feat!: keep the peer's close code over qmux and raw QUIC, on web-transport-trait 0.5

[kixelated]

Problem

A client loses the application close code its server sent (Session::abort(App(4011)), Request::reject(Reject::App(4011))) over WebSocket/qmux (ws://) and raw QUIC (moqt://). Instead it sees Transport("connection closed"), the close as text inside Transport(..), or its own Session(Internal). WebTransport is fine. Both bugs are upstream.

Approach

Fixed at the source:

• fix(qmux): keep the first close reason web-transport#399: qmux keeps the first close reason. The reader's EOF / WebSocket Close, a second peer close, and a local close() no longer overwrite it. accept_* / open_* / recv_datagram report it instead of a bare Closed.
• fix(web-transport-moq): report a raw QUIC peer's close code noq#11: web-transport-moq decodes a raw QUIC peer's ApplicationClosed code directly, and close() after the connection closed keeps the peer's reason.

Releasing qmux from web-transport main also ships #390's set_priority(i32), so the trait goes to 0.5 (moq-dev/web-transport#401, moq-dev/noq#13). This PR follows it:

• rs/moq-tokio/tests/close_code.rs runs the issue's three cases (abort after accept, abort then drop, reject during handshake), 5 fresh connections each over https://, moqt://, and ws://. It fails on the current pins for every moqt:// and ws:// connection.
• Pins web-transport-moq 1.3.2, qmux 0.6, web-transport-trait 0.5, web-transport-iroh 0.8, and web-transport-wasm 0.7. The set_priority impls take i32. moq-net keeps its u8 send orders and widens them at the trait boundary.

Verified with a local [patch.crates-io] on the four upstream branches merged pairwise (not committed): cargo check --workspace --all-targets passes; close_code passes; moq-net, moq-tokio, moq-uring, and moq-e2ee tests pass; clippy is clean.

Draft until the upstream crates are released (web-transport-moq is already pinned to the planned 1.3.2). Then refresh Cargo.lock and CI can build.

@moq/qmux already keeps the first close, so no JS change.

Impact

• Breaking (Rust): moq_net::web_transport_trait (re-exported) is now 0.5. SendStream::set_priority takes i32. moq_tokio::transport::Session<S> requires a trait-0.5 session. Hence dev.
• Wire: none.

Alternatives

• A qmux 0.5.2 cherry-picked from the tag would have kept this on main. The maintainer chose to release main with the trait bump.
• Widening moq-net's own send order to i32 is left for later. Nothing needs it yet.

Follow-ups

• Raw QUIC sessions in web-transport-moq still map stream RESET/STOP codes through the HTTP/3 space.

Closes #4249

(Written by Opus 5.5)

🤖 Generated with Claude Code

[kixelated]

Decisions

• The noq fork ships one 1.3.2 bundling noq#11, noq#12, and noq#13.
• web-transport-moq stops re-exporting generic.

(Written by Opus 5.5)

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-27T00:48:49.878794Z	d43956d	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d43956dfcf

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Commit the updated dependency lockfile

The manifest now requires qmux 0.6 and the new WebTransport versions, but the unchanged Cargo.lock still pins qmux 0.5.1, web-transport-trait 0.4.0, and the previous adapter versions. The required quest validation invokes cargo run --locked, and cargo check --help defines --locked as asserting that Cargo.lock remains unchanged, so just check and CI cannot resolve this workspace after these manifest changes. Update and commit Cargo.lock with the released dependencies.

AGENTS.md reference: quest/AGENTS.md:L67-L72

Useful? React with 👍 / 👎.