As a starter this documentation may help: http://ipython.readthedocs.io/en/stable/development/index.html
I'm not involved in Blender development, so you shouldn't take my views as anything official. But my opinion is that a Jupyter integration for Blender would be useful for users who write Python addons, but don't want to compile the sources themselves. Using Blender as a Python module would require compilation. Implementing the kernel protocol through asyncio would mean it can be run from any blender binary distribution. The Jupyter notebook server would then have a kernel config which starts Blender with a script, this script installs asyncio and listens on the specified port. Blender could be run with or without GUI. Here is one implementation of such an asyncio based kernel: https://github.com/takluyver/aiokernel/blob/master/aiokernel.py The problem with that is the zeromq protocol implementations for asyncio. The former code uses something called "zentedeschia", and there is also aiomzmq and I think pyzmq has another one. But they all depend on pyzmq which in turn requires compiled extensions. I haven't been able to load pyzmq into Blender at all, so far. Other than this little import, there should be nothing in the way of Jupyter/Blender integration. _______________________________________________ Bf-committers mailing list [email protected] https://lists.blender.org/mailman/listinfo/bf-committers
