msrd0

Hi, thanks for your PR! I am not convinced that just not specifying a port is a good way to indicate a unix socket. Rather, I think this will lead...

Ideally, it allows to either set `port` and `listen` or `unix_socket`. But this is work. Less ideally, we keep `port` optional with some default value like 8080, and add a...

This will be fixed with the next release of hickory-dns: https://github.com/hickory-dns/hickory-dns/pull/2845, https://github.com/hickory-dns/hickory-dns/pull/2854

Should be fixed after #42

You can try using https://github.com/luckyturtledev/mstickereditor, it can also import stickers compatible with this repository and I hacked webm support (using ffmpeg) into that one a while ago.

For the version with no parameters, I used play.rust-lang.org which claims to use 1.7.1-stable (2024-09-04 eeb90cd). For the version with `trailing_comma = "Never"`, I used rustfmt 1.7.1-nightly (5aea140 2024-08-20).

I very strongly believe that forcing an exact version of tokio is a grave mistake. It will cause massive incompatibilities with users of this crate when they mix with other...

That is not true. I'm not sure about resolver 2 from edition 2021, but with previous versions the resolver always considered dev dependencies, and e.g. enabled features in release mode...

We should discuss the approach we take. "Spurious" CI failures, i.e. ones not caused by the PR itself, are very annoying, so the current situation is undesirable. I see the...

> How is that relevant to adding a Cargo.lock to this repository, as done by this PR? It means we (maintainers/contributors/ci) won't see the breaking change introduced by dependencies, but...