Fork me on GitHub
#data-science
<
2020-01-06
>
kharbo07:01:14

Clojupyter 0.3.0beta1 released Clojupyter-0.3.0 is now released as beta1. The focus of the release has been to improve the structure of the implementation, with a much larger share of pure functions resulting in improved testability and clearer separation of layers. The new kernel communicates with its environment exclusively using core.async channels with ZeroMQ communication split off in a separate layer and communication with the internally spawned NREPL server based on a combination of interceptors and core logic implemented as pure functions. The release fixes reported issues #94 (Use tools.deps for dependencies), #103 (Imagemagick needed to install Clojupyter), #104 (Clojupyter not installable in non-base Conda environment), and #105 (Can’t interrupt the kernel on Windows), closing all open bugs in the project issue list. The release also includes support for Jupyter COMM messages (`COMM_OPEN`, COMM_CLOSE, COMM, COMM_INFO_REQUEST, COMM_INFO_REPLY) which are used in Jupyter to synchronize state between the front-end (Jupyter Notebook and Lab) and language kernel when using interactive widgets. The support will serve as the basis for exploring the possibility of adding support for widgets to Clojupyter. To signify the fairly fundamental changes and additions to Clojupyter in this release the version number is bumped to 0.3.0. The beta1 release is available on Anaconda Cloud with version=0.3.0beta1 and build=101 with label beta; it can thus be installed using the command:

conda install -c simplect/label/beta clojupyter=0.3.0beta1=101
Version 0.2.3 build 2 remains the default conda install until 0.3.0 comes out of beta. The beta1 library is available on Clojars with coordinates “0.3.0-BETA1”. The code is available on the release/0.3.0beta1 branch of the project repository. Reports of experiences using beta1 are welcome, please report to the Clojupyter issue list https://github.com/clojupyter/clojupyter/issues/109.