Arch Linux User Repository

@markuschaaf, the setup with rules adding 'scanner' group is package-specific. Sane's way is to add libsane_matched=yes to scanner devices and let some 'base' rule to react to this env. Particularly, /usr/lib/udev/rules.d/66-saned.rules runs /usr/bin/setfacl -m g:saned:rw $env{DEVNAME} for these.

You should probably tweak your service launchers to start under 'saned' group instead of 'scanner'.

Alternatively, add a copy of that rule in, say, /etc/udev/rules.d/66-scanner.rules with 'saned' group replaced by 'scanner'.

Note: both tweaks will leave all scanners (not just those supported by brscan4) equally accessible.

Group based permission management is deprecated in Arch and repo packages are moving away from it. It's hard to blame AUR package for doing exactly the same what repo package did. It's more a standard than a fashion. One of the points of systemd dynamic permission model is to differentiate between local and remote user permissions. Users are free to modify this behavior as they wish.

What's the purpose of using hwdb, besides "because you can"? Supporting the scanner group has a real world use case: I can use the scanner from other services beside saned and over ssh, which is what I do. Now the package is broken for me and I have to roll my own. The udev-rules I had sent @Harvey worked for everyone. I don't know what's the reason for those commenters here pushing their unjustified but fashionable ideas.

Sorry, still didn't manage to bring my Arch installation to the device.

I can say that systemd-udevd accepts the new file at least.

Ok, although I am not fully convinced yet I made up my mind to sync with sane's behavior. Pushed a new package Version with udev-hwdb support. Lets see what happens (Fingers crossed).

I think when provided udev+hwdb approach will work here then other packages will follow (if they are still maintained). Someone must go first.

@ettavolt: Did you test it? I only have networked devices here so this is rather like flying blind for me :( I had a look at your PKGBUILD. Does this work for you? Meanwhile I looked at the other brscan incarnations (2/3/5) in AUR, none of them is using udev+hwdb... I feel if we decide to go this way, at least all should do the same. And the Arch wiki should be corrected as well: https://wiki.archlinux.org/index.php/CUPS/Troubleshooting#Conflict_with_SANE

@ettavolt: Did you test it? I only have networked devices here so this is rather like flying blind for me :( I had a look at your PKGBUILD. Does this work for you? Meanwhile I looked at the other brsan incarnations (2/3/5) in AUR, none of them is using udev+hwdb...

Yes, like this: https://github.com/ettavolt/aur-brscan4/commit/3680c46f83ec7909e433cebfae4d0c92cdb98238

Let me test it though. ☺

My understanding is if you add right device id to hwdb then systemd will take care of permissions (it will grant permission for currently logged user regardless of any groups set).

I think the request from https://aur.archlinux.org/packages/brscan4/#comment-793321 was about generating .hwdb file with device ids and install it to /usr/lib/udev/hwdb.d/ instead of creating custom udev rules which grant permissions for generated device id.