✍️ Parse Markdown and LaTeX outputs into AST

[agoose77]

This PR closes #2114 by parsing Markdown and LaTeX outputs into the MyST AST.

Note

Part of initiative #1026

[changeset-bot]

🦋 Changeset detected

Latest commit: 38c6351

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 3 packages

Name	Type
myst-cli	Patch
mystmd	Patch
myst-migrate	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[stevejpurves]

@agoose77 I am right in thinking that this does nothing to provide for markdown or latex rending when using in browser compute correct (we may already have LaTeX)? and if we in future support MyST rendering in outputs then in a similar vein they would not be rendered by the native Jupyter outputs areas we use for in-browser compute.

What is your longer term view on getting that support? are you thinking that we will move away from using native jupyter services code for in-browser support? or you think this will just stay unsupported`?

[agoose77]

@stevejpurves I think this is an area where a conversation will be easiest! I'm totally OK with there being some limitations between thebe and build-time experiences, and I think that there are a few different ways we can think about it.

[stevejpurves]

@agoose77 this is part of the issue, significant differences between build-time renderered / static content and results of in-browser may not be OK at all for a wider group.

I'm happy for a conversation if it can generate some output / assessment of the impact of the change that can be visibly discussed and decided on in a larger group. (i.e. significant enough to warrant an enhancement proposal)

[agoose77]

@stevejpurves can you elaborate on what you mean by "a wider group"?

I'm going to save this for our meeting because it will be a lot quicker to do, but in brief:

significant differences between build-time renderered / static content and results of in-browser may not be OK at all for a wider group.

I think this is the wrong reference frame. The existing status-quo involves major differences (not working) between static and Thebe compute outputs for Markdown and LaTeX. What we're talking about is inverting that, so that build-time is more feature-rich than compute-time. I think that's a reasonable (and necessary) trade-off.

the most important thing for us to do here is have clear user-stories and personas in mind for who is using MyST/Thebe, and how they are using them. In my mind, there are a lot of differences. Where they intersect, we can and should add ways to reduce the gap (such as opt-in ignoring recomputed cell outputs, etc.)

[stevejpurves]

@agoose77 let's try and get some written down output from our meeting, clear users stories would be a good output from that!

[agoose77]

Steve and I are preparing some user stories to frame this work in https://hackmd.io/H7ERZTycQlKy52JIlEWhDw