Adapt ForceEquations for Sphinx 8.2

[AA-Turner]

cc @mgeier xref sphinx-doc/sphinx#13346

A

[mgeier]

Thanks for this PR!

This works in most cases, but sadly not in all.

The problem is that there are multiple situations where math/LaTeX stuff can appear:

• Markdown cells, inline math
• Markdown cells, block math
• Code output cells

All three are handled separately and it's quite fiddly.

I tested it in #824 and this shows the problem with Markdown block math: https://nbsphinx--824.org.readthedocs.build/en/824/a-markdown-file-block.html

I don't know yet why that happens (I've actually expected problems with Markdown inline math, but that seems to work fine), and I have limited time to investigate ...

[AA-Turner]

I don't know yet why that happens (I've actually expected problems with Markdown inline math, but that seems to work fine), and I have limited time to investigate ...

The new behaviour includes MathJax if a math or math_block node is traversed in the document. Does the custom markdown parsing here construct math nodes?

A

[AA-Turner]

Alternatively, you could call app.set_html_assets_policy('always'), which forcibly includes the mathjax script in every page.

A

[mgeier]

The new behaviour includes MathJax if a math or math_block node is traversed in the document. Does the custom markdown parsing here construct math nodes?

It turns out that the block math (that doesn't work correctly) ends up in custom :nbsphinx-math:`...` roles, which I've defined there:

nbsphinx/src/nbsphinx/__init__.py

Lines 648 to 650 in 215bf0d

	.. role:: nbsphinx-math(raw)
	:format: latex + html
	:class: math

How can I make those be recognized as math?

Alternatively, you could call app.set_html_assets_policy('always'), which forcibly includes the mathjax script in every page.

OK, thanks, I think I can use this as work-around (slightly worse than the current work-around) in case we don't find a better solution.

[mgeier]

Out of curiosity: Why are you putting [:2] there?
Shouldn't it behave identically when you drop it?

[mgeier]

In #836, I have added example equations to the docs that would show (if rebased) that this PR doesn't work for block math.

[gabalafou]

I created an alternative solution:

• Fix nbsphinx_assume_equations for newer versions of Sphinx #840