ci: Run semver checks even on non-host targets

Author: tgross35

.github/workflows/ci.yaml

@@ -17,6 +17,7 @@ env:
   RUSTDOCFLAGS: -Dwarnings
   RUSTFLAGS: -Dwarnings
   RUST_BACKTRACE: full
+  SERIES: "1.0" # 1.0 alpha series
 
 defaults:
   run:
@@ -79,6 +80,10 @@ jobs:
         uses: taiki-e/install-action@cargo-semver-checks
         if: matrix.toolchain == 'stable'
 
+      - name: Retrieve semver baseline
+        if: matrix.toolchain == 'stable'
+        run: ./ci/semver.sh
+
       # FIXME(ci): These `du` statements are temporary for debugging cache
       - name: Target size before restoring cache
         run: du -sh target | sort -k 2 || true
@@ -100,6 +105,7 @@ jobs:
 
           python3 ci/verify-build.py \
             --toolchain "$TOOLCHAIN" \
+            ${BASELINE_CRATE_DIR:+"--baseline-crate-dir" "$BASELINE_CRATE_DIR"} \
             ${{ matrix.only && format('--only "{0}"', matrix.only) }} \
             ${{ matrix.half && format('--half "{0}"', matrix.half) }}
       - name: Target size after job completion

ci/semver.sh

@@ -0,0 +1,34 @@
+#!/bin/bash
+# Download a baseline crate to run semver checks against
+
+set -euxo pipefail
+
+# https://crates.io/data-access#api
+
+# Need to include versions to get the default and stable versions
+meta=$(curl -L https://index.crates.io/li/bc/libc)
+
+# Versions to check against
+if [ "${SERIES:-}" = "0.2" ]; then
+    pat="^0.2"
+elif [ "${SERIES:-}" = "1.0" ]; then
+    pat="^1.0"
+else
+    echo "SERIES must be set to either '0.2' or '1.0'"
+    exit 1
+fi
+
+# Find the most recent version matching a pattern.
+release=$(
+    echo "$meta" |
+    jq -er --slurp --arg pat "$pat" 'map(select(.vers | test($pat))) | last'
+)
+version=$(echo "$release" | jq -r '.vers')
+
+crate_dir="libc-$version"
+curl -L "https://static.crates.io/crates/libc/libc-$version.crate" | tar xzf -
+
+# Need to convince Cargo it's not part of our workspace
+echo '[workspace]' >> "$crate_dir/Cargo.toml"
+
+echo "BASELINE_CRATE_DIR=$(realpath "$crate_dir")" >> "$GITHUB_ENV"

ci/verify-build.py

@@ -9,7 +9,8 @@
 import time
 from dataclasses import dataclass, field
 from enum import Enum, IntEnum
-from typing import Optional
+from pathlib import Path
+from typing import Optional, Sequence
 
 
 ESC_CYAN = "\033[1;36m"
@@ -35,6 +36,7 @@ class Cfg:
     toolchain: Toolchain = field(init=False)
     host_target: str = field(init=False)
     os_: Os = field(init=False)
+    baseline_crate_dir: Optional[Path]
 
     def __post_init__(self):
         rustc_output = check_output(["rustc", f"+{self.toolchain_name}", "-vV"])
@@ -200,13 +202,13 @@ def __post_init__(self):
 ]
 
 
-def eprint(*args, **kw):
+def eprint(*args, **kw) -> None:
     print(*args, file=sys.stderr, **kw)
 
 
-def xtrace(args: list[str], /, env: Optional[dict[str, str]]):
+def xtrace(args: Sequence[str | Path], /, env: Optional[dict[str, str]]) -> None:
     """Print commands before running them."""
-    astr = " ".join(args)
+    astr = " ".join(str(arg) for arg in args)
     if env is None:
         eprint(f"+ {astr}")
     else:
@@ -215,17 +217,20 @@ def xtrace(args: list[str], /, env: Optional[dict[str, str]]):
         eprint(f"+ {estr} {astr}")
 
 
-def check_output(args: list[str], /, env: Optional[dict[str, str]] = None) -> str:
+def check_output(
+    args: Sequence[str | Path], /, env: Optional[dict[str, str]] = None
+) -> str:
     xtrace(args, env=env)
     return sp.check_output(args, env=env, encoding="utf8")
 
 
-def run(args: list[str], /, env: Optional[dict[str, str]] = None):
+def run(args: Sequence[str | Path], /, env: Optional[dict[str, str]] = None) -> None:
     xtrace(args, env=env)
     sp.run(args, env=env, check=True)
 
 
-def check_dup_targets():
+def check_dup_targets() -> None:
+    """Ensure there are no duplicate targets in the list."""
     all = set()
     duplicates = set()
     for target in TARGETS:
@@ -235,7 +240,78 @@ def check_dup_targets():
     assert len(duplicates) == 0, f"duplicate targets: {duplicates}"
 
 
-def test_target(cfg: Cfg, target: Target):
+def do_semver_checks(cfg: Cfg, target: Target) -> None:
+    tname = target.name
+    if cfg.toolchain != Toolchain.STABLE:
+        eprint("Skipping semver checks")
+        return
+
+    if tname == cfg.host_target:
+        # FIXME(semver): can't pass `--target` to `cargo-semver-checks` so we can only
+        # run directly on the host target
+        eprint("Running semver checks on host")
+        run(
+            [
+                "cargo",
+                "semver-checks",
+                "--only-explicit-features",
+                "--features=std,extra_traits",
+            ]
+        )
+        return
+
+    if cfg.baseline_crate_dir is None:
+        eprint(
+            "Non-host target: --baseline-crate-dir must be specified to \
+            run semver-checks"
+        )
+        return
+
+    if not target.dist:
+        eprint("Skipping semver checks on non-dist target")
+        return
+
+    # Since semver-checks doesn't work with `--target`, we build the json our self and
+    # hand it over.
+    eprint("Running semver checks with cross compilation")
+
+    env = os.environ.copy()
+    env.setdefault("RUSTFLAGS", "")
+    env.setdefault("RUSTDOCFLAGS", "")
+    # Needed for rustdoc json
+    env["RUSTC_BOOTSTRAP"] = "1"
+    # Unset everything that would cause us to get warnings for the baseline
+    env["RUSTFLAGS"] += " -Awarnings"
+    env["RUSTDOCFLAGS"] += " -Awarnings"
+    env.pop("LIBC_CI", None)
+
+    cmd = ["cargo", "rustdoc", "--target", tname]
+    rustdoc_args = ["--", "-Zunstable-options", "--output-format=json"]
+
+    # Build the current crate and the baseline crate, which CI should have downloaded
+    run([*cmd, *rustdoc_args], env=env)
+    run(
+        [*cmd, "--manifest-path", cfg.baseline_crate_dir / "Cargo.toml", *rustdoc_args],
+        env=env,
+    )
+
+    baseline = cfg.baseline_crate_dir / "target" / tname / "doc" / "libc.json"
+    current = Path("target") / tname / "doc" / "libc.json"
+
+    run(
+        [
+            "cargo",
+            "semver-checks",
+            "--baseline-rustdoc",
+            baseline,
+            "--current-rustdoc",
+            current,
+        ]
+    )
+
+
+def test_target(cfg: Cfg, target: Target) -> None:
+    """Run tests for a single target."""
     start = time.time()
     env = os.environ.copy()
     env.setdefault("RUSTFLAGS", "")
@@ -261,14 +337,15 @@ def test_target(cfg: Cfg, target: Target):
     if not target.dist:
         # If we can't download a `core`, we need to build it
         cmd += ["-Zbuild-std=core"]
-        # FIXME: With `build-std` feature, `compiler_builtins` emits a lot of lint warnings.
+        # FIXME: With `the build-std` feature, `compiler_builtins` emits a lot of
+        # lint warnings.
         env["RUSTFLAGS"] += " -Aimproper_ctypes_definitions"
     else:
         run(["rustup", "target", "add", tname, "--toolchain", cfg.toolchain_name])
 
     # Test with expected combinations of features
     run(cmd, env=env)
-    run(cmd + ["--features=extra_traits"], env=env)
+    run([*cmd, "--features=extra_traits"], env=env)
 
     # Check with different env for 64-bit time_t
     if target_os == "linux" and target_bits == "32":
@@ -286,49 +363,38 @@ def test_target(cfg: Cfg, target: Target):
         run(cmd, env=env | {"RUST_LIBC_UNSTABLE_MUSL_V1_2_3": "1"})
 
     # Test again without default features, i.e. without `std`
-    run(cmd + ["--no-default-features"])
-    run(cmd + ["--no-default-features", "--features=extra_traits"])
+    run([*cmd, "--no-default-features"])
+    run([*cmd, "--no-default-features", "--features=extra_traits"])
 
     # Ensure the crate will build when used as a dependency of `std`
     if cfg.nightly():
-        run(cmd + ["--no-default-features", "--features=rustc-dep-of-std"])
+        run([*cmd, "--no-default-features", "--features=rustc-dep-of-std"])
 
     # For freebsd targets, check with the different versions we support
     # if on nightly or stable
     if "freebsd" in tname and cfg.toolchain >= Toolchain.STABLE:
         for version in FREEBSD_VERSIONS:
             run(cmd, env=env | {"RUST_LIBC_UNSTABLE_FREEBSD_VERSION": str(version)})
             run(
-                cmd + ["--no-default-features"],
+                [*cmd, "--no-default-features"],
                 env=env | {"RUST_LIBC_UNSTABLE_FREEBSD_VERSION": str(version)},
             )
 
-    is_stable = cfg.toolchain == Toolchain.STABLE
-    # FIXME(semver): can't pass `--target` to `cargo-semver-checks` so we restrict to
-    # the host target
-    is_host = tname == cfg.host_target
-    if is_stable and is_host:
-        eprint("Running semver checks")
-        run(
-            [
-                "cargo",
-                "semver-checks",
-                "--only-explicit-features",
-                "--features=std,extra_traits",
-            ]
-        )
-    else:
-        eprint("Skipping semver checks")
+    do_semver_checks(cfg, target)
 
     elapsed = round(time.time() - start, 2)
     eprint(f"Finished checking target {tname} in {elapsed} seconds")
 
 
-def main():
+def main() -> None:
     p = argparse.ArgumentParser()
     p.add_argument("--toolchain", required=True, help="Rust toolchain")
     p.add_argument("--only", help="only targets matching this regex")
     p.add_argument("--skip", help="skip targets matching this regex")
+    p.add_argument(
+        "--baseline-crate-dir",
+        help="specify the directory of the crate to run semver checks against",
+    )
     p.add_argument(
         "--half",
         type=int,
@@ -337,7 +403,10 @@ def main():
     )
     args = p.parse_args()
 
-    cfg = Cfg(toolchain_name=args.toolchain)
+    cfg = Cfg(
+        toolchain_name=args.toolchain,
+        baseline_crate_dir=args.baseline_crate_dir and Path(args.baseline_crate_dir),
+    )
     eprint(f"Config: {cfg}")
     eprint("Python version: ", sys.version)
     check_dup_targets()