First off, thank you for these great comments, I’ll try to address them one by one:
- “jupyter pixiedust install” should only be used if you want to use Spark locally and doesn’t want to deal with the arcane steps of configuring the kernel. To be clear, if you don’t care about Spark then don’t use this step. Using pip install pixiedust is sufficient and will work the same.
- Spark is optional. PixieDust works the same with a plain Python kernel.
- This is great feedback, can you please create a GitHub issue: https://github.com/ibm-watson-data-lab/pixiedust
- Yep, good point. Please create a GitHub issue.
- This is unfortunately a known problem that is very hard to prevent due to a limitation on how Notebooks work. It also happen the same using the command line pdb. I recently added a button that lets you quit the debugger (will be available in 1.1.12) which should mitigate the issue.
Hope that helps