Homebrew: Sandbox Cleanup Ripples Into a Path Length Bug
A pair of sandbox and bottle reliability changes landed this week, but one of them — nesting temp directories for better isolation — broke gpgme's socket paths on macOS and needed a same-week follow-up fix. Several smaller quality-of-life patches for cask permissions, formula creation, and dependency reporting are also in flight.
Duration: PT2M30S
Episode overview
This episode is a short developer briefing from Homebrew.
It explains recent repository work in plain language.
- Show: Homebrew
- Published: 2026-09-12T13:14:37Z
- Audio duration: PT2M30S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
It's September 12th, 2026, and this is Homebrew.
The big story this week is a good reminder that infrastructure cleanups have blast radius. Mike McQuaid's PR 23940 unified how sandboxes handle temporary directories across fetch, build, test, and install, giving each sandbox its own private space for temp files and sockets. Solid idea — but it added length to every…
Second theme: tightening up what Homebrew reports versus what's actually true on disk. PR 23938, also from McQuaid, stops missing bottles from silently downgrading supported configurations, bounds stale metadata refresh retries, and turns off automatic pkgconf reinstalls after failed fetches — all aimed at making…
Smaller but notable: PR 23942 moves Golden Gate dependent testing onto GitHub-hosted macOS 27 runners now that Apple's silicon image is available, cutting reliance on self-hosted infrastructure. PR 23948, still open, adds an ownership check before cask cleanup falls back to sudo chown, so users won't get an…
What to watch: keep an eye on formula builds that create Unix sockets in deep paths — the macOS length limit bit gpgme once already. And expect the info and cask permission PRs to land soon; both are…
That's…