What features would you like to see?

[szymonmaszke]

As the title says, post your ideas below.

[szymonmaszke]

Current ideas in order of importance and amount of additional work needed:

• Move plugin infrastructure to jupytext
• Major source code refactor
• Bug fixing (notebooks with spaces and other open issues)
• Binding for inserting above/below/here python block
• Support for hisaknown/jupyterkernel.vim
• ultisnips integration (other snippet plugins?)
• Run kernel independently from notebooks (connect to remote kernels as well) (help? xd)
• Closer integration with Jupyter notebooks (maybe using jupytext?)
• Automatic notebook scrolling like (see an example here)

[malramsay64]

So I had a go at implementing support for jupytext here. While I did manage to get it working, my opinion is that to properly implement jupytext support requires a rethink of the workflow around using it. Like having the markdown alongside the notebook, that is the the paired representation in jupytext.

An alternative approach is to use a .py representation so that all the autocompletion, code linting, code formatting etc, all works without much additional configuration.

[DerWeh]

Thanks for your great project.
I think the focus should be to get a more interactive and seamless interaction with the notebooks. Currently it is not really workable. If I edit something in vim, I have to reload my notebook with lots of hassles, then rerun the calculations ... Currently it is still to cumbersome.

jupyterkernel.vim seams like one possible direction. I am currently still working with the plain old vim-ipython and simple .py files. While it is not nearly as pretty, it simply provides a better workflow (and seamlessly integrates with language-servers).
I think in order to be snappy, the plugin needs some possibility to run a cell from within vim. Till then this plugin will just be a way to write some additional text to notebooks.

I am looking forward to your future work, keep going.

[vnijs]

You may be interested in "https://github.com/mwouts/jupytext" which makes it possible to "link", e.g., a python or R(markdown) script to a jypyter notebook. You can edit the notebook and the linked file will be auto-updated. You can also edit the linked file, e.g., in (space)vim, and the notebook will be updated as well. Very nice!

I also saw the follow issue posted on 'spacevim' (https://github.com/SpaceVim/SpaceVim/issues/1854) but I don't think this has happened yet but it is listed as a "milestone".

One thing that would be fantastic in vimpyter (and spacevim) would be to send Python (or R) code to a running jupyter kernel. This is something that atom + hydrogen currently does a fantastic job at. See link below

https://nteract.gitbooks.io/hydrogen/docs/Usage/GettingStarted.html

You can even connect to jupyter kernels (e.g., R and Python) running in local or remote docker containers which is something I use all the time. Works really great ... from atom ... but I would love to have this work form (space)vim.

https://github.com/nteract/hydrogen/blob/master/docs/Usage/RemoteKernelConnection.md

[szymonmaszke]

Yeah, I think removing notedown dependency and changing it to jupytext is the road I will take when time allows.

[wsdjeg]

@szymonmaszke Thanks for your plugin, I am spacevim author, and I will add new layer for ipynb filetype, and I will fork this plugin, make some change, of cause will sent PR to this repo.

[anosillus]

Hello, I like your plugin.
But I wanted to use python lint tools and some python plugins.
So I prefer like to use ipynb-py-convert tools instead of notedown.
ipynb-py-convert convert an ipynb-file into a python file.(not a markdown file).
I prefer like ipynb-py-convert to jupyterxt. It is faster and easy to use, convert style looks better.

I have folked your plugin and using ipynb-py-convert tools yet.
If user's want to use ipynb-py-convert. I can send pull requests.

[tvatter]

See #21

[mikh3x4]

Is work on this plugin still happening? I'm really excited about having vim - jupyter integration and I'm trying to decide on what workflow is best