test(interop): the clients against real sites over the internet, and quic-interop-runner

Open
#57 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
15/100
Issue type
Feature
Clarity
Mostly clear
Activity status
Active
Tech stack
c

Research direction

The existing local harness is tests/servers/run.sh, and the clients involved are h2_client and h3_client. Start by reading those two entry points and the script to see how a run is launched and how results are reported. The work has two parts: running the clients against public origins with recorded results, and wrapping the client and server in the quic-interop-runner Docker images. Done means both runs are written up in H2H3.md with every failure fixed or listed as a known limit.

Written by the indexing model from the issue text.

Description

P0: release blocker

Every interop test so far is local: a simulated network, ngtcp2, nghttp2 and nghttp3 in memory, and real servers on 127.0.0.1 (tests/servers/run.sh). CDNs behave differently from the reference stacks (Retry under load, small MTUs, connection-ID rotation, 0-RTT policies, GOAWAY timing), and real paths add loss and reordering the simulator doesn't model.

  1. The clients against real sites over the internet: h2_client (TLS + ALPN) and h3_client against Google, Cloudflare, Fastly, Akamai, Meta and a few more large origins. Record handshake success, resumption and 0-RTT acceptance, PMTU reached, and body integrity (lengths, and hashes where known). Run it by hand or on a schedule, not on every PR, since it depends on the network.
  2. quic-interop-runner, the standard QUIC interop matrix, with our client and server in its Docker images, against at least ngtcp2, quiche, quic-go, picoquic, msquic and aioquic. Its test cases (handshake, transfer, retry, resumption, zerortt, multiplexing, keyupdate, chacha20, v2, blackhole, and so on) are the credible public evidence for a QUIC stack.

Done when: both runs are documented in H2H3.md with results, and every failure is either fixed or listed as a known limit.

Dominant language
C
Stars
0
Forks
0
Avg merge
20h 4m
Merged PRs (30d)
25

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from public/cursed-http

All issues in public/cursed-http

Similar issues

More C issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.