diff --git a/.agents/rules/rust.md b/.agents/rules/rust.md index adaa035d3a..8e60c0f457 100644 --- a/.agents/rules/rust.md +++ b/.agents/rules/rust.md @@ -53,12 +53,26 @@ When adding or updating a Rust crate dependency: and `PathBuf`; never round-trip `OsString -> String -> PathBuf` via `to_string_lossy()`. Return `std::process::ExitCode` from `main()` and `run_cli`. +* **Symbolic constants in `match` arms**: When string tokens (subcommand names, + CLI flags, or wire-format tags) appear across multiple sites (`match` arms, + registries, or error messages), group them as constants in a module (`mod + flags`, `mod subcommands`) and match via qualified paths (`flags::TEMPLATE`, + `subcommands::ZIPPER`). A misspelled bare identifier in a Rust `match` arm + parses as a catch-all variable binding, whereas a module-qualified path fails + compilation (`E0531`). Keep raw string literals in tests so tests catch typos + in constant values, and pass `case=no` to `keep-sorted` when `_` vs. + uppercase ASCII would invert runtime lowercase sort order (`"zip_main_maker"` + < `"zipper"`, whereas `ZIPPER` < `ZIP_MAIN_MAKER`). * **Collections**: Default to `Vec` and `HashMap`/`HashSet` over - `BTreeMap`/`BTreeSet` unless sorted-key traversal is a data-structure + `BTreeMap`/`BTreeSet` (and over `Vec<(K, V)>` linear scans when insertion + order is not required) unless sorted-key traversal is a data-structure invariant. Sort a `Vec` at formatting or output boundaries when deterministic - order is required. + order is required; omit defensive key sorting when template placeholders are + disjoint delimited tokens (e.g., `%KEY%`). * **Cohesive file organization**: Place core logic in library crates - (`src/lib.rs`) so unit tests call library functions directly. Group CLI + (`src/lib.rs`) so unit tests call library functions directly, and extract + shared wire/manifest parsers into a small workspace library crate rather than + duplicating line parsing across sibling `shrubbery` crates. Group CLI option structs, `parse_args`, and `run_cli` together in one contiguous section rather than splitting `parse_args` at the top and `run_cli` at the bottom around domain logic. @@ -94,8 +108,11 @@ When adding or updating a Rust crate dependency: `allow-*-in-tests` settings only recognize `#[test]` functions and `#[cfg(test)]` items, so helper functions in integration-test crates (`tests/*/*_test.rs`) would be flagged. Code under `#[cfg(unix)]` must - leave no `unused_mut`/`useless_vec` behind on Windows; rebind instead of - conditionally `push`ing. + leave no `unused_mut`/`useless_vec` behind on Windows: initialize base + collections as fixed-size arrays (`[...]`) or strings rather than `vec![...]` + before rebinding under `#[cfg(unix)]` (spot-check the non-Unix branch locally + on Linux by temporarily flipping `#[cfg(unix)]` to `#[cfg(windows)]` before + pushing). * **WORKSPACE-mode stub**: Any new `@rules_rust//...` flag or aspect referenced from `.bazelrc` must also exist in `tests/modules/rules_rust_stub/`, which WORKSPACE CI jobs use in place of diff --git a/.agents/skills/analyze-ci-failure/scripts/BUILD.bazel b/.agents/skills/analyze-ci-failure/scripts/BUILD.bazel new file mode 100644 index 0000000000..d990069c63 --- /dev/null +++ b/.agents/skills/analyze-ci-failure/scripts/BUILD.bazel @@ -0,0 +1,18 @@ +load("//python:py_library.bzl", "py_library") +load("//python:py_test.bzl", "py_test") + +package(default_visibility = ["//visibility:private"]) + +py_library( + name = "analyze_ci_failure", + testonly = True, + srcs = ["analyze_ci_failure.py"], + imports = ["."], +) + +py_test( + name = "analyze_ci_failure_test", + size = "small", + srcs = ["analyze_ci_failure_test.py"], + deps = [":analyze_ci_failure"], +) diff --git a/.agents/skills/analyze-ci-failure/scripts/analyze_ci_failure.py b/.agents/skills/analyze-ci-failure/scripts/analyze_ci_failure.py index e11334a41c..094d4a54fb 100755 --- a/.agents/skills/analyze-ci-failure/scripts/analyze_ci_failure.py +++ b/.agents/skills/analyze-ci-failure/scripts/analyze_ci_failure.py @@ -8,7 +8,10 @@ import urllib.parse import urllib.request -_ANSI_ESCAPE = re.compile(r"\x1B(?:[@-Z\\-_]|\[[0-?]*[ -/]*[@-~])") +_ANSI_ESCAPE = re.compile( + r"\x1B(?:[_\]][^\x07\x1B]*(?:\x07|\x1B\\)|[@-Z\\-_]|\[[0-?]*[ -/]*[@-~])" +) +_BUILDKITE_TIMESTAMP = re.compile(r"^_?bk;t=\d+\s*") _HTTP_5XX_DOWNLOAD_ERROR = re.compile(r"get returned 5\d\d\b", re.IGNORECASE) _HTTP_GATEWAY_ERRORS = ( @@ -19,38 +22,57 @@ # keep-sorted end ) -_ERROR_KEYWORDS = _HTTP_GATEWAY_ERRORS + ( +_DNS_OR_GIT_MIRROR_ERRORS = ( + # keep-sorted start + "could not resolve host:", + "getting/updating git mirror:", + "temporary failure in name resolution", + # keep-sorted end +) + +_MODEXTWD_CLEANUP_ERRORS = ( # keep-sorted start - "##[error]", - "connection reset by peer", - "critical path", - "curl: (", - "errno 5", - "error downloading [", - "error during connect:", - "error waiting for container", - "error:", - "error]", - "exception", - "exit code", - "exit-code", - "failed:", - "fatal", - "fatal:", - "filenotfounderror", - "get returned 5", - "input/output error", - "no such package", - "no such target", - "status 125", - "traceback", - "unable to find bazelisk version", - "user command error:", - "would be reformatted", - "would reformat:", + "failed to clean up module context directory:", # keep-sorted end ) +_ERROR_KEYWORDS = ( + _HTTP_GATEWAY_ERRORS + + _DNS_OR_GIT_MIRROR_ERRORS + + _MODEXTWD_CLEANUP_ERRORS + + ( + # keep-sorted start + "##[error]", + "connection reset by peer", + "critical path", + "curl: (", + "errno 5", + "error downloading [", + "error during connect:", + "error waiting for container", + "error:", + "error]", + "exception", + "exit code", + "exit-code", + "failed:", + "fatal", + "fatal:", + "filenotfounderror", + "get returned 5", + "input/output error", + "no such package", + "no such target", + "status 125", + "traceback", + "unable to find bazelisk version", + "user command error:", + "would be reformatted", + "would reformat:", + # keep-sorted end + ) +) + def fetch_log(job_name, build_id, job_id, output_path): if ( @@ -200,8 +222,8 @@ def parse_log(log_path): errors = [] for line in lines: clean_line = _ANSI_ESCAPE.sub("", line).strip() - # Clean buildkite timestamp prefix: _bk;t=... - clean_line = re.sub(r"^_bk;t=\d+\s*", "", clean_line) + # Strip pre-stripped `_bk;t=` / `bk;t=` prefixes in addition to APC sequences. + clean_line = _BUILDKITE_TIMESTAMP.sub("", clean_line) lower_line = clean_line.lower() if clean_line and any(kw in lower_line for kw in _ERROR_KEYWORDS): errors.append(clean_line) @@ -272,6 +294,19 @@ def create_plan(job_name, log_path, errors): "GitHub Actions `setup-bazel` step (`Unable to find Bazelisk version`). " "This is an infrastructure network flake, not a codebase bug." ) + elif _contains_any(errors, _DNS_OR_GIT_MIRROR_ERRORS): + is_flake = True + flake_reason = ( + "Transient CI runner DNS resolution or git mirror checkout failure. " + "This is an infrastructure network flake, not a codebase bug." + ) + elif _contains_any(errors, _MODEXTWD_CLEANUP_ERRORS): + is_flake = True + flake_reason = ( + "Transient Windows file-lock race while Bazel cleaned up a temporary " + "`modextwd` module extension directory (`Permission denied`). " + "This is an infrastructure flake, not a codebase bug." + ) classification = ( "⚡ **Classification**: **Infrastructure / Flake Issue** (Not a codebase bug)" diff --git a/.agents/skills/analyze-ci-failure/scripts/analyze_ci_failure_test.py b/.agents/skills/analyze-ci-failure/scripts/analyze_ci_failure_test.py index 05a6089c98..bb41dc34ec 100755 --- a/.agents/skills/analyze-ci-failure/scripts/analyze_ci_failure_test.py +++ b/.agents/skills/analyze-ci-failure/scripts/analyze_ci_failure_test.py @@ -74,6 +74,44 @@ def test_benign_download_progress_lines_not_captured_by_parse_log(self) -> None: errors, ["ERROR: Analysis of target '//foo:bar' failed: build failed"] ) + def test_buildkite_apc_timestamp_prefix_stripped_from_errors(self) -> None: + errors, _ = self._classify_log( + "\x1b_bk;t=1794137028816\x07ERROR: Build did NOT complete successfully\n" + "_bk;t=1794137028817 FAILED: Build did NOT complete successfully\n" + ) + self.assertEqual( + errors, + [ + "ERROR: Build did NOT complete successfully", + "FAILED: Build did NOT complete successfully", + ], + ) + + def test_runner_dns_and_git_mirror_checkout_classified_as_flake(self) -> None: + for log_text in ( + "fatal: unable to access 'https://github.com/bazel-contrib/rules_python.git/': Could not resolve host: github.com\n", + "Error: getting/updating git mirror: exit status 128\n", + "HTTPSConnectionPool(host='docs.python.org', port=443): Temporary failure in name resolution\n", + ): + with self.subTest(log_text=log_text): + _, plan = self._classify_log(log_text) + self.assertIn(_FLAKE_BANNER, plan) + + def test_windows_modextwd_cleanup_permission_denied_classified_as_flake( + self, + ) -> None: + errors, plan = self._classify_log( + "WARNING: errors encountered while analyzing target '//tests/py_zipapp:system_python_bin'\n" + "Failed to clean up module context directory: C:/b/nzh6davs/modextwd/rules_rust++crate/cargo-bazel.exe (Permission denied)\n" + ) + self.assertEqual( + errors, + [ + "Failed to clean up module context directory: C:/b/nzh6davs/modextwd/rules_rust++crate/cargo-bazel.exe (Permission denied)" + ], + ) + self.assertIn(_FLAKE_BANNER, plan) + if __name__ == "__main__": unittest.main() diff --git a/.agents/skills/retry-ci-job/SKILL.md b/.agents/skills/retry-ci-job/SKILL.md index 6737ebdedb..9909b2e8f2 100644 --- a/.agents/skills/retry-ci-job/SKILL.md +++ b/.agents/skills/retry-ci-job/SKILL.md @@ -55,7 +55,9 @@ automatically waits for an in-progress workflow run to finish before triggering 4. **Rebuild Full Build**: Re-runs the entire pipeline/workflow build from scratch via `bk build - rebuild` or `gh run rerun`: + rebuild` or `gh run rerun`. Prefer default in-place job retries (mode 1 or + 3, without `--rebuild`) when most Buildkite jobs have already passed, so + passing jobs are preserved; reserve `--rebuild` for mass early failures: ```bash ./.agents/skills/retry-ci-job/scripts/retry_ci_jobs.py \ --pr --rebuild diff --git a/.agents/skills/review-code/review-rust-prompt.md b/.agents/skills/review-code/review-rust-prompt.md index 07da9e3f14..6842f34ce5 100644 --- a/.agents/skills/review-code/review-rust-prompt.md +++ b/.agents/skills/review-code/review-rust-prompt.md @@ -39,8 +39,12 @@ in `git diff`: - Verify `Vec` and `HashMap`/`HashSet` are preferred over `BTreeMap`/ `BTreeSet` unless sorted traversal is an invariant, CLI types/`parse_args`/ `run_cli` are grouped cohesively (no top/bottom CLI sandwich), tests use - real temporary directories instead of mocked filesystem IO, and `rustfmt` - + `bazel test --config=fast-tests` pass. + real temporary directories instead of mocked filesystem IO, repeated + string tokens in `match` arms / registries / diagnostics use + module-qualified constants (`mod::CONST`), sibling crates share format + parsers instead of duplicating them, `#[cfg(unix)]` test fixtures avoid + leaving `vec![...]` (`clippy::useless_vec`) on Windows, and `rustfmt` + + `bazel test --config=fast-tests` pass. 5. Guardrails: Never nitpick generated `Cargo.lock` or `Cargo.Bazel.lock` formatting; only flag stale lockfiles or unvetted/outdated crate versions.