pp. 365-375.

.ti 0
3.  McCauley, E. J., and P. J. Drongowski, \*(lqKSOS\-The Design of a Secure
Operating System,\*(rq \fIAFIPS Conference Proceedings\fR,
National Computer Conference,
Vol. 48, 1979, pp. 345-353.

.ti 0
4.  Crocker, David H., \fIFramework and Functions of the \*(lqMS\*(rq Personal
Message System\fR, The RAND Corporation, R-2134-ARPA, December 1977.

.ti 0
5.  Thompson, K., and D. M. Ritchie, \fIUNIX Programmer's Manual\fR, 6th ed.,
Western Electric Company, May 1975 (available only to UNIX licensees).

.ti 0
6. Crocker, D. H.,
\*(lqStandard for the Format of ARPA Internet Text Messages,\*(rq
\fIRFC822\fR,
August 1982.
.de $c
.ce
.b "\\s12\\$1\\s0"   \" 12 Point Bold Header
.(x y
.sp
\\$1
.)x
.sp 3
..
.++ P
.bp 1
.fo ''''
.he ''-%-''
.+c "READ THIS"
.pp
Although the \fIMH\fR system was originally developed by the RAND Corporation,
and is now in the public domain,
the RAND Corporation assumes no responsibility for \fIMH\fR
or this particular version of \fIMH\fR.
.pp
In addition,
the Regents of the University of California issue the following
\fBdisclaimer\fR in regard to the UCI version of \fIMH\fR:
.sp 1
.in +.5i
\*(lqAlthough each program has been tested by its contributor,
no warranty, express or implied,
is made by the contributor or the University of California,
as to the accuracy and functioning of the program
and related program material,
nor shall the fact of distribution constitute any such warranty,
and no responsibility is assumed by the contributor
or the University of California in connection herewith.\*(rq
.in -.5i
.pp
This version of \fIMH\fR is in the public domain,
and as such,
there are no real restrictions on its use.
The \fIMH\fR source code and documentation have no licensing restrictions
whatsoever.
As a courtesy,
the authors ask only that you provide appropriate credit to the RAND
Corporation and the University of California for having developed the software.
.pp
\fIMH\fR is a software package that is supported neither by the RAND
Corporation nor the University of California.
However,
since we do use the software ourselves and plan to continue using
(and improving) \fIMH\fR,
bug reports and their associated fixes should be reported back to us so that
we may include them in future releases.
The current computer mailbox for \fIMH\fR is \fBBug\[e-mail]\fR
(in the ARPA Internet),
and \fB...!ucbvax!ucivax!bug\-mh\fR (UUCP).
Presently,
there are two Internet discussion groups, \fBMH\[e-mail]\fR
and \fBMH\[e-mail]\fR.  \fBMH\-Workers\fP is for people
discussing code changes to \fIMH\fP.  \fBM