Merge branch 'main' into tail-call-windows

Author: mdboom

PROVISIONING.md

@@ -85,3 +85,16 @@ Install [git for Windows](https://git-scm.com/download/win).
 ### Install Python
 
 Install Python 3.9 or later from [python.org](https://python.org), and install for all users.
+
+### Install Clang 19
+
+For support for building with Clang and the tail-call interpreter, install
+the right version of the windows-msvc clang+llvm package from
+https://github.com/llvm/llvm-project/releases ([direct link to clang+llvm 19.1.6
+for x86_64](https://github.com/llvm/llvm-project/releases/download/llvmorg-19.1.6/clang+llvm-19.1.6-x86_64-pc-windows-msvc.tar.xz))
+and install it by copying its contents to `C:\Program Files\LLVM`, such that
+`C:\Program Files\LLVM\bin\clang-cl.exe` and `C:\Program
+Files\LLVM\lib\clang\19\` exist. (The version and installation
+directory are passed to `PCbuild\build.bat` in
+`bench_runner/templates/_benchmark.src.yml`, using `/p:LLVMInstallDir` and
+`/p:LLVMToolsVersion`.)

bench_runner/result.py

@@ -901,6 +901,8 @@ def load_all_results(
             key=lambda x: (
                 x.parsed_version,
                 x.commit_datetime,
+                tuple(x.flags),
+                x.filename,  # Just to produce a stable ordering
             ),
             reverse=True,
         )

bench_runner/templates/_benchmark.src.yml

@@ -126,7 +126,7 @@ jobs:
       - name: Running pyperformance
         if: ${{ steps.should_run.outputs.should_run != 'false' }}
         run: |
-          venv\Scripts\python.exe -m bench_runner run_benchmarks benchmark cpython\$env:BUILD_DEST\python.exe ${{ inputs.fork }} ${{ inputs.ref }} ${{ inputs.benchmarks || 'all' }} "${{ env.flags }}" --run_id ${{ github.run_id }}
+          venv\Scripts\python.exe -m bench_runner run_benchmarks benchmark ${{ (inputs.nogil == true && '(get-item cpython/$env:BUILD_DEST/python3.*.exe).FullName' || 'cpython/$env:BUILD_DEST/python.exe') }} ${{ inputs.fork }} ${{ inputs.ref }} ${{ inputs.benchmarks || 'all' }} "${{ env.flags }}" --run_id ${{ github.run_id }}
       # Pull again, since another job may have committed results in the meantime
       - name: Pull benchmarking
         if: ${{ steps.should_run.outputs.should_run != 'false' }}

pyproject.toml

@@ -31,7 +31,7 @@ test = [
     "black==25.1.0",
     "filelock==3.18.0",
     "flake8==7.1.2",
-    "pyright==1.1.396",
+    "pyright==1.1.397",
     "pytest==8.3.5",
     "pytest-xdist==3.6.1",
 ]