Support Jupytext

[MarcoGorelli]

closes #668

[MarcoGorelli]

Fairly confident about this - I've tried it out on a few projects, and haven't noticed any issues

[MarcoGorelli]

Just noticed this from pymc-examples:

--- a/myst_nbs/gaussian_processes/GP-MaunaLoa.myst.md
+++ b/myst_nbs/gaussian_processes/GP-MaunaLoa.myst.md
@@ -4,13 +4,11 @@ jupytext:
     extension: .md
     format_name: myst
     format_version: 0.13
-    jupytext_version: 1.13.7
+    jupytext_version: 1.14.1
 kernelspec:
   display_name: Python 3
   language: python
   name: python3
-substitutions:
-  extra_dependencies: bokeh

probably a blocker

EDIT: resolved, reported to jupytext mwouts/jupytext#994

[MarcoGorelli]

Another blocker:

-kernelspec:
-  name: python3
-  display_name: python3
 jupytext:
   text_representation:
     extension: .md
     format_name: myst
-    format_version: '0.13'
-    jupytext_version: 1.13.8
-execution:
-  timeout: 300
+    format_version: 0.13
+    jupytext_version: 1.14.1
+kernelspec:
+  display_name: python3
+  name: python3

maybe, should just...not overwrite the header? not sure if:

• to consider this a bug in jupytext, and just defer to them
• to not overwrite the header, ever

tempted to go with the latter, though it's perhaps slightly trickier? return header from read_notebook and pass it to write_notebook. use jupytext.jupytext.writes instead of jupytext.jupytext.write, and then replace the header before writing

gonna go with the former, but will try to get around it where possible. check mwouts/jupytext#656 for the execution one

[MarcoGorelli]

I've tried this out on a bunch of repos, and added tests, and it seems to work fine

Tempted to just merge, more issues will probably only be discovered if this gets out and people start using it

[moshemoshe137]

Is support for mdformat being removed? It seems to disappear in 44d8ac8.

[MarcoGorelli]

I removed it from the docs, but you can still use it

[MarcoGorelli]

cc @OriolAbril as you're using jupytext, in case this is useful to you