Thought experiment: a Leo web app based on Jupyter

[Edward K. Ream]

The idea is to fork the IPython/Jupyter code and make it the basis of Leo as a web app.

This has been a great thought experiment, regardless of what happens. It has given rise to the following ideas:

1. Extend the vr plugin to handle pyplot scripts.  A prototype is already running.
2. Create a server that will showcase .leo files and that will allow users to share .leo files.
    This page could be called a prototype, but much more work would be necessary.
3. Add features to Leo that would facilitate web development.

It's been useful to study the Jupyter code, for at least the following reasons:

- The code uses a range of interesting technologies, including the Tornado server.
- We can assume that the code is close to best practices.
- Little code is written in javascript. Most is in Python or .css or jinja2.
- Jupyter has an interesting security model, which allows sharing on sites like this one.
- Jupyter allows lengthy computations to be run asynchronously.

Summary

- Adding Leo's features to Jupyter notebooks would be a lot of work for little gain. 
- It is easy for Jupyter users to find and share Jupyter notebooks safely.
- We should investigate ways of having Leo users find and share .leo files via the web.

Your comments, please.

Edward

[Edward K. Ream]

On Wednesday, September 28, 2016 at 11:29:32 AM UTC-5, Edward K. Ream wrote:

> We should investigate ways of having Leo users find and share .leo files via the web.

There are many to share already: posting to SourceForge, GitHub, etc.  It would also be possible to set up a dedicated web server, but I'm not sure that would add much.

I suspect that security issues, as discussed here in the Jupyter docs, are much more serious in a web app than they are in Leo.  I think we are pretty safe provided we know who actually contributed each .leo file.  This will be the case when sharing via GitHub.

Otoh, I could be totally mistaken about this. Any comments?

Edward