Am Tue, Dec 06, 2022 at 04:27:53PM +0000 schrieb Stuart Henderson:
> On 2022/12/06 16:50, Martin Ziemer wrote:
> > Am Tue, Dec 06, 2022 at 03:42:10PM +0000 schrieb Stuart Henderson:
> > > On 2022/12/06 16:38, Martin Ziemer wrote:
> > > > Am Tue, Dec 06, 2022 at 03:24:51PM +0000 schrieb Stuart Henderson:
> > > > > On 2022/12/06 15:22, Stuart Henderson wrote:
> > > > > > I have left the getmails patch alone for now as I can't test it but
> > > > > > the
> > > > > > pgrep invocation is wrong, it should probably search for something
> > > > > > like
> > > > > > "^/bin/sh /usr/local/bin/getmails$" and then I expect the set -e
> > > > > > will
> > > > > > work.
> > > > > ...or here's an (untested) version with that proposed change.
> > > > > (sorry for the spam!)
> > > > > +-PID_GETMAILS=$(pgrep -U $UID_BY_ID '^getmails$')
> > > > > ++PID_GETMAILS=$(pgrep -U $UID_BY_ID '^/bin/sh
> > > > > /usr/local/bin/getmails$')
> > > > Just tested the getmails change: it still exits at the pgrep line.
> > > Try pgrep -f [...]
> > Does not work either.
> > The problem is a premature end of the whole script, of pgrep finds
> > noting, instead of just filling the variable to empty.
> Have a poke around with pgrep while the script is running and see
> what's needed to get it to match then; in particular this part of the
> diff breaks the whole reason they're using pgrep:
>
> +-if [ "x$PID_GETMAILS" != "x$$" ]; then
> ++if [ "x${PID_GETMAILS}x" != "xx" ]; then
Today i got my hands on a Debian and a FreeBSD system.
On Debian getmails is not distributed with the package. If i use
getmails from git there, i get the same error.
On FreeBSD getmails is distributed unpatched. It shows the same error
(and tries to start getmail from /usr/bin/getmail)
So i see 3 Options for us:
1. We do not distribute the script in the package
2. We patch getmails in a way like the diff below and install it (This
version i use on my systems)
3. We ship a patched version as example
I tend to say for the moment not shipping it would be the safest way,
until the version in upstream is better.
The original getmail had no script for multiple configuration files,
so we will get no problems with compatibility.
--- /usr/obj/ports/getmail-6.18.10/getmail6-6.18.10/getmails Sun Sep 18
19:56:20 2022
+++ /usr/local/bin/getmails Fri Dec 9 08:15:44 2022
@@ -28,8 +28,8 @@ BASE1=${1##*/}
[ "$BASE1" != "${BASE1#$2}" ] && return 0 || return 1
}
UID_BY_ID=$(id -u)
-PID_GETMAILS=$(pgrep -U $UID_BY_ID '^getmails$')
-if [ "x$PID_GETMAILS" != "x$$" ]; then
+PID_GETMAILS=$(pgrep -fU $UID_BY_ID '/usr/local/bin/getmails' | sed "s/$$//"
| tr -d '\n' )
+if [ "x${PID_GETMAILS}x" != "xx" ]; then
echo "The getmails script is already running as PID=\"$PID_GETMAILS\"
." >&2
exit 1
fi
@@ -44,7 +44,7 @@ if [ -f $getmailrcdir/stop ]; then
echo "Do not run getmail ... (if not, remove $getmailrcdir/stop)" >&2
exit 1
fi
-rcfiles="/usr/bin/getmail"
+rcfiles="/usr/local/bin/getmail"
# Address concerns raised by #863856
# emacs backup files: foo~ foo#
# vim backup files: foo~ foo.swp
@@ -57,7 +57,8 @@ if $para ; then
! endwith "$file" '#' && \
! startswith "$file" 'oldmail-' && \
! endwith "$file" '.swp' && \
- ! endwith "$file" '.bak' ; then
+ ! endwith "$file" '.bak' && \
+ [ -f "$file" ]; then
$rcfiles --rcfile "$file" "$@" &
pids="$pids $!"
fi
@@ -79,7 +80,8 @@ else
! endwith "$file" '#' && \
! startswith "$file" 'oldmail-' && \
! endwith "$file" '.swp' && \
- ! endwith "$file" '.bak' ; then
+ ! endwith "$file" '.bak' && \
+ [ -f "$file" ]; then
rcfiles="$rcfiles --rcfile \"$file\""
fi
done
