kittu@SGI.COM (Krishna Swaroop Kolluri) (01/23/91)
I have a mh/mh-e question, and I wonder if someone can answer it. Do you know of a good quick hack that can rebind the D key and subsequent X in mh-e show mode to mark a message to be moved to a folder such as +archive, such that I never really delete messages, just move them to a folder that I can expunge once a month or so, using cron. I have tried the following.. in my .mh_profile file I put the following line rmmproc: /usr/local/bin/myrmm where myrmm is the following shell script #!/bin/sh refile +archive $* But that didn't work. Since refile also looks at rmmproc, it went into a loop and multiple copies of the deleted message got refiled into the archive folder. Any help would be much appreciated. kittu.. ------- Krishna Swaroop Kolluri 9U-530 kittu@sgi.com Silicon Graphics, Inc. Off:(415)335-1859
karlton@sgi.com (Phil Karlton) (01/23/91)
I have a somewhat simpler scheme for moving deleted messages into the trash folder. Since I am a 'csh' user, I have an alias defined for rmm: "refile +trash !*". This takes care of the commands issued from a shell. Since I use xmh, I also have ~/bin/rmm which contains "refile -src $* +trash". To clean things up, I have a crontab entry which deletes anything from ~Mail/trash/ which is over 2 weeks old. PK
jerry@ORA.ORA.COM (Jerry Peek) (01/25/91)
Jeff Honig wrote about a shell script of his: > ...Note that I > use mv instead of refile because I belive it will improve performance. I had a shell script that did the same thing. I set the script in my MH profile as the "rmmproc:". Around the same time, I noticed that my sequences were getting screwed up. I didn't figure it out for a while... I realized that the "refile" command will update your sequences... but "mv" doesn't. So if you use any sequences, you'll probably want to use refile instead of mv. One problem with refile, though, is that it changes the current message. With the default rmm, you can remove any messages and not change the current message... but once you make it use "refile", then "rmm" *will* change the current message. The Nutshell Handbook on MH (and the archives I just posted to comp.sources.misc) have a script called "rmmer" that I use. It gets around both of those problems. It's a big hairy script, though, and I'd be glad to hear from people who can think of better ways. --Jerry Peek, jerry@ora.com
smarks@eng.sun.COM (Stuart Marks) (01/25/91)
I had a shell script that did the same thing. I set the script in
my MH profile as the "rmmproc:". Around the same time, I noticed that my
sequences were getting screwed up. I didn't figure it out for a while...
I realized that the "refile" command will update your sequences... but
"mv" doesn't. So if you use any sequences, you'll probably want to use
refile instead of mv.
I don't understand. I have a perl script that I use as my rmmproc, and it
seems to work fine with sequences. Note: I don't use this script in place
of rmm; I let rmm invoke it through the rmmproc mechanism. rmm seems to
update the sequences before it invokes the rmmproc.
Unfortunately, refile uses the rmmproc, so I end up with an extra link to
every refiled message in the archive folder. To work around this problem,
I've added a -normmproc option to refile that causes it to use unlink(2)
instead of invoking the rmmproc. I can supply a patch for this if anyone
is interested (it's really simple).
Here's the perl script. It works on perl 3.0 patchlevel 18 (I haven't
tried it on a higher patchlevel yet; yes, I'm behind). It has the
advantage of being much faster than shell scripts, because it avoids forks
of "mhpath new" and "mv" for each message. However, it doesn't annotate
the messages as they are archived. That should be pretty simple to add.
It touches them so that "find -mtime ..." can be used to expire them
effectively. It also detects if you're rmm'ing from the archive folder,
and if you are, simply unlinks the messages.
Improvements welcome. Enjoy.
s'marks
=============================== cut here =================================
#! /bin/sh
# @(#)rmmsafe 2.1 90/07/19
#
# Implements an MH rmmproc. Moves deleted messages to the folder indicated
# by Deleted-Folder in the user's MH profile. Called with target folder as
# the working directory, and file names (message numbers) as arguments.
perl - "$@" << \EOF
# Initialize some variables to default values.
$dfolder = 'wastebasket';
$home = $ENV{'HOME'};
$path = $home . '/Mail';
$profile = $ENV{'MH'};
# Subroutine to extract a profile entry. Assumes $_ is the line from
# the profile.
sub profentry {
chop; # remove trailing newline
($junk, $_) = split(/: */, $_, 2); # split into label and value
s/ *$//; # remove trailing whitespace
return $_;
}
# If the MH environment variable isn't set, use $HOME/.mh_profile instead.
if ( ! $profile ) {
$profile = $ENV{'HOME'} . '/.mh_profile';
}
# Scan the profile for some data: mail path and deleted-folder.
if (open(PROFILE, $profile)) {
while (<PROFILE>) {
if (/^Path:/) {
$path = do profentry();
# If mail path doesn't begin with a /, prepend $HOME.
if ( $path !~ m#^/# ) {
$path = $home . '/' . $path;
}
next;
}
if (/^Deleted-Folder:/) {
$dfolder = do profentry();
next;
}
}
close(PROFILE);
}
# Absolute pathname of the delete-folder.
$dfolderpath = $path . '/' . $dfolder;
die "rmmsafe: folder +$dfolder not found\n"
unless -d $dfolderpath;
# Get timestamp for touching files.
$now = time;
# Get device and inode numbers of current directory and the deleted-folder.
# If cwd is the deleted-folder, really remove messages. Otherwise, move
# messages from here to the deleted-folder.
($curdev, $curino, $junk) = stat('.');
($deldev, $delino, $junk) = stat($dfolderpath);
if ( $curdev == $deldev && $curino == $delino ) {
unlink @ARGV;
} else {
# Scan the deleted-folder and look for the highest numbered file.
# Non-numeric filenames are treated as zero by perl.
$next = 0;
opendir(DIR, $dfolderpath)
|| die "rmmsafe: can't open $dfolderpath\n";
while ($_ = readdir(DIR)) {
$next = $_
if ($_ > $next);
}
# For each named message, touch it, and move it to the next
# available message number in the deleted-folder.
foreach $msg (@ARGV) {
++$next;
# print "rename $msg $dfolderpath/$next\n";
utime $now, $now, $msg;
rename($msg, $dfolderpath . '/' . $next) ||
die "rmmsafe: can't move $msg to $dfolderpath/$next\n";
}
}
EOFjerry@ORA.ORA.COM (Jerry Peek) (01/25/91)
I wrote, and Stuart Marks replied: > I realized that the "refile" command will update your sequences... but > "mv" doesn't. So if you use any sequences, you'll probably want to use > refile instead of mv. > > I don't understand. I have a perl script that I use as my rmmproc, and it > seems to work fine with sequences. Note: I don't use this script in place > of rmm; I let rmm invoke it through the rmmproc mechanism. rmm seems to > update the sequences before it invokes the rmmproc. Stuart seems to be right. I've been using MH since the early 1980s... I "found the problem" in rmm a long time ago and was sure of it then. I can't reproduce it now, though, and my old rmmproc script is long gone. I'll look into this some more and straighten it out. Thanks, Stuart. --Jerry Peek, jerry@ora.com