> By its own admission, the project says: "The reliability layer is a pretty naive sliding window implementation". A bit scary if you really want to have guaran

> By its own admission, the project says: "The reliability layer is a pretty naive sliding window implementation". A bit scary if you really want to have guarantee message is delivered. A "naive sliding window implementation" is the basis of reliable delivery in TCP. I think all they're saying here is that they don't have concept like selective ACKs. But SACKs are basically an optimization, not an issue of correctness.