Thomas Eizinger
Thomas Eizinger
### Describe your use case We are using this in CI and it contributes about ~1-2min to the CI run of each crate despite: - `cargo semver-checks` being installed -...
Extracted out of #3391. We don't actually need this for #3391 though because we've added a compatibility layer during deserialization. But, it will be good to remove that compat layer...
Tracking issue for the plan and progress of replacing `webrtc-rs` with our own, product-tailored connectivity library. For a prior discussion, see https://github.com/firezone/firezone/issues/1962. ### Design After lots of prototyping and brainstorming...
The spec recommends checking for the username that created the allocation when it is being interacted with after (e.g. when binding a channel or refreshing it). See https://www.rfc-editor.org/rfc/rfc8656#name-general-behavior.
See https://github.com/firezone/firezone/pull/3528 for a longer discussion on this topic. There is always the brute-force method of just adding all local interfaces as host candidates but that creates a combinatorial explosion...
The allocated port is in fact a unique identifier of the allocation. The use of a separate ID is unnecessary and could be removed.
I am guessing we will eventually want to rate limit how much bandwidth a user can use. I am writing this issue so we don't forget about it. Any ideas...
The above mentioned function is currently marked `async` but doesn't use `await`: https://github.com/hickory-dns/hickory-dns/blob/d07455476cfba08a661c09f303af2cfbb8c1a5cc/crates/async-std-resolver/src/lib.rs#L141-L143 When digging deeper, @oblique found out that the underlying file IO with `/etc/resolv.conf` is using blocking access:...
#### The associated forum post URL from `https://forum.rclone.org` I didn't post to the forum as I'd assume that a segmentation fault bug should be tracked here instead. #### What is...