Repository navigation
ci: install qemu on Ubuntu 26.04 runners; fix the rustdoc link error - #409
Conversation
Two failures in CI that no code change caused. 1. neon-simd/parity-qemu (and the four qemu jobs in simd-matrix): the ubuntu-latest image is now Ubuntu 26.04, where `qemu-user-static` is a virtual package with no installation candidate, so `apt-get install` exits 100 before anything is built (PR #408, a docs-only change, went red this way). New scripts/install-qemu-user.sh installs `qemu-user-static` where it still exists, else `qemu-user` plus links for the `qemu-<arch>-static` names the parity scripts call. Every qemu step calls it. Both branches were run on Ubuntu 24.04 (QEMU_INSTALL_FORCE_PLAIN=1 takes the 26.04 branch), and scripts/neon-parity.sh passes through the linked plain qemu-aarch64. 2. docs/nightly has been red on master since at least 2026-10-29: `table[0]` in a doc comment in src/simd_stride.rs was read as an intra-doc link (rustdoc -D warnings). It is now in backticks; `cargo +nightly doc --no-deps --features approx,serde,rayon` with RUSTDOCFLAGS=-D warnings builds. CLAUDE.md's install instruction is corrected in place. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMQJzQD5KvNm8BtbsWG
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars.
 |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
The qemu install fix works on the new image (qemu 10.2.1, both -static names resolve), and it exposed a second effect of the same image change. realization/avx failed with "Fatal glibc error: CPU does not support x86-64-v3": ubuntu-latest is now Ubuntu 26.04 amd64v3, whose glibc requires x86-64-v3, so no host-linked binary starts under `qemu -cpu SandyBridge` or `-cpu Nehalem`. Those two arms exist to run on exactly such CPUs. They are pinned to ubuntu-24.04 with the reason inline; moving them back needs a static (musl) build of the parity binary and the lib test binary. Every other job stays on ubuntu-latest; the cpu-guard job emulates Haswell (x86-64-v3) and is unaffected. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMQJzQD5KvNm8BtbsWG
|
The new qemu install worked: the log shows qemu 10.2.1 with both The next commit pins those two jobs to Generated by Claude Code |
|
@codex review Generated by Claude Code |
|
Codex Review: Didn't find any major issues. Can't wait for the next one! Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
Two CI failures that no code change caused.
1. qemu jobs:
qemu-user-staticis gone on the new runner imageubuntu-latestis now Ubuntu 26.04. There,qemu-user-staticis a virtual package with no installation candidate, soapt-get installexits 100 before anything builds. PR #408, which changes one Markdown file, went red onneon-simd/parity-qemuthis way. The same step is in foursimd-matrixjobs.scripts/install-qemu-user.shinstallsqemu-user-staticwhere it still exists. Otherwise it installsqemu-userand links theqemu-aarch64-static/qemu-x86_64-staticnames the parity scripts call.QEMU_INSTALL_FORCE_PLAIN=1taking the 26.04 branch;scripts/neon-parity.sh→NEON SIMD parity: PASSthrough the linked plainqemu-aarch64.2.
docs/nightly: red on master since at least 29 Octtable[0]in a doc comment insrc/simd_stride.rswas read as an intra-doc link, and rustdoc runs with-D warnings. It is now in backticks.cargo +nightly doc --no-deps --features approx,serde,rayonwithRUSTDOCFLAGS=-D warningsbuilds.Not fixed here
mirion master is cancelled at the 6-hour job limit on every push. This is why each master CI run takes about 6 hours and ends red. It needs its own look: which tests miri spends the time on, and whether to filter them.CLAUDE.md's install instruction is corrected in place.
Generated by Claude Code