Package Details: postsrsd 2.0.10-1

Git Clone URL: https://aur.archlinux.org/postsrsd.git (read-only, click to copy)
Package Base: postsrsd
Description: Provides the Sender Rewriting Scheme (SRS) via TCP-based lookup tables for Postfix
Upstream URL: https://github.com/roehling/postsrsd
Licenses: GPL2
Submitter: fordprefect
Maintainer: fordprefect
Last Packager: fordprefect
Votes: 3
Popularity: 0.013926
First Submitted: 2016-03-30 16:15 (UTC)
Last Updated: 2024-07-23 08:33 (UTC)

Latest Comments

« First ‹ Previous 1 2 3 4 5 6 Next › Last »

fordprefect commented on 2020-11-02 14:28 (UTC)

@Thaodan: Thanks for your suggestions. However, I prefer explanations over plain diff dumps. I see why we'd want to set systemd in the cmake flags, but why install an empty directory?

Thaodan commented on 2020-11-02 03:25 (UTC) (edited on 2020-11-02 03:25 (UTC) by Thaodan)

Needed to patch the pkgbuild by this to build in a clean-chroot: The indendation is not important but forcing the init systemd detection and the install.

diff --git a/PKGBUILD b/PKGBUILD
index 19c9507..eafd6af 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -37,7 +37,11 @@ build() {
   [[ -d build ]] && rm -rf build
   mkdir build
   cd build
-  cmake .. -DCMAKE_INSTALL_PREFIX=/usr -DCMAKE_BUILD_TYPE=Release -DGENERATE_SRS_SECRET=OFF -DCONFIG_DIR=/etc/postsrsd
+  cmake .. -DCMAKE_INSTALL_PREFIX=/usr \
+        -DCMAKE_BUILD_TYPE=Release \
+        -DGENERATE_SRS_SECRET=OFF \
+        -DCONFIG_DIR=/etc/postsrs \
+        -DINIT_FLAVOR=systemd
   make all
 }

@@ -49,6 +53,7 @@ package() {

   #rm -rf $pkgdir/usr/lib
   mv "$pkgdir/usr/sbin" "$pkgdir/usr/bin"
+  install -m755 -d "$pkgdir/etc/postsrsd"
   sed -e 's/^\(RUN_AS=\)nobody/#\1postsrsd/;s/\(\/etc\/postsrsd\)\(\.secret\)/\1\/postsrsd\2/' \
       -e 's/^\(# is \).*$/\1localhost\.localdomain/'< postsrsd.default > "$pkgdir/etc/postsrsd/postsrsd"

jonhoo commented on 2019-07-02 15:44 (UTC) (edited on 2019-07-02 15:45 (UTC) by jonhoo)

It looks like sysusers.d is broken again. I now get:

Jul 02 10:23:17 vm-mx systemd-sysusers[270]: Failed to parse UID: 'mail': Invalid argument

I think the issue is that you must give a numeric UID. You can't give the name of an existing user. It looks like - is the value to use for automatic assignment, which is probably what we want.

fordprefect commented on 2019-04-30 06:57 (UTC)

@Jonhoo: you are right, we need type u here. Thank you.

jonhoo commented on 2019-04-29 18:51 (UTC)

It looks like the sysusers.d file doesn't have the right format? I get:

[/usr/lib/sysusers.d/postsrsd.conf:2] Lines of type 'm' don't take a GECOS field.

Removing the GECOS field seems to make things work just fine

fordprefect commented on 2019-03-06 13:02 (UTC)

@bertptrs: thanks for finding the fix before I even had time to start searching!

bertptrs commented on 2019-03-06 12:39 (UTC)

I've created my own updated PKGBUILD that fixes both of the issues, which were coincidentally caused by the same thing. It adds a single commit as a patch until upstream makes a new release. https://ptpb.pw/Bv4S

fordprefect commented on 2019-03-04 11:49 (UTC)

@bertptrs: thanks for notification. being short on time atm i just bumped the version without extensive testing. I will look into this in the coming days and push a fix.

bertptrs commented on 2019-03-04 11:45 (UTC)

Hey, I'm having trouble running this PKGBUILD. It appears that starting version 1.6, a few things have changed.

First: the directory /etc/postsrsd is no longer created by make install. This causes a problem on line 39 [1] of your PKGBUIDL, where you try to pipe into this directory. A simple mkdir -p "$pkgdir/etc/postsrsd" somewhere before that line should fix it.

Second: upstream changed their detection of systemd-based systems to something that doesn't actually work. I've notified upstream here [2].

[1] https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=postsrsd#n39 [2] https://github.com/roehling/postsrsd/issues/91

fordprefect commented on 2019-01-14 10:36 (UTC)

i omitted the bump on purpose, since it triggers a rebuild which is unneeded imo. on second thought though it corrects the location of the file, so you might be right here.