From 57e934d7d23457cf370e3e5461773b803ca4bde4 Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Thu, 22 Oct 2026 20:56:16 -0700 Subject: [PATCH 1/4] chore(quest): claim quest/m1/gh-test-gate.md Quest-Claim: 0f4625602d62664e From 334467f7dfb91b65cdd41754e69baf0b381e7adb Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Thu, 22 Oct 2026 20:59:14 -0700 Subject: [PATCH 2/4] ci: run the tooling tests on every check `just gh test` covers the CI selector, release, size, and benchmark tooling, but only Nightly ran it, so a change that broke a fixture merged green. The gh module runs on every diff, so its test phase now runs the suite in every `just check` and queue Check. Nightly's `gate` covers it, so its separate matrix entry goes. Co-Authored-By: Claude Opus 5.5 --- .github/ffi.md | 2 +- .github/justfile | 2 +- .github/workflows/nightly.yml | 3 +-- quest/m1/README.md | 1 - quest/m1/gh-test-gate.md | 27 --------------------------- quest/m2/audit-ci.md | 4 ++-- sh/dispatch.sh | 7 +++++-- sh/js/README.md | 2 +- 8 files changed, 11 insertions(+), 37 deletions(-) delete mode 100644 quest/m1/gh-test-gate.md diff --git a/.github/ffi.md b/.github/ffi.md index f8cfb1ffef0..c028d950c49 100644 --- a/.github/ffi.md +++ b/.github/ffi.md @@ -248,7 +248,7 @@ packages each target from its consumers succeed. It verifies the archive sidecar and manifest before CMake uses it, with no source fallback. An unrelated coordinator failure does not suppress a successful C++ target. Tag releases refuse same-run inputs and keep -the explicit release pin path. `just cpp check` and nightly `just gh test` also +the explicit release pin path. `just cpp check` and `just gh test` also cover hash refusal, archive traversal/symlink refusal, reproducible packaging, publication refusal, and CMake's build-free prebuilt path with malformed-input fixtures. diff --git a/.github/justfile b/.github/justfile index 0bfbf22fba3..b9bd97093a9 100644 --- a/.github/justfile +++ b/.github/justfile @@ -25,7 +25,7 @@ check: bun sh/gh/check-runs.ts python3 bench/relay.test.py -# Tests of repository tooling; nightly, not on every pull request. +# Tests of repository tooling; every `just check` and `just ci` runs them. test: python3 sh/coverage.test.py python3 sh/size/report.test.py diff --git a/.github/workflows/nightly.yml b/.github/workflows/nightly.yml index edca5a31179..1947b1a13a0 100644 --- a/.github/workflows/nightly.yml +++ b/.github/workflows/nightly.yml @@ -318,7 +318,6 @@ jobs: # `test drain` stands up two relays and drains one under a live JS # viewer: diff-independent, since a break can arrive through the relay, # js/net, or the wire between them. - # `gh test` exercises workflow and release tooling, which no pull request runs. # `rs quic-sim` sweeps moq-quic's simulated workloads from a fresh seed; PRs run a fixed 64. # `test drill impaired` runs the relay drills over a freshly seeded shaper; PRs run loopback. # `rs cli-export` runs moq-cli's export process tests, which take tens of seconds of real time. @@ -328,7 +327,7 @@ jobs: # `rs multicast` runs the mDNS tests, which loop multicast back on the runner. # `rs render-sw` renders through wgpu on Mesa's software Vulkan driver. # `test qmux` runs js/qmux against the in-tree Rust qmux, per draft. - recipe: ["gate", "rs doctest --workspace", "rs loom", "test drill-sensitivity", "rs uring", "test drain", "gh test", "test json-properties", "rs quic-sim", "test drill impaired", "rs cli-export", "rs quic-stress", "rs audio-devices", "rs multicast", "rs render-sw", "test qmux"] + recipe: ["gate", "rs doctest --workspace", "rs loom", "test drill-sensitivity", "rs uring", "test drain", "test json-properties", "rs quic-sim", "test drill impaired", "rs cli-export", "rs quic-stress", "rs audio-devices", "rs multicast", "rs render-sw", "test qmux"] include: - recipe: rs send-depth nightly: true diff --git a/quest/m1/README.md b/quest/m1/README.md index afb1c5d6779..69f50ccc452 100644 --- a/quest/m1/README.md +++ b/quest/m1/README.md @@ -135,7 +135,6 @@ tracks, so per-group stats stop costing cluster announces. - [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 - [NixOS leaf dials its root](/quest/m1/nix-cluster-connect.md) - the NixOS relay module sets the cluster variable the relay reads, so a leaf joins its root instead of running standalone - [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 - [CI housekeeping](/quest/m1/ci-housekeeping.md) - validate queue nextest at 24 threads and remove the actionlint queue ignore after its release diff --git a/quest/m1/gh-test-gate.md b/quest/m1/gh-test-gate.md deleted file mode 100644 index ef1865d8f10..00000000000 --- a/quest/m1/gh-test-gate.md +++ /dev/null @@ -1,27 +0,0 @@ -# [XS] Tooling tests gate every merge - -## Goal - -Every merge-queue Check runs `just gh test`, the tests of the CI selector, -release, size, and benchmark tooling, so a change to the machinery that picks -coverage cannot merge without its own tests. Nightly keeps running it as part -of the full gate. - -## Plan - -Decided 2026-10-22 from the repository improvement audit: run it on every -Check rather than select it by path. The whole recipe took about 10 s locally -(about 3% of a median 5.5-minute queue entry), and path selection would let a -broken selector skip the very tests that catch it. - -- The run that planned this found `gh test` red on `main`: - `sh/gh/setup.test.ts` expects Check to be the last step of its job, and - [moq-dev/moq#6155](https://github.com/moq-dev/moq/pull/6155) appended - "Upload build timings" after it. It stayed unnoticed because no PR runs the - suite. [moq-dev/moq#6325](https://github.com/moq-dev/moq/pull/6325) fixes - that fixture; land after it, or fix it here if it has not merged. -- Run it from the `gh` module's check in `sh/dispatch.sh` (or an always-selected - module) and add any tools it needs (`python3`) to that module's tool list. -- Update the recipe comment in `.github/justfile` and the `gh test` comment in - Nightly's gate matrix, which both say no pull request runs it. Keep it in - Nightly only if the full gate there does not already cover it. diff --git a/quest/m2/audit-ci.md b/quest/m2/audit-ci.md index 9888cf20aa9..b1e6ddb2720 100644 --- a/quest/m2/audit-ci.md +++ b/quest/m2/audit-ci.md @@ -104,8 +104,8 @@ verdicts: merged green and reddened the next night. That happened three times in two days (the setup fixture, the removed `writer` bench, and the unregistered jitter and SUBSCRIBE_TRACKS benches the m1 run registered). - [Tooling tests gate every merge](/quest/m1/gh-test-gate.md) moves it into - every Check; confirm it landed. + [moq-dev/moq#6480](https://github.com/moq-dev/moq/pull/6480) moved it into + every Check; confirm no such break recurs. - Platform (macOS, Windows) now runs only in Nightly and on release PRs, not in the merge queue. A Windows- or macOS-only regression surfaces first in Nightly, as the reconnect test did. diff --git a/sh/dispatch.sh b/sh/dispatch.sh index 45395e399ca..a4d89791344 100755 --- a/sh/dispatch.sh +++ b/sh/dispatch.sh @@ -158,7 +158,7 @@ declare -A tools=( [toml]='taplo' [nix]='nixfmt' [justfile]='' - [gh]='actionlint bun' + [gh]='actionlint bun python3 cargo' ) # CI and the local check share one selection and one lint/test dispatch. @@ -314,15 +314,18 @@ for module in "${selected[@]}"; do done # The bindings run their tests in the preamble; these suites run once here. +# gh's tooling tests run on every diff: path selection would let a broken +# selector skip the very tests that catch it. if [[ "$verb" == check ]]; then for module in "${selected[@]}"; do case "$module" in - rs | js | py) begin "test $module" ;; + rs | js | py | gh) begin "test $module" ;; esac case "$module" in rs) just rs test-changed "$rs_list" ;; js) just js test-changed "$js_list" ;; py) just py test ;; + gh) just gh test ;; esac finish done diff --git a/sh/js/README.md b/sh/js/README.md index 12d7ff9b184..35594f6f23d 100644 --- a/sh/js/README.md +++ b/sh/js/README.md @@ -38,6 +38,6 @@ nixpkgs moves Playwright, update the reported manifests and regenerate `bun.lock with `bun install` in the new shell. Smoke has its own flake and freshness check. Run `python3 sh/js/workspace.test.py` for graph and execution regressions. The same -suite runs in `just gh test` nightly. To inspect work without executing it, pass +suite runs in `just gh test` on every check. To inspect work without executing it, pass a file containing relevant paths, one per line, to `python3 sh/js/workspace.py plan LISTFILE`, or pass `--all`. From 2993739108641a5168b78514b6ac4df33ba2692a Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Fri, 23 Oct 2026 09:06:08 -0700 Subject: [PATCH 3/4] fix(ci): require every gh test executable up front The always-selected gh module now runs `just gh test`, whose package and lint tests need rpm, nfpm, dpkg-deb and clippy-driver. List them so a local run skips the module and MOQ_STRICT reports them up front instead of failing mid-suite. Co-Authored-By: Claude Opus 5.5 --- sh/dispatch.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sh/dispatch.sh b/sh/dispatch.sh index a4d89791344..1c97fd2daa8 100755 --- a/sh/dispatch.sh +++ b/sh/dispatch.sh @@ -158,7 +158,7 @@ declare -A tools=( [toml]='taplo' [nix]='nixfmt' [justfile]='' - [gh]='actionlint bun python3 cargo' + [gh]='actionlint bun python3 cargo clippy-driver dpkg-deb nfpm rpm' ) # CI and the local check share one selection and one lint/test dispatch. From a95a15c7db325d2e7bc725c2d8f9c864359d0433 Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Fri, 23 Oct 2026 09:08:11 -0700 Subject: [PATCH 4/4] fix(ci): run the relay's http_auth bench nightly moq-dev/moq#6488 added the bench target without registering it, which the newly gated `just gh test` catches. Co-Authored-By: Claude Opus 5.5 --- sh/gh/nightly.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sh/gh/nightly.sh b/sh/gh/nightly.sh index 96ac983266c..3d8b85fdf19 100755 --- a/sh/gh/nightly.sh +++ b/sh/gh/nightly.sh @@ -80,7 +80,7 @@ case "$1" in # independently for each target. Keep the explicit targets: '*' would add coverage. criterion shared --no-fail-fast -p moq-relay -p moq-stats -p moq-mux -p moq-rtmp -p moq-uring -p moq-pattern \ -p moq-quic --features moq-stats/bench,moq-quic/fuzz \ - --bench http_idle --bench auth --bench producer_tick --bench decode --bench aggregate \ + --bench http_idle --bench auth --bench http_auth --bench producer_tick --bench decode --bench aggregate \ --bench catalog --bench ts_schedule --bench export --bench containers --bench matching \ --bench chunk_interleaving --bench worker_turns \ --bench streams --bench receive --bench endpoint --bench wire