How I measured our local GitLab runner usage against GitLab.com compute minutes
Using glab and the GitLab API to turn a self-hosted runner's job durations into GitLab.com SaaS compute-minute estimates.
Blog
10⁻¹X Engineer stuff.
Seven ways to host a GitLab CI runner in 2026, priced and compared
A comparison of home hardware, GitLab.com, AWS, ARM64 cloud, local AMD hardware, and Hetzner for a monorepo running 2,000 CI jobs a month.
Removing the brace-expansion patch after moving minimatch to v10
A recurring SyntaxError traced to a brace-expansion patch and an ESM export-shape mismatch: how moving minimatch to v10 removed one patch while a minimatch@3 compatibility patch remained necessary.
Clearing thirty Renovate merge requests in one pass
A retrospective of clearing a 30+ MR dependency backlog in a repository with JavaScript, Java, Python, and Terraform projects, including CI, supply-chain, and typecheck changes.
Tty-aware shell wrappers for OSC 8 hyperlinks in eza, ripgrep, and fd
Shell functions that add OSC 8 hyperlink flags to eza, ripgrep, and fd only when stdout is a terminal, so clickable paths never leak into pipes or files.
eza --hyperlink eats the next positional argument
eza v0.23.5 made --hyperlink a clap optional-value flag, so a bare --hyperlink swallows the next argument; attach the value with =.
What Changed in pnpm Since 11.7 Until 11.13.1
Seven pnpm releases shipped between 11.7 and 11.13.1, and the one that maps to my config is a peak-memory fix for the minimumReleaseAge and trustPolicy resolution path.
AWS auth in a Gondolin VM: why SigV4 breaks the secrets bridge
The Gondolin bridge substitutes placeholders in request headers, but the AWS SDK signs the request over the placeholder value. The substitution happens too late and every call returns SignatureDoesNotMatch. The fix is to mount the host's ~/.aws/ read-only instead.
Why Lychee checked 336 links in CI and 6,633 locally
An unquoted recursive glob made Lychee scan seven shallow HTML files in GitHub Actions while the quoted local command reached the full generated Astro site.
Dependency caching belongs at the registry layer, not the package store
Caching the package manager store treats a symptom. A pull-through proxy registry is the layer where cross-runner dependency caching actually compounds.
_EXPERIMENTAL_DAGGER_CACHE_CONFIG is inert in Dagger 0.21
A source trace explains why Dagger's BuildKit cache bridge works on 0.20.1, disappears in 0.21, and leaves type=gha with no effect.
Practical cache strategies for ephemeral Dagger CI
A decision guide to persistent engines, host-side cache bridges, registry proxies, and cold execution after Dagger 0.21 removed BuildKit remote export.
Hardcoded tool versions in a GitHub Actions step drift from mise.lock under Renovate
Renovate bumps mise.lock but a version pinned inside a workflow run step stays stale, and MISE_LOCKED=1 then aborts. Install tools by name and let the lockfile decide.
TypeScript 7 is fast, but the Compiler API is gone. Here is what that breaks.
The Go rewrite delivers 10x faster type-checking. It also removes the programmatic API that eslint, Dagger, and every framework template type-checker depends on.
Shipping visual regression on React Native: Maestro baselines, an RNTL v14 migration, and killing Storybook
How I closed two testing gaps on a React Native/Expo app, visual regression and CI integration, by adding Maestro baselines, migrating RNTL to v14, and deleting Storybook RN entirely.