On Sat, Dec 14, 2024 at 12:08:08AM +0100, Christoph Liebender wrote: > Hello there everyone. > > This is my first try at contributing to the OpenBSD ports tree: > > https://liebender.dev/patches/openbsd-ports/wstunnel-port.tar.gz > > wstunnel is useful in case you want to connect to your OpenBSD box, given > that > you are behind some firewall that restricts traffic to certain ports - for > example, you can configure wstunnel to listen on :443 and connect to your > box > via that port, to then route all your traffic to the wireguard interface of > your server. There are other usecases though, but this is my primary one. > > Now, there are two patches included, even though wstunnel compiles on > OpenBSD > amd64 just fine. I had to apply these because the original Cargo.toml > references a github repository as a crate: fastwebsockets. In this case, the > developer of wstunnel made changes to the original crate which they did not > (yet?) upstream, and their fork does not include any tags for the Cargo.toml > to reference. This is why I am patching Cargo.toml to reference the original > crate and then apply the changes of the wstunnel developer on top. I edited > crates.inc manually in this case to add fastwebsockets-0.8.0. Is there a > better > way? `make modcargo-gen-crates` apparently only reads the unpatched, > original > Cargo.toml. > > Anyway, I'm glad to receive any feedback or comments. Thanks. > > - Christoph
New ports should be as attachments. The user.list patch should be sent inlined. Regarding Cargo.toml, I don't know any Rust nor its ecosystem, but it seems to be possible to specify a Git hash, according to https://doc.rust-lang.org/cargo/reference/specifying-dependencies.html in the way of -fastwebsockets = { git = "https://github.com/erebe/fastwebsockets.git";, features = ["upgrade", "simd", "unstable-split"] } +fastwebsockets = { git = "https://github.com/erebe/fastwebsockets.git";, rev = "deadbad0", features = ["upgrade", "simd", "unstable-split"] } Also, this depends on rustls-0.23.19, which already uses aws-lc-rs as the cryptography backend. I believe that will give issues on IBT-enabled hardware.
