Ricardo Wurmus writes: > Christopher Baines writes: > >> I've tried builing locally with the following command, and I get >> something that suggests there may be a problem, but I'm not sure where >> to go from here. >> >> $ ./pre-inst-env guix build --no-grafts --rounds=2 --check ruby >> >> guix build: error: build failed: derivation `/gnu/store/37rb11ivy4w6ydn0px386k158ixbgczz-ruby-2.5.1.drv' may not be deterministic > > You can add “-K” to keep the second build. Then you can compare the two > directories with diffoscope. > > (You only need “--check” if you already have a local build; “--rounds=2” > is for an initial build that should be performed twice.) Thanks, this works really well. Comparing the two outputs with diffoscope --exclude-command stat shows that there are some timestamp issues, some with the ri documentation, and others with the gemspec's for the bundled gems. Taking the same approach for ruby 2.4, ruby 2.5 actually seems a lot better. Ruby 2.4 seems to have some file ordering issues as well (although maybe they just didn't crop up for my build of Ruby 2.5) as well as some of the binary files differing.