We’ve just published IPython Parallel 7.0, a pretty big release. I’d love your help testing and feedback on the new features.
Major new changes:
- a new Python API for starting/stopping/signaling/restarting clusters
- Support for sending signals to remote engines
- a new BroadcastView for “do on all” operations with dramatically improved performance for large numbers of engines, c/o Tom-Olav Bøyum’s Master’s thesis at UiO.
- JupyterLab extension for managing clusters, based on dask-labextension
- Improved streaming of output, and interactive waits for results
- Solved longstanding registration timeout issue with large numbers of engines starting at once
and lots of other fixes and improvements along the way. More information in the changelog.
Special thanks to Bodo for their support of IPython Parallel this year, giving me time to do things I’ve been meaning to do for years!