Max Inden

Q: Why not send PMTUD probe in coalesced packet?

1

Currently we don't send a PMTUD probe in a coalesced packet: https://github.com/mozilla/neqo/blob/ea6cde436bcf67e71deba0809978312fbf3ee734/neqo-transport/src/connection/mod.rs#L2456-L2463 Why is that? Do we, for the sake of stability, not send probes in initial and handshake space,...

QUIC datagram outbound backpressure

2

Neqo currently has a default queue size for outbound QUIC datagrams of `10`: https://github.com/mozilla/neqo/blob/f1df4233212d9707572e799f7b912074e40b6290/neqo-transport/src/connection/params.rs#L42 When exceeding that queue size, `neqo-transport` drops at the head of the queue: https://github.com/mozilla/neqo/blob/f1df4233212d9707572e799f7b912074e40b6290/neqo-transport/src/quic_datagrams.rs#L158-L167 The queue...

iOS does not ship with ``, see https://github.com/mozilla/mtu/issues/82. https://github.com/mozilla/mtu/pull/83 dropped support for the Apple OS variants, see e.g. code below: https://github.com/mozilla/mtu/blob/8329753853c5cc0b695a8b72f953f8a1b9828c24/build.rs#L47-L52 https://github.com/mozilla/mtu/blob/8329753853c5cc0b695a8b72f953f8a1b9828c24/src/lib.rs#L101-L107 We should add support for `target_os` `ios`, `tvos`...

Given a WebTransport session where a server initiates a stream to the client, but does not (yet) write to it, the client will correctly emit a `Http3ClientEvent::WebTransport(WebTransportEvent::NewSession(_))`, but in addition...

We currently implement WebTransport draft 4: https://github.com/mozilla/neqo/blob/0f4a2060047594446c81113ea9c6a09dd5b35d34/neqo-http3/src/settings.rs#L23 See corresponding settings definition in [draft 4 IANA section](https://www.ietf.org/archive/id/draft-ietf-webtrans-http3-04.html#name-http-3-settings-parameter-r). As of writing this issue, there exists [draft 13](https://datatracker.ietf.org/doc/draft-ietf-webtrans-http3/13/). Related: https://github.com/mozilla/neqo/issues/1287 See also [changelog...

`VecDeque::truncate_front` isn't stable yet. https://github.com/rust-lang/rust/issues/140667 Instead we do `VecDeque::rotate_left` followed by a `VecDeque::truncate`. https://github.com/mozilla/neqo/blob/97c3fd4f59513f84e8aeefbf9f4d488b1dfc0b2f/neqo-transport/src/send_stream.rs#L548-L550 That is surprisingly fast. https://github.com/mozilla/neqo/pull/2680 Still, once stable, we should switch to `VecDeque::truncate_front`.

When reading off a stream trough `RecvStream::read`, unsurprisingly a lot of time is spent in (1) `copy_from_slice` and (2) de-allocating the now-no-longer-needed memory. https://github.com/mozilla/neqo/blob/33ec4462d21b681df15e1647b79be05d426d8df5/neqo-transport/src/recv_stream.rs#L334-L374 Instead of the caller providing a...

https://github.com/bheisler/criterion.rs/pull/532 introduced the `--baseline-lenient` flag. Given a missing baseline with `--baseline`, hint at the `--baseline-lenient` flag.

Send window update on STREAM_DATA_BLOCKED rx?

6

If I understand the code below correctly, Quiche currently ignores incoming `StreamDataBlocked` frames, i.e. ignores if a sender signals that it is blocked on the provided stream data flow control...

udp: retry failed GSO sendmsg on EIO without GSO

2

We have multiple reports of Linux systems that might: 1. Pass our GSO support check at runtime https://github.com/quinn-rs/quinn/blob/93b6d01605147b9763ee1b1b381a6feb9fcd454e/quinn-udp/src/unix.rs#L823-L825 2. But then fail with `EIO` on the first GSO `sendmsg` https://github.com/quinn-rs/quinn/blob/93b6d01605147b9763ee1b1b381a6feb9fcd454e/quinn-udp/src/unix.rs#L345-L349...