Packaging and Deployment
summary
Shows concrete, repeatable packaging: use conda environments to install Python, numpy, scipy and switch between 2.7/3.x; conda uses a SAT solver and a simple tar.bz2 format with an info/index.json manifest. Describes channels, binstar.org hosting, miniconda/anaconda installers, conda build/conda package workflows and conda convert for cross‑platform packages. Gives numbers: ~200–230 tested conda packages, ~2,000 packages on binstar, conda pip build succeeds for many but fails when setup.py imports numpy before declaring it, so some recipes require manual fixes.
AI-written summary · about
chapters (13)
- 0:00Packaging and deployment intro
- 1:38Packaging in open source
- 3:41Dependency hell and conda
- 8:13Scipy and numpy struggles
- 13:30Conda core principles
- 22:30Installing miniconda
- 26:46Managing conda envs
- 36:47Demo of conda envs
- 41:20Conda package format
- 47:22Conda env benefits
- 52:15Building conda packages
- 1:06:05Hosting packages on binstar
- 1:11:41Cross-platform conda
related talks
02h 45mPublishing your Python project, the conda way
37mEmbrace conda packages.
29mManaging GNU Radio Installations with Conda
37mAutomated deployment of Python packages for development
09mBuilding a Hermetic Python Environment with Conda in Bazel
41mTaming the toxic python environment on your laptop
36mPython Packaging from Init to Deploy
31mFrom 0 to Continuous Delivery in 30 minutes.