167 wires and counting

$ follow Go

Keep up with Go in about 3 minutes: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

or

fair warning: these emails are deeply technical. diffs, version numbers, CVEs, benchmark deltas. if that's not your idea of a good read, this isn't your newsletter.

Folds into your digest — weekly by default, monthly if you prefer. Unsubscribe in one click.

$ status

wire 2026-10-17
stories 43

© 2026 RepoJournal Home Showcase How it works Privacy

$ the-wire · showcase

Mach-O linking fix on Darwin ARM64, and a 2.7x SGEMM baseline

By RepoJournal · Filed · About Go · Composed from the cited sources · methodology

The day's real work is all in the compiler and runtime toolchain: a Darwin ARM64 external-linking failure gets patched, ARM64 SIMD grows 64-bit element reductions, and an SVE SGEMM benchmark puts SIMD at roughly three times scalar throughput.

cmd/link: don't add label symbol for unallocated symbols golang/go

by Cherry Mui

External linking on Darwin ARM64 could fail after the Mach-O symbol and string tables were moved earlier: gensymlate saw those pseudo-symbols, tried to emit label symbols for the large ones, and hit a missing section because they belong to none. The fix filters unallocated symbols out of gensymlate. Anyone cross-compiling or external-linking Go binaries for darwin/arm64 should pick this up.

simd/archsimd: add an SVE SGEMM test and assembly benchmark baseline golang/go

by Junyang Shao

A new archsimd test implements Float32 C = A*B for arbitrary m, k, n with the SVE API and checks it against a scalar version, alongside an assembly-kernel benchmark baseline. In the committed numbers, the SIMD path runs 64x64x64 at 9.688 GFLOPS versus 2.954 GFLOPS for scalar, though the SIMD kernel has 32,893 ns per op of unexplained time between the two listed benchmark lines.

simd: add ARM64 64-bit element Reduce* methods golang/go

by Alexander Musman

ReduceSum now exists for Int64x2, Uint64x2 and Float64x2, built on ConcatAddPairs(x).GetElem(0), with ReduceMax and ReduceMin added for Float64x2 only. The Int64/Uint64 min and max reducers are absent because VSMAXP/VUMAXP lack 2D support, so integer 64-bit width reductions are not part of this API.

_content/tour: build and run code snippets as go1.26 modules golang/website

by Alberto Donizetti

Locally the tour writes a synthetic go.mod to run snippets in module mode; with no go directive it fell back to go1.16 semantics and the generics snippets stopped building. It now declares go1.26. Running the tour from source is usable again.

_content/blog: add post for gopls interactive refactoring golang/website

by Hongxiang Jiang

go.dev now carries a post on gopls interactive refactoring, pitched at developer experience rather than internals and scoped to that feature; its date was later set to 2026-10-16, and previews run at localhost:6060/go.dev/blog/interactive-refactoring. Elsewhere the long tail is cosmetic: pkgsite corrected a quoted Docker Compose default that made the frontend read the literal "true", with the s...

Quick answers

What shipped in Go on October 17, 2026?
The day's real work is all in the compiler and runtime toolchain: a Darwin ARM64 external-linking failure gets patched, ARM64 SIMD grows 64-bit element reductions, and an SVE SGEMM benchmark puts SIMD at roughly three times scalar throughput. In total, 43 commits landed.
Who contributed to Go on October 17, 2026?
9 developers shipped this update, including Alberto Donizetti, Cherry Mui, Ville Vesilehto, Alexander Musman, Junyang Shao, Hana Kim, Zachary Spar, and Nicholas S. Husin, and 1 more.
What were the notable Go updates?
cmd/link: don't add label symbol for unallocated symbols, simd/archsimd: add an SVE SGEMM test and assembly benchmark baseline, and simd: add ARM64 64-bit element Reduce* methods.