Perry Lorier

Since you asked for opinions: 1) If neli::{NlSocket,NlSocketHandle} share a function in common with neli::tokio::NlSocket, they should ideally have the same function signature and do the same thing (modulo asyncronous...

You're right, I had forgotten I convert things to a std::os::Error very early on and thought that it was neli doing that. My apologies. I played with it a little...

This looks fantastic! Thank you very much, this will make debugging much much more pleasant and much easier to understand what's going wrong. I've left a comment in the PR...

> In my redesign, I'd like to make all methods take &self which should allow you to simply wrap it in an Arc. Would this be satisfactory for you? As...

I believe that Linux only supports a message going to a single multicast group, so yes, `recvfrom()`s should only have the single multicast group. I looked a bit at https://github.com/jbaublitz/neli/blob/v0.7.0-dev...

Okay, I've finally got some time to perform a full migration. In general I think this is a good change, here's my stream of thought process as I do the...

Ah, and the next one is that `x.get_attr_handle()` →`x.attrs().get_attr_handle()`. And `NlError` appears to have just become `SocketError`. The new builder pattern means I can't pattern match on messages anymore because...

Hrm, now I'm stuck. I often take the Nlmsghdr, then extract out the payload, but I can't move the payload out nor can I clone it, I must take a...

`v0.7.0-rc1` is missing the "rt" feature for tokio (it's only part of dev-dependencies), so `tokio::spawn` doesn't exist which means the "tokio" feature of neli fails to compile from outside of...

nit: `resolve_genl_family()` returns a `Result`, but `send` requires a `NlType` as it's first argument. Could `resolve_genl_family()` instead return a `Result`? This keeps everything type safe and makes it less likely...