Fast and reproducible tests, packaging, and deploys with...
summary
Claims that Pantsbuild makes tests packaging and deploys fast and reliable by running every step inside hermetic environments that isolate process side‑effects. It models inputs and outputs as immutable digests stored in content-addressable storage (lmdb under the hood) so repeated runs are memoized and remote caching can reuse results across the team. The build system materializes only needed files into temporary dirs, cutting redundant work and making CI and local runs both faster and verifiably correct.
AI-written summary · about
chapters (12)
- 0:00Start
- 1:53Pants is a build system
- 3:45Pants 2 is a Python tool for Python People
- 4:59Better performance with the same underlying tools
- 6:58Remote cacheing and execution
- 10:39Processes are Rules too
- 11:59Dependency versions
- 13:36Tools don't need to run in compatible environments
- 13:51input files + environment output artefacts + side effects
- 23:33Content-Addressable Storage
- 26:04Modelling a process
- 27:00Execution environment Host OS Predictable env Materialised digest
related talks
18mReproducible Builds with Bazel
51mTread Lightly When Building and Testing Your Python Project
50mDesigning tests for ML libraries
28mBuilding reproducible Python applications for secured environments
39mA Pythonic Approach to Continuous Delivery
28mTesting Complex Python Dependencies With Docker Containers
26mCLI, CI, DIY: Crafting Python Based Dev Tools in a DevOps Vacuum
34mAutomating infrastructure at SA Home Loans with Python (and friends)