urkellabs / brontide Goto Github PK

According to Rust Forge: https://forge.rust-lang.org, Rust 1.36 will be out in 3 days.

At that point, our feature "Stream" will work on beta instead of just nightly. We should adjust our CI accordingly so that we are testing the stream feature on beta and nightly.

Additionally, we can probably turn allow fail nightly back on since we don't absolutely depend on it for the stream feature anymore.

https://github.com/trailofbits/siderophile

https://github.com/carllerche/loom

• License mentioning HSD and LND
• Finish up Readme w/ examples.
• More tests
• Make Brontide Stream a feature, so you can compile w/o nightly
• update a few fuzzing targets, and integrate into CI - add feature comp into CI
• Crate visibilities
• Documentation Comments
• Custom Debug types
• Specific tests in cipher -> As in what a tag should be depending on the input.
• Crate Bump and Release

right now we rely on the helper function of read_message to parse the next message in the brontide stream. I believe, that we could actually implement the same functionality into the stream trait, so that just polling the stream will return when the entire message is ready to be read, and not just potentially the header.

https://github.com/RustSec/cargo-audit

net2 crate has been deprecated; use socket2 instead

The net2 crate has been deprecated
and users are encouraged to considered socket2 instead.

See advisory page for additional details.

We now enabled async-std unstable (for racing), and so we are able to use the Stream timeout trait function.

This allows us to remove the double ?? on the timeout functions.

https://noiseprotocol.org/noise.html#security-considerations Check all of these

Re: https://blog.yoshuawuyts.com/rust-streams/

Our errors are fine for right now, but I think we can make them a lot easier to parse by breaking them into a few separate enums.

It would be good to have an enum for Act Errors, as well as one for Packet errors. We should probably just remove the Strings that are nested inside of them as well, as the error explains itself quite good enough.

Rust currently does not allow the ? shorthand to be used on options for our errors to catch NoneError. Once this following RFC is merged in: rust-lang/rust#42327, we will be able to replace our ok_or statements w/ just a simple ?.