"Complex" Katex formulas can't be rendered using MDX

[nikoladev]

I don't know the exact terms to describe it, but "complex" Katex formulas prevent MDX from rendering. (Feel free to change the title into something more specific)

I've figured out a workaround by passing the formula as a string to a custom component that runs the string through katex.renderToString and then passes the output to dangerouslySetInnerHTML. That might help in pinpointing where the problem occurs.

There is a minimal reproducible example here with instructions on how to make it fail as well as the workaround in action.

[ChristianMurphy]

Sorry you run into some trouble @nikoladev

Looking to further reduce the problem I tried
Katex + MDX 1 runtime: https://codesandbox.io/s/katex-example-mdx-1-937f6
and Katex + MDX 2 runtime: https://codesandbox.io/s/katex-example-mdx-2-em1jg
Both worked 🤔

gatsby-plugin-mdx has forked parts of the MDX pipeline, which has been problematic in the past.
It's managed by the Gatsby team independent of the MDX team https://github.com/gatsbyjs/gatsby/tree/master/packages/gatsby-plugin-mdx
I suspect, this has something to do with gatsby-plugin-mdx

[nikoladev]

@wooorm It does and I did try it before, even toggling plugins one by one to see if anything sticks. But unfortunately it didn't. Thanks for thinking along :)

Though, seeing you here reminds me that I'm using remark-math@3 because gatsby-plugin-mdx depends on remark@10. Off the top of your head, do you think that this could have anything to do with that combination? Would remark@13 combined with remark-math@4 magically solve this?

I've also opened an issue at Gatsby's repo.

[wooorm]

remark@13 w/ remark-math@4 would not solve this.
remark@10 works with remark-math@3. I believe everything, as Christian pointed out, is working okay on our side, but something in gatsby is acting up?

[nikoladev]

Yeah, I figured that was the case. Was just hoping for some deus ex machina :p

Thanks!

[wooorm]

I did see that several unified projects that Gatsby is using are from 3 years ago. E.g., mdast-util-to-hast is at v3, whereas we’re at v10 already

[nikoladev]

Thanks for investigating! I've mentioned it in the issue