Oskar Mansfeld
Oskar Mansfeld
This initial draft PR just scopes the update from our implementation to RFC 9438 in code comments, updates RFC references and doesn't change any code yet. I mostly did this...
For #2858 it would make sense to introduce some metrics for the slow start phase that can be tracked in Firefox telemetry. I'm thinking of the following: - `slow_start_exit_ratio`: ratio...
As discussed in #2535 From the RFC: > A QUIC implementation can easily determine a spurious fast retransmit if a QUIC packet is acknowledged after it has been marked as...
- Adds `CongestionControlStats` struct with `congestion_events` and `spurious_congestion_events` stats. - Passes `CongestionControlStats` down to the congestion controller. - Adds mechanism to detect spurious congestion events by logging lost packet numbers...
This implements changing CUBIC_ALPHA to `1.0` when `w_est >= cwnd_prior` as per . - added `self.alpha` field - added `self.cwnd_prior` field - existing functions now use `self.alpha` instead of constant...
We used to only increase `w_est` by multiples of `max_datagram_size`, thus not using up all of the `new_acked_bytes`. Therefore we had to calculate how many bytes were actually used, subtract...
All the main changes to CUBIC have been merged in #2967, but we do have some leftovers. Both of these draft PRs are review-ready in theory but will need some...
> There are a bunch of `f64` values in this code (not necessarily new with this PR, also before) that are inherently integers. For example, everything measured in bytes, etc....
> Before we start reviewing/merging PRs that change behavior, I'd like to discuss better (CI) testing for CC, and getting that merged first. > This isn't (only) about making sure...
https://github.com/mozilla/neqo/blob/67ad82d7ae7a72e67034ab285854522918a8b0af/neqo-transport/src/cc/cubic.rs#L276-L280 The above function can be changed to `fn on_app_limited(&mut self) {}` and all tests still pass. As per [the RFC](https://datatracker.ietf.org/doc/html/rfc9438#name-behavior-for-application-li): >CUBIC does not increase its congestion window if a...