Request
Add an aarch64-linux (a.k.a. arm64-linux) binary to the release assets, alongside the existing x86_64-linux and arm64-darwin ones.
Current state
As of v0.4.0, the release ships three assets:
scip-clang-x86_64-linuxscip-clang-dev-x86_64-linuxscip-clang-arm64-darwin
So ARM is covered on macOS but not on Linux. Users on aarch64-linux (ARM servers, AWS Graviton, ARM CI runners, ARM dev machines) currently have to build scip-clang from source — which means compiling LLVM/Clang locally, a heavy step that a published binary would remove.
What seems to be missing
Looking at .github/workflows/release.yml, the build-and-upload-artifacts matrix has three jobs, all on ubuntu-24.04-…-amd64 or macos-14 — there's no Linux ARM runner in the matrix. Adding one appears to be a single matrix entry targeting an arm64 Linux runner (e.g. the GitHub-hosted ubuntu-24.04-arm, or a graph-team arm64 runner mirroring the amd64 one), since:
- the upload step already globs
./*-release-artifacts/* and is platform-agnostic (per the "Download everything to avoid spelling out the different platforms" comment), so a new artifact would be picked up with no further change;
- the build itself is the same Bazel/LLVM path, just on aarch64.
The main cost is presumably the runner (LLVM/Clang-from-source + LTO wants a sizeable box), not the workflow change.
Why it matters
A stock aarch64-linux binary would let ARM Linux users install without a local LLVM compile, matching the out-of-the-box experience x86_64 and macOS ARM users already have.
Happy to help test an arm64 build if that's useful.
Request
Add an
aarch64-linux(a.k.a.arm64-linux) binary to the release assets, alongside the existingx86_64-linuxandarm64-darwinones.Current state
As of v0.4.0, the release ships three assets:
scip-clang-x86_64-linuxscip-clang-dev-x86_64-linuxscip-clang-arm64-darwinSo ARM is covered on macOS but not on Linux. Users on
aarch64-linux(ARM servers, AWS Graviton, ARM CI runners, ARM dev machines) currently have to build scip-clang from source — which means compiling LLVM/Clang locally, a heavy step that a published binary would remove.What seems to be missing
Looking at
.github/workflows/release.yml, thebuild-and-upload-artifactsmatrix has three jobs, all onubuntu-24.04-…-amd64ormacos-14— there's no Linux ARM runner in the matrix. Adding one appears to be a single matrix entry targeting an arm64 Linux runner (e.g. the GitHub-hostedubuntu-24.04-arm, or a graph-team arm64 runner mirroring the amd64 one), since:./*-release-artifacts/*and is platform-agnostic (per the "Download everything to avoid spelling out the different platforms" comment), so a new artifact would be picked up with no further change;The main cost is presumably the runner (LLVM/Clang-from-source + LTO wants a sizeable box), not the workflow change.
Why it matters
A stock
aarch64-linuxbinary would let ARM Linux users install without a local LLVM compile, matching the out-of-the-box experience x86_64 and macOS ARM users already have.Happy to help test an arm64 build if that's useful.