perf(net)!: box the transport so the session code compiles once

quest/m1/README.md

@@ -129,7 +129,6 @@ tracks, so per-group stats stop costing cluster announces.
 - [Track live signal](/quest/m1/track-live.md) - a reader can wait for its track's feed to be live, so a player skips old cached groups while a fresh upstream subscription starts
 - [Producers drop the phantom catalog type](/quest/m1/catalog-phantom.md) - moq-video, moq-audio and moq-mux producers lose an `E: CatalogExt` parameter that only `PhantomData` holds
 - [JSON encoder options without a downcast](/quest/m1/json-encoder-options.md) - moq-mux's `json_snapshot` takes its `deltas` switch through a typed path any custom catalog entry can use, not an `Any` downcast
-- [moq-net compiles once](/quest/m1/net-erase-transport.md) - the session drivers stop monomorphizing per transport and per crate, saving several MiB of the relay, FFI library, `libmoq` and CLI
 - [Audio devices on ARM](/quest/m1/audio-devices-arm.md) - `just rs audio-devices` passes on the hosted ARM runner
 - [Tooling tests gate every merge](/quest/m1/gh-test-gate.md) - every queue Check runs `just gh test`, so a change to the CI selector or release tooling cannot merge without its own tests
 - [Wayland display id](/quest/m1/wayland-display-id.md) - a requested display id off X11 is refused or honored, never silently replaced by a portal pick shared across publications

quest/m1/audit-compile-time.md

@@ -24,8 +24,8 @@ incremental). The [CI workflow audit](/quest/m2/audit-ci.md) owns job-level CI
 cost and caching; the [dependency audit](/quest/m2/audit-deps.md) owns
 deduplicating and upgrading dependencies.
 
-What to check first: whether [moq-net compiles once](/quest/m1/net-erase-transport.md)
-landed. If so, rerun `cargo llvm-lines` on the moq-tokio lib and compare the
+What to check first: [moq-net compiles once](https://github.com/moq-dev/moq/pull/6176)
+landed, so rerun `cargo llvm-lines` on the moq-tokio lib and compare the
 transport share below, then rerun the thin-LTO size comparison against the
 smaller archive. Then whether `cargo-llvm-lines` joined the pinned shell.
 
@@ -137,7 +137,7 @@ Not measured: thin sizes for linux_arm64, darwin_arm64 and windows_amd64;
 build time above two jobs; wall-clock runtime.
 
 Decision (maintainer, 2026-10-20): keep fat LTO, and revisit once
-[moq-net compiles once](/quest/m1/net-erase-transport.md) shrinks the archive.
+[moq-net compiles once](https://github.com/moq-dev/moq/pull/6176) shrinks the archive.
 Thin saves about 40% of the final unit's wall time on two jobs, but release
 builds run only nightly and on release PRs, off the merge-queue gate. It costs
 10 MiB of the Go archive's headroom under a hard publish limit while that
@@ -146,7 +146,7 @@ builds join the PR gate.
 
 ### Results of the m0 run
 
-- Planned: [moq-net compiles once](/quest/m1/net-erase-transport.md) removes
+- Planned: [moq-net compiles once](https://github.com/moq-dev/moq/pull/6176) removes
   the per-transport copies, the largest attributable dev IR (43% of moq-tokio)
   and a share of the release fat-LTO module.
 - Explained, no action: the release `moq` unit is all LLVM backend work on one

quest/m1/audit-size.md

@@ -19,9 +19,9 @@ Only a change that needs a maintainer decision (public API, scope) or is
 independently landable and too large to review alongside the rest becomes a
 follow-up through `/quest-plan`.
 
-Checkpointed 2026-10-21, a day after the m0 run, with nothing to plan but
-the quests below. The run finishes after
-[moq-net compiles once](/quest/m1/net-erase-transport.md) lands, so the audit
+Checkpointed 2026-10-21, a day after the m0 run, with nothing to plan. The
+run waited for moq-net to compile once, which landed in
+[moq-dev/moq#6176](https://github.com/moq-dev/moq/pull/6176), so the audit
 before the breaking cut sees its result and the rest of m1.
 
 What to check first: whether the nightly `sizes` job (`sh/size/`) has run.
@@ -30,7 +30,7 @@ The Linux release-link fix
 2026-10-22 UTC; verify the first scheduled Nightly after it before treating
 the sizes history as available. Once it runs, its history replaces the local
 method below. Then
-re-attribute the native rows for the erasure quest, and measure `@moq/net`
+re-attribute the native rows after the erasure, and measure `@moq/net`
 with the track compression codec in `js/net`.
 
 ### Baselines
@@ -66,13 +66,14 @@ run's FFI (31.86 MiB) and CLI (48.14 MiB) figures for it do not reproduce.
   from the message parameter codec (`parameters.ts`, `filter.ts`), so a lite
   session never loads them. First load fell 5.1 to 5.5 KB gzip in every
   bundle; the same test fails if SETUP reaches `parameters.ts` again.
-- Planned: [moq-net compiles once](/quest/m1/net-erase-transport.md). On
-  7d2a91fe66, `moq-net` is 9.5 MiB of the relay's text and 7.7 MiB of the FFI
-  library's; one copy of each function would be about 2.5 MiB in both, so the
-  ceiling is about 7.0 and 5.2 MiB. In the relay, the binary crate (a 16-line
-  `main.rs`) instantiates 3.1 MiB of it, `moq-tokio` 3.2 MiB and the library
-  1.9 MiB; identical instantiations repeated across crates are 5.0 MiB. They
-  are not byte-identical after LTO: `-Wl,--icf=all` saves only 0.23 MiB.
+- Fixed: moq-net boxes the transport, so its session drivers compile once
+  instead of per transport and per instantiating crate
+  ([moq-dev/moq#6176](https://github.com/moq-dev/moq/pull/6176)). Stripped
+  profiling builds fell 31.8 to 22.5 MiB (relay), 49.1 to 42.5 MiB (CLI),
+  28.0 to 21.8 MiB (`libmoq_ffi.so`), 254.6 to 207.8 MiB (unstripped
+  `libmoq.a`) and 862 to 786 KB gzipped (`@moq/wasm`). Before it, the relay
+  binary crate, `moq-tokio` and the library each instantiated their own copy;
+  `-Wl,--icf=all` saved only 0.23 MiB, since the copies differ after LTO.
 - Planned: [the catalog publish path compiles once](/quest/m2/catalog-publish-once.md)
   (`moq-json`'s diff serializer, +0.14 MiB of the CLI this run).
 - Explained, no action: native growth since 2b28f5eac is `moq-net` features
@@ -102,10 +103,6 @@ v0 mangling suffix) on local `profiling` builds of both revisions, the wasm
 name section from a `wasm-release` build with `CARGO_PROFILE_WASM_RELEASE_STRIP=none`,
 and esbuild metafiles from `sh/size/bundles.ts` for JS.
 
-## Required
-
-- [moq-net compiles once](/quest/m1/net-erase-transport.md) - re-measure the native rows once the drivers compile once
-
 ## Related
 
 - [FFI size profile](/quest/m2/ffi-size-profile.md) - whether the bindings ship at `opt-level = "s"`

quest/m1/net-internals.md

@@ -25,7 +25,3 @@ Dropped: renaming the over-long `session_delivery_*` bench IDs, since it
 loses their history for a cosmetic gain.
 
 Public API: none. Wire: none.
-
-## Required
-
-- [moq-net compiles once](/quest/m1/net-erase-transport.md) - boxes the transport these cleanups build on

quest/m2/ietf-adapter-session-errors.md

@@ -29,7 +29,3 @@ ProtocolViolation while a uni or datagram poll is pending, and the client
 must see a terminal error and not redial.
 
 Public API: none. Wire: none.
-
-## Related
-
-- [moq-net compiles once](/quest/m1/net-erase-transport.md) - its open PR [moq-dev/moq#6176](https://github.com/moq-dev/moq/pull/6176) touches the same adapter

quest/m2/uring-transport-machines.md

@@ -3,8 +3,8 @@
 ## Goal
 
 The relay fanout benchmark measures what `moq_uring::transport::Local`'s
-per-call thread check costs on the ring once
-[moq-net compiles once](/quest/m1/net-erase-transport.md) lands, and the
+per-call thread check costs on the ring now that
+[moq-net compiles once](https://github.com/moq-dev/moq/pull/6176), and the
 check is removed only if it shows up. `transport::poll::Boxable` stays the
 single bound.
 
@@ -24,10 +24,6 @@ check, and remove it only if the difference shows.
 Public API: none unless the check goes, which removes
 `moq_uring::transport::Local`. Wire: none.
 
-## Required
-
-- [moq-net compiles once](/quest/m1/net-erase-transport.md) - boxes the transport the check wraps
-
 ## Related
 
 - [Relay io_uring packages](/quest/m2/relay-io-uring-package.md) - not blocked on this; `Local` already gives parity, and only its cost is measured here

rs/moq-audio/src/decode/consumer.rs

@@ -781,18 +781,12 @@ mod tests {
 			.with_subscriber(to.clone());
 		tokio::join!(
 			async {
-				let (session, driver) = client
-					.connect_lite(moq_net::time::now(), client_transport)
-					.await
-					.unwrap();
+				let (session, driver) = client.connect(moq_net::time::now(), client_transport).await.unwrap();
 				tokio::spawn(moq_net::time::run(driver));
 				session
 			},
 			async {
-				let (session, driver) = server
-					.accept_lite(moq_net::time::now(), server_transport)
-					.await
-					.unwrap();
+				let (session, driver) = server.accept(moq_net::time::now(), server_transport).await.unwrap();
 				tokio::spawn(moq_net::time::run(driver));
 				session
 			}

rs/moq-mux/src/container/test_util.rs

@@ -259,18 +259,12 @@ impl Link {
 		// Each driver runs as soon as its handshake resolves: the accept waits on the client's SETUP.
 		let sessions = tokio::join!(
 			async {
-				let (session, driver) = client
-					.connect_lite(moq_net::time::now(), client_transport)
-					.await
-					.unwrap();
+				let (session, driver) = client.connect(moq_net::time::now(), client_transport).await.unwrap();
 				tokio::spawn(moq_net::time::run(driver));
 				session
 			},
 			async {
-				let (session, driver) = server
-					.accept_lite(moq_net::time::now(), server_transport)
-					.await
-					.unwrap();
+				let (session, driver) = server.accept(moq_net::time::now(), server_transport).await.unwrap();
 				tokio::spawn(moq_net::time::run(driver));
 				session
 			}
@@ -304,18 +298,12 @@ impl Link {
 			.with_publisher(from.consume());
 		let sessions = tokio::join!(
 			async {
-				let (session, driver) = client
-					.connect_lite(moq_net::time::now(), client_transport)
-					.await
-					.unwrap();
+				let (session, driver) = client.connect(moq_net::time::now(), client_transport).await.unwrap();
 				tokio::spawn(moq_net::time::run(driver));
 				session
 			},
 			async {
-				let (session, driver) = server
-					.accept_lite(moq_net::time::now(), server_transport)
-					.await
-					.unwrap();
+				let (session, driver) = server.accept(moq_net::time::now(), server_transport).await.unwrap();
 				tokio::spawn(moq_net::time::run(driver));
 				session
 			}

rs/moq-net/MIGRATE.md

@@ -4,6 +4,14 @@ How to update your code for each breaking change, newest first.
 
 ## Unreleased
 
+### Sessions box their transport (#6176)
+
+`Driver` and `server::Handshake` lose their transport type parameter: write `moq_net::Driver` and `moq_net::server::Handshake`.
+`Client::connect_lite`, `Server::accept_lite`, and `Server::accept_request_lite` are gone; call `connect`, `accept`, or `accept_request`, which accept every ALPN.
+Every session now takes a `transport::poll::Boxable` transport, whose streams must also be `Sync` on native.
+A `!Send` transport wraps itself in a type that checks the thread on use, such as `moq_uring::transport::Local`.
+`Client::connect` and `Server::accept_request` are plain functions returning `impl Future`, so call them as before with `.await`.
+
 ### `track::Producer::largest()` is a getter
 
 `track::Producer::largest(&self)` returns the track's largest location now, `None` before any, instead of streaming each newer one; `track::Producer::poll_largest` is gone.