Windows support: dual-stack IPv4/IPv6 sockets, setup.bat, and just dev#1732
Merged
Conversation
Quinn and the relay's axum HTTP/HTTPS listeners bind a single `[::]` socket and rely on the OS to route both address families. Linux does, but Windows defaults IPV6_V6ONLY to on, so an IPv6 socket silently drops all IPv4 traffic. A relay on `[::]` is then unreachable over 127.0.0.1, and a client on `[::]` cannot dial an IPv4 server (the existing pick_addr IPv4-mapping only works on a dual-stack socket). Clear IPV6_V6ONLY before binding via socket2. A new bind_udp helper backs the quinn and noq client/server sockets; a new public bind_tcp helper backs the relay and CLI HTTP/HTTPS listeners via axum_server::from_tcp. Verified end-to-end over IPv4 (127.0.0.1) on Windows: HTTP cert fetch and a QUIC publish/subscribe round-trip both succeed (relay logs the client as ::ffff:127.0.0.1). Both were connection-refused before. Public API: adds moq_native::bind_tcp (additive). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add a winget-based setup.bat and a Windows setup page (registered in the sidebar and linked from the setup index). The script installs the Rust + JS toolchain, the MSVC C++ Build Tools, the ffmpeg CLI used by the demo, and gh/uv. Guard the demo free-port picker so it falls back to the start port when lsof is unavailable (Git Bash on Windows). With ffmpeg present and run from Git Bash, `just dev` now works end-to-end on Windows: relay, ffmpeg publish, and the web UI at localhost:5173. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
WalkthroughThis PR adds dual-stack IPv4/IPv6 socket binding utilities ( 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches✨ Simplify code
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
setup.bat (1)
1-73:⚠️ Potential issue | 🟡 Minor | ⚡ Quick winUse CRLF line endings for batch reliability on Windows.
This
.batfile is currently LF-only; convert it to CRLF to avoid cmd parsing edge cases on Windows.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@setup.bat` around lines 1 - 73, The setup.bat file is currently using LF (Unix-style) line endings, which can cause parsing issues on Windows when the script is executed. Convert all line endings throughout the entire setup.bat file from LF to CRLF (Windows-style carriage return + line feed). This can be done using a text editor's line ending conversion feature (most modern editors like VS Code have this functionality) or by running a command-line tool designed for this purpose. Ensure the entire file, from the opening `@echo` off to the closing endlocal, uses CRLF line endings consistently.Source: Linters/SAST tools
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@rs/moq-native/src/util.rs`:
- Around line 118-125: The bind_udp_ipv6_is_dual_stack test currently assumes
unconditional IPv6 and only_v6 socket option support, which creates CI flakiness
on platforms without these capabilities. Since bind_udp is intentionally
best-effort, modify the test to gracefully handle cases where IPv6 binding or
reading the only_v6 socket option is not supported. Use conditional logic (such
as skip or early return) to handle failures in the bind_udp call at line 122 or
the only_v6call at line 124, rather than panicking on unsupported platforms.
---
Outside diff comments:
In `@setup.bat`:
- Around line 1-73: The setup.bat file is currently using LF (Unix-style) line
endings, which can cause parsing issues on Windows when the script is executed.
Convert all line endings throughout the entire setup.bat file from LF to CRLF
(Windows-style carriage return + line feed). This can be done using a text
editor's line ending conversion feature (most modern editors like VS Code have
this functionality) or by running a command-line tool designed for this purpose.
Ensure the entire file, from the opening `@echo` off to the closing endlocal, uses
CRLF line endings consistently.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 2bf90321-3a82-43fc-ba06-b6c686f6eabd
⛔ Files ignored due to path filters (1)
Cargo.lockis excluded by!**/*.lock
📒 Files selected for processing (12)
demo/justfiledoc/.vitepress/config.tsdoc/setup/index.mddoc/setup/windows.mdrs/moq-cli/src/web.rsrs/moq-native/Cargo.tomlrs/moq-native/src/lib.rsrs/moq-native/src/noq.rsrs/moq-native/src/quinn.rsrs/moq-native/src/util.rsrs/moq-relay/src/web.rssetup.bat
Comment on lines
+118
to
+125
| #[test] | ||
| fn bind_udp_ipv6_is_dual_stack() { | ||
| // An IPv6 wildcard bind should come back dual-stack so IPv4 traffic | ||
| // reaches it. socket2 lets us read the option back to confirm. | ||
| let socket = bind_udp("[::]:0".parse().unwrap()).unwrap(); | ||
| let socket = socket2::Socket::from(socket); | ||
| assert!(!socket.only_v6().unwrap(), "IPv6 socket should be dual-stack"); | ||
| } |
Contributor
There was a problem hiding this comment.
Harden the IPv6 dual-stack test for platform capability variance.
bind_udp is intentionally best-effort, but Line [122] and Line [124] currently require unconditional IPv6 + only_v6 support, which can fail on valid environments and create CI flakiness.
Suggested adjustment
#[test]
fn bind_udp_ipv6_is_dual_stack() {
- let socket = bind_udp("[::]:0".parse().unwrap()).unwrap();
+ let socket = match bind_udp("[::]:0".parse().unwrap()) {
+ Ok(socket) => socket,
+ Err(err)
+ if matches!(
+ err.kind(),
+ std::io::ErrorKind::AddrNotAvailable | std::io::ErrorKind::Unsupported
+ ) =>
+ {
+ return; // host/runner doesn't support this capability
+ }
+ Err(err) => panic!("unexpected bind failure: {err}"),
+ };
let socket = socket2::Socket::from(socket);
- assert!(!socket.only_v6().unwrap(), "IPv6 socket should be dual-stack");
+ if let Ok(only_v6) = socket.only_v6() {
+ assert!(!only_v6, "IPv6 socket should be dual-stack");
+ }
}📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| #[test] | |
| fn bind_udp_ipv6_is_dual_stack() { | |
| // An IPv6 wildcard bind should come back dual-stack so IPv4 traffic | |
| // reaches it. socket2 lets us read the option back to confirm. | |
| let socket = bind_udp("[::]:0".parse().unwrap()).unwrap(); | |
| let socket = socket2::Socket::from(socket); | |
| assert!(!socket.only_v6().unwrap(), "IPv6 socket should be dual-stack"); | |
| } | |
| #[test] | |
| fn bind_udp_ipv6_is_dual_stack() { | |
| // An IPv6 wildcard bind should come back dual-stack so IPv4 traffic | |
| // reaches it. socket2 lets us read the option back to confirm. | |
| let socket = match bind_udp("[::]:0".parse().unwrap()) { | |
| Ok(socket) => socket, | |
| Err(err) | |
| if matches!( | |
| err.kind(), | |
| std::io::ErrorKind::AddrNotAvailable | std::io::ErrorKind::Unsupported | |
| ) => | |
| { | |
| return; // host/runner doesn't support this capability | |
| } | |
| Err(err) => panic!("unexpected bind failure: {err}"), | |
| }; | |
| let socket = socket2::Socket::from(socket); | |
| if let Ok(only_v6) = socket.only_v6() { | |
| assert!(!only_v6, "IPv6 socket should be dual-stack"); | |
| } | |
| } |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@rs/moq-native/src/util.rs` around lines 118 - 125, The
bind_udp_ipv6_is_dual_stack test currently assumes unconditional IPv6 and
only_v6 socket option support, which creates CI flakiness on platforms without
these capabilities. Since bind_udp is intentionally best-effort, modify the test
to gracefully handle cases where IPv6 binding or reading the only_v6 socket
option is not supported. Use conditional logic (such as skip or early return) to
handle failures in the bind_udp call at line 122 or the only_v6call at line 124,
rather than panicking on unsupported platforms.
Move the dual-stack socket helpers out of the internal `util` module into a new `pub mod bind`, so callers read `moq_native::bind::tcp` / `bind::udp` and the crate-internal sockets use `crate::bind::udp`. `udp` is now public alongside `tcp`. Also drops the public-to-private doc links the flat `bind_tcp` re-export carried. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
kixelated
enabled auto-merge (squash)
June 14, 2026 19:21
kixelated
disabled auto-merge
June 14, 2026 19:24
Merged
kixelated
added a commit
that referenced
this pull request
Jun 14, 2026
PR #1732 added an autolink to http://localhost:4443/certificate.sha256 in setup/windows.md, which VitePress can't reach at build time, failing the doc build (and every PR's `just ci`). Generalize the existing localhost exemption from a single :5173 literal to a regex covering any localhost URL. Folded in here so this PR is green on its own; supersedes #1734. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Two related changes that get MoQ building, running, and developing on Windows.
1. Dual-stack IPv6 sockets (the bug)
Quinn and the relay's axum HTTP/HTTPS listeners bind a single
[::]socket and rely on the OS to route both address families. Linux does, but Windows defaultsIPV6_V6ONLYto on, so an IPv6 socket silently drops all IPv4 traffic. The result on Windows:[::]is unreachable over127.0.0.1(both QUIC/UDP and the cert/WebSocket HTTP listener), and[::]cannot dial an IPv4 server (the existingpick_addrIPv4-mapping only works on a dual-stack socket).Fix: clear
IPV6_V6ONLYbefore binding, viasocket2. The helpers live in a newpub mod bind:bind::udpbacks the quinn and noq client/server sockets (crate::bind::udp).bind::tcpbacks the relay and CLI HTTP/HTTPS listeners throughaxum_server::from_tcp.2. Windows onboarding
setup.batat the repo root: awinget-based installer for the Rust + JS toolchain, the MSVC C++ Build Tools, theffmpegCLI (demo media), andgh/uv. Idempotent.doc/setup/windows.md(registered in the sidebar, linked from the setup index).lsofbehindcommand -vand falls back to the start port when it is missing (Git Bash ships nolsof). This unblocksjust dev.Why
Out of the box on Windows:
cargo buildworked only afterrustup update(deps need a recent stable), the relay/clients couldn't talk over IPv4, andjust devfailed onlsof+ the missing POSIX shell.Verification (Windows 11, MSVC toolchain)
moq-native+moq-relaytests pass (the 16auth::testsfailures are pre-existing and reproduce on the base tree: a Windows temp-path-as-URL fixture issue, unrelated to this change).cargo docwith-D warningsis clean onmoq-native.127.0.0.1): HTTP cert fetch and a QUIC publish/subscribe clock round-trip both succeed; the relay logs the client as::ffff:127.0.0.1. Both were connection-refused before.just devfrom a Git Bash terminal runs the relay, publishes Big Buck Bunny via ffmpeg, and serves the web UI atlocalhost:5173(playback confirmed).Reviewer notes
moq_native::bindmodule withbind::tcpandbind::udp(additive, non-breaking). Targetsmainper the branch rules (bug fix + additive).cargo clippyon this machine flags ~20 pre-existingresult_large_errlints inquinn.rs/tls.rs; they come from clippy 1.96 being newer than the Nix-pinned toolchain, not from this change. None are in the new code.setup.batwas run on a real machine; thejust devchain was smoke-tested end to end.(Written by Claude)