The Classics Never Go Out of Style: An Empirical Study of Downgrades from Bazel
summary
Analysis of 542 open-source Bazel adopters finds 61 projects (11.2%) abandon Bazel after sustained use, with a median 638 days of adoption. Thematic coding of 212 abandonment documents surfaces seven recurring reasons: perceived maintenance/troubleshooting (≈20%), platform support/interoperability gaps (≈26%), perceived parity with native toolchains (≈22%), contributor onboarding barriers (≈14%), experimentation (≈4%), coordination issues, and community trends. Abandoners move to alternatives ranging from Go Build, CMake, Make to language-specific managers like SwiftPM; 8.7% of projects instead migrate to Bazel.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:49About the speaker
- 4:38Bazel’s promise and challenges
- 8:13Studying Bazel abandonment
- 10:42Prevalence of Bazel abandonment
- 15:04Reasons to abandon Bazel
- 22:50Replacements for Bazel
- 23:55Projects upgrading to Bazel
- 25:43Summary and lessons
- 26:49Q&A
related talks
20mAdopting Bazel in a quickly scaling organization
10mBazelCon vision and wrap-up
18mDesigning a language Agnostic CI using Bazel queries
22mThrough the (Open Source) Looking Glass
58mResponsibly Abandoning Open Source Projects
01h 01mGuide to Destroying Codebases the Demise of Clever Code
16mMigrating Twitter's monorepo from Pants to Bazel
19mBazel persistent workers