v0.12.1 has the wrong `/dev/` path for `enumerate_dev_nodes_blocking`

[jtroo]

Thanks for the work on this crate!

A user of my project discovered a bug in v0.12.1 where the dev node paths have changed since v0.12.0. It remains an iterator with a single entry as expected, but the single entry has an unexpected path.

https://github.com/jtroo/kanata/issues/455

The symlink which Kanata creates for the virtual output device points at /sys/devices/virtual/input/input8/event1 which is a folder with a bunch of metadata rather than the expected /dev/input/event1. The Created device /path/to/device message shows the same incorrect path.

[mattfbacon]

I traced this back to 50a33695. The commit message is not really correct because in addition to removing the loop and returning Ok(None) at the end it also changes the behavior by removing the path manipulation. I would open a PR but I think this repository is abandoned.

[jeff-hiner]

As far as I know the crate is still active. I'm still a reviewer and have merge perms, but I don't have permissions to publish. If you submit a PR I'll take a look.

[emberian]

not quite abandoned, i'll merge and release any important PRs. i've recently resumed my open source work after long sabbatical, haven't gotten back to this crate yet

[mattfbacon]

Good to know

[wolfwood]

a bisect using kanata points to d0231e3c626316b18de59eec5d1ff8a344d94c9e as the commit responsible. reverting this commit on top of 0.12.1 allows kanata to retain the previous behavior.

I've made PR #147 with the revert. Ideally though would be a 0.12.2 release that cherrypicks the revert on top of 0.12.1. I have a branch with that here: f17826fed3b3c69cbc0b262f02f0581351d47008

[coolreader18]

Released 0.12.2!

[wolfwood]

Thanks @coolreader18 ! This works in my testing.