INT26: wave H, the late picks and the corpus re-pin to b40aa8e (1.0.0-rc.1 candidate)

.github/workflows/ci.yml

@@ -22,10 +22,22 @@ env:
   CCACHE_COMPRESS: "1"
   CCACHE_MAXSIZE: 500M
 
+# The heavy jobs run on the organization's 16-core larger runner, and the macOS
+# legs on GitHub's M2 runner for the macos-26 image, for a push, the schedule, a
+# dispatch and a pull request from a branch of this repository. A larger runner
+# bills the organization even for a public repository, so every other event --
+# a fork's pull request above all -- runs on the free standard runner.
+# Each verification sizes its parallelism to the runner that took it.
+# scripts/ci_preflight.py pins every runner, time limit and job count, and
+# docs/ci.md gives each limit's measured basis.
 jobs:
   preflight:
-    runs-on: ubuntu-24.04
-    timeout-minutes: 10
+    runs-on: ${{ (github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository)) && 'pf-linux-x64-16' || 'ubuntu-24.04' }}
+    # Its verifier-tests stage is one serial process: 475 s at 0d76a099's tree
+    # on the maintainers' verification hosts, which ran it 1.93-1.95 times as
+    # fast as the standard runner on the same trees, so about 15 min here. Ten
+    # minutes cancelled main's run at 0d76a099 593 s into it.
+    timeout-minutes: 45
     steps:
       - uses: actions/checkout@v4
         with:
@@ -67,14 +79,27 @@ jobs:
           if-no-files-found: warn
 
   build:
-    timeout-minutes: 45
+    # Each leg keeps its name, build (<image>, <type>), whatever the matrix carries.
+    name: build (${{ matrix.os }}, ${{ matrix.build_type }})
+    # A full-set leg waits on the test_ci_verify CTest row, 976-1288 s on the
+    # standard runners at 0d76a099, which more cores do not shorten. There the
+    # Release legs, which a fork's pull request runs in full, took 30.3 and
+    # 32.0 min, and the macOS Debug leg 37.0 min.
+    timeout-minutes: 75
     strategy:
       fail-fast: false
       matrix:
         os: [ubuntu-24.04, macos-26]
         build_type: [Release, Debug]
+        include:
+          - os: ubuntu-24.04
+            larger_runner: pf-linux-x64-16
+          - os: macos-26
+            larger_runner: macos-26-xlarge
 
-    runs-on: ${{ matrix.os }}
+    # matrix.os is the standard runner a fork's pull request gets; its
+    # larger_runner runs the same image with more cores.
+    runs-on: ${{ (github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository)) && matrix.larger_runner || matrix.os }}
 
     steps:
       - uses: actions/checkout@v4
@@ -121,7 +146,7 @@ jobs:
           # ratio -- for a runner whose CPU accounting is unusable; record the
           # measured reason here and in docs/ci.md before setting it.
           PINEFORGE_RUNTIME_BUDGET_CANDIDATE_ONLY: '0'
-        run: python3 scripts/ci_verify.py ${{ matrix.build_type == 'Release' && 'release' || 'debug' }} --build-dir build --jobs 4 --ccache ${{ github.event_name == 'pull_request' && matrix.build_type == 'Debug' && '--exclude-label slow' || '' }}
+        run: python3 scripts/ci_verify.py ${{ matrix.build_type == 'Release' && 'release' || 'debug' }} --build-dir build --jobs "$(getconf _NPROCESSORS_ONLN)" --ccache ${{ github.event_name == 'pull_request' && matrix.build_type == 'Debug' && '--exclude-label slow' || '' }}
 
       - name: Stage and summarize diagnostics
         if: always()
@@ -142,9 +167,11 @@ jobs:
   # instrumentation propagates into every test binary (CMakeLists.txt option
   # PINEFORGE_ENABLE_SANITIZERS).
   sanitizers:
-    runs-on: ubuntu-24.04
+    runs-on: ${{ (github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository)) && 'pf-linux-x64-16' || 'ubuntu-24.04' }}
     # A full run (push to main, manual dispatch) gives its CTest stage an hour
-    # after a build of about 30 minutes; the PR set keeps its 30-minute stage.
+    # after the build; the PR set keeps its 30-minute stage, after about 30
+    # minutes of build and ABI providers on the standard runner a fork's pull
+    # request gets.
     timeout-minutes: 120
     env:
       ASAN_OPTIONS: detect_leaks=1:halt_on_error=1:abort_on_error=1
@@ -173,7 +200,7 @@ jobs:
             ccache-${{ runner.os }}-${{ runner.arch }}-sanitizers-
 
       - name: Verify (sanitizers)
-        run: python3 scripts/ci_verify.py sanitizers --build-dir build-asan --jobs 4 --ccache ${{ github.event_name == 'pull_request' && '--exclude-label slow' || '' }}
+        run: python3 scripts/ci_verify.py sanitizers --build-dir build-asan --jobs "$(getconf _NPROCESSORS_ONLN)" --ccache ${{ github.event_name == 'pull_request' && '--exclude-label slow' || '' }}
 
       - name: Stage and summarize diagnostics
         if: always()
@@ -194,8 +221,10 @@ jobs:
   # modules are built from the kernel alone, and every examples/native host
   # runs as an example_* row asserting both its exit code and its summary line.
   kernel-only:
-    runs-on: ubuntu-24.04
-    timeout-minutes: 45
+    runs-on: ${{ (github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository)) && 'pf-linux-x64-16' || 'ubuntu-24.04' }}
+    # 22.3 min on the standard runner at 0d76a099, where a fork's pull request
+    # still runs the whole kernel set.
+    timeout-minutes: 60
     steps:
       - uses: actions/checkout@v4
         with:
@@ -220,7 +249,7 @@ jobs:
             ccache-${{ runner.os }}-${{ runner.arch }}-kernel-
 
       - name: Verify (kernel)
-        run: python3 scripts/ci_verify.py kernel --build-dir build-kernel --jobs 4 --ccache
+        run: python3 scripts/ci_verify.py kernel --build-dir build-kernel --jobs "$(getconf _NPROCESSORS_ONLN)" --ccache
 
       - name: Stage and summarize diagnostics
         if: always()
@@ -255,6 +284,7 @@ jobs:
     name: build
     needs: [preflight, build, sanitizers, native-live, kernel-only, corpus-parity-subset]
     if: always()
+    # A few seconds of shell: the standard runner, for every event.
     runs-on: ubuntu-24.04
     timeout-minutes: 5
     steps:

.github/workflows/corpus-parity.yml

@@ -17,17 +17,17 @@ name: Corpus parity
 # the runtime and all 312 corpus strategies build in ~105 s at JOBS=8, and then
 # the run phase -- which run_corpus.sh executes serially, one ctypes process per
 # probe over a 221861-bar feed -- took 1792 s, with 94 s more to judge the
-# result. The dominant term is single-core serial work, which a 4-vCPU hosted
-# runner runs slower rather than faster, so the whole population cannot be held
-# to the ~25 min a required pull-request check is budgeted for. It therefore
-# runs nightly and on demand, plus on any pull request that touches the pin or
-# the parity tooling itself.
+# result. The dominant term is single-core serial work, which a runner's extra
+# cores do not shorten, so the whole population cannot be held to the ~25 min
+# a required pull-request check is budgeted for. It therefore runs nightly and
+# on demand, plus on any pull request that touches the pin or the parity
+# tooling itself.
 #
 # PR EARLY SIGNAL. The `corpus-parity-subset` job below
-# re-runs the 30 probes scripts/corpus_parity_subset.txt names -- in parallel,
+# re-runs the 54 probes scripts/corpus_parity_subset.txt names -- in parallel,
 # they write into disjoint directories -- and judges them against the same
 # pinned sha256 rows the nightly uses. Measured on the same laptop at JOBS=8,
-# under sibling load: derive 2 s, build the runtime + the 30 strategy .so 68 s
+# under sibling load, when the list held its first 30 probes: derive 2 s, build the runtime + the 30 strategy .so 68 s
 # from clean, run 45 s wall for 80 s of probe CPU, judge <1 s; 94 s end to end
 # over an up-to-date build directory, under 3 min cold.
 # .github/workflows/ci.yml calls it as a reusable workflow and includes it in
@@ -80,12 +80,19 @@ env:
   # pin is almost entirely cache hits.
   CCACHE_MAXSIZE: 2G
 
+# Both jobs run on the organization's 16-core larger runner for a push, the
+# schedule, a dispatch and a pull request from a branch of this repository;
+# every other event, a fork's pull request above all (this workflow's own or
+# CI's: in a workflow_call the github context is the caller's), runs on the
+# free standard runner. The build uses every core of the runner that took the
+# job. scripts/ci_preflight.py pins the runners and limits.
 jobs:
   # The whole population. Nightly, on demand, and on a pull request that
   # touches the pin or the parity tooling. Never reached by `mode: subset`.
   corpus-parity:
     if: inputs.mode != 'subset'
-    runs-on: ubuntu-24.04
+    runs-on: ${{ (github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository)) && 'pf-linux-x64-16' || 'ubuntu-24.04' }}
+    # 16.3-26.2 min on the standard runner a fork's pull request still gets.
     timeout-minutes: 120
     steps:
       - uses: actions/checkout@v4
@@ -129,9 +136,8 @@ jobs:
         shell: bash
         env:
           BUILD_DIR: build-corpus-parity
-          JOBS: "4"
           CMAKE_CXX_COMPILER_LAUNCHER: ccache
-        run: ./scripts/check_corpus_parity.sh
+        run: JOBS="$(getconf _NPROCESSORS_ONLN)" ./scripts/check_corpus_parity.sh
 
       - name: Collect the drifted trades
         if: failure()
@@ -156,11 +162,12 @@ jobs:
           path: corpus-parity-diagnostics/
           if-no-files-found: warn
 
-  # The advisory PR subset: the 30 probes scripts/corpus_parity_subset.txt
+  # The advisory PR subset: the 54 probes scripts/corpus_parity_subset.txt
   # names, judged against the same pinned sha256 rows. Reached through CI.
   corpus-parity-subset:
     if: inputs.mode == 'subset'
-    runs-on: ubuntu-24.04
+    runs-on: ${{ (github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository)) && 'pf-linux-x64-16' || 'ubuntu-24.04' }}
+    # 1.8-4.7 min on the standard runner.
     timeout-minutes: 30
     steps:
       # Credentials stay persisted, exactly as the nightly job above: the
@@ -218,7 +225,7 @@ jobs:
       # A restored cache is not evidence. Every LFS pointer the corpus commits
       # under data/ records its content's sha256; check the bytes on disk
       # against it, so a corrupt or stale cache is "the feed is wrong" here and
-      # not "30 probes drifted" three steps later.
+      # not "54 probes drifted" three steps later.
       - name: Verify the chart feed against its Git-LFS pointers
         shell: bash
         run: |
@@ -245,9 +252,8 @@ jobs:
         shell: bash
         env:
           BUILD_DIR: build-corpus-parity-subset
-          JOBS: "4"
           CMAKE_CXX_COMPILER_LAUNCHER: ccache
-        run: ./scripts/check_corpus_parity.sh --subset
+        run: JOBS="$(getconf _NPROCESSORS_ONLN)" ./scripts/check_corpus_parity.sh --subset
 
       - name: Collect the drifted trades
         if: failure()

.github/workflows/native-live.yml

@@ -27,10 +27,17 @@ env:
   CURL_VERSION: "8.14.1"
   CURL_SHA256: "f4619a1e2474c4bbfedc88a7c2191209c8334b48fa1f4e53fd584cc12e9120dd"
 
+# The organization's 16-core larger runner for a push, the schedule, a
+# dispatch and a pull request from a branch of this repository; every other
+# event, a fork's pull request above all, runs on the free standard runner. In
+# a workflow_call the github context is the caller's, so a fork's pull request
+# through CI lands there too. scripts/ci_preflight.py pins the runner and limit.
 jobs:
   native-live:
-    runs-on: ubuntu-24.04
-    timeout-minutes: 45
+    runs-on: ${{ (github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository)) && 'pf-linux-x64-16' || 'ubuntu-24.04' }}
+    # The full set waits on the test_ci_verify CTest row, 1034 s on the
+    # standard runner at 0d76a099, which more cores do not shorten.
+    timeout-minutes: 60
     steps:
       - uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262
         with:
@@ -105,7 +112,7 @@ jobs:
           cmake --install curl-build 2>&1 | tee curl-install.log
 
       - name: Verify native live
-        run: python3 scripts/ci_verify.py native --build-dir build-live --jobs 4 --generator Ninja --curl-dir "${{ github.workspace }}/build-native-deps/curl-install/lib/cmake/CURL" --ccache --require-websocket ${{ inputs.exclude_slow && '--exclude-label slow' || '' }}
+        run: python3 scripts/ci_verify.py native --build-dir build-live --jobs "$(getconf _NPROCESSORS_ONLN)" --generator Ninja --curl-dir "${{ github.workspace }}/build-native-deps/curl-install/lib/cmake/CURL" --ccache --require-websocket ${{ inputs.exclude_slow && '--exclude-label slow' || '' }}
 
       - name: Stage and summarize diagnostics
         if: always()