Skip to content

Latest commit

 

History

History
46 lines (37 loc) · 2.91 KB

clojupyter-and-conda.md

File metadata and controls

46 lines (37 loc) · 2.91 KB
Raw

Clojupyter and Conda

Anaconda is the recommended way to install Jupyter and so we can reasonably assume that many or most Jupyter users have Anaconda on their machine. Clojupyter includes Conda support to make it easier for Jupyter users to get started with Clojupyter/Clojure.

Core Conda Concepts at a high level

Documentation for building Conda packages can be here.

At a high level Conda enables you to create a reusable recipe for building a package. A package is anything you install using your package manager. A conda package is a compressed tarfile containing the module to be installed plus instructions on how to install it. You use conda build to build a package. Conda channels contain packages and conform to a standard structure and contain an index of available packages. Conda is able to install from channels and uses the indexes in the channel to solve for requirements and dependencies. A conda package has a version number and conda package versions have a build number which enable you to build versions multiple times and keep them distinct from a package management point of view.

Clojupyter's use of Conda

Clojupyter's use of Conda is rudimentary at best, sufficient to get started distributing Clojupyter using Anaconda Cloud. Key features of Clojupyter's conda support:

  • As simple as possible, enough to get started
  • 3 platforms are supported: Linux, MacOS and Windows, all in 64-bit variant
  • Clojupyter's version number is the basis of the Clojupyter package version number (major, minor, incremental, qualifier)
  • Build numbers start at 1
  • The Anaconda Cloud channel is simplect
  • The Clojupyter package can be installed, upgraded and removed using conda
  • Only a single Clojupyter kernel can be conda-installed at a time into a given, conda-managed environment (see below)

Clojupyter kernels in a conda-managed environment

Conda tries to deliver a complete package management solution most of which we believe is not relevant for Clojupyter users, the focus of Clojupyter's support is leveraging the Anaconda Cloud for distribution and enabling the user to do conda install on the supported platforms - beyond that the implementation is as limited as possible. To keep things simple, and to align with the version-and-buildnum concepts of conda, we support only a single conda-installed version of Clojupyter into a conda-managed environment at time. However, you are not necessarily precluded from using multiple conda-installed Clojupyter versions since conda enables you to create multiple conda environments: You should be able to keep multiple versions/builds on your machine using conda's environment management facilities, see Conda's documentation for Managing Environments for more details.