New comment by 0x5c on void-packages repository

https://github.com/void-linux/void-packages/issues/38034#issuecomment-1182770242

Comment:
The problem is much worse than it seems at first. The internal CCID driver is not what's new; it's quite old and there's traces of it breaking pcscd all over the web from before gnupg 2.3 (`2021-04-07`). Here's some bug reports and blog posts from 2019;
- https://github.com/LudovicRousseau/PCSC/issues/65
- https://ludovicrousseau.blogspot.com/2019/06/gnupg-and-pcsc-conflicts.html
- https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=925312

What's new is that gnupg 2.3 dropped the automatic fallback to pcscd when the internal driver fails (https://dev.gnupg.org/T4673). That fallback was what we all used since void and (most) distros never shipped any udev rules that would allow the built-in driver to work.

I think that option 1 would be fine, but would imo require at minimum a `README.void` file.

But it's unclear to me if option 3 is realistic at all, seeing how there's no official or otherwise recognised set of udev rules, and the Debian ones are so barebones that they don't even include any Yubico devices from after the Yubikey 4, like my Yubikey 5. We'd have to maintain a set of rules and that seems like a major task.
I'd say option 2 might be best.

Regarding my previous comment on plugdev/uaccess, it seems it might not be relevant since rules like the ones Debian ships appear to use a completely different permission mechanism that I don't quite comprehend. Maybe it's some weirder logind stuff that might not work here, or Debian-specific things.