Hey Ihor, Apologies for the delay, I’ve been recovering from an RSI and had to save my limited keystrokes for other things. (Take care of yourself folks!).
On 2025-07-13 09:57, Ihor Radchenko wrote: > I feel that using preview machinery will be quite an abuse. Agreed, overloading `org-preview-latex-process-alist' for MathML is certainly a kludge. I only meant to suggest that the structure is similar in that it could be a multi-step process, i.e. not all MathML converters can convert directly to `<math>' fragments, and those that export HTML would need to be processed further. > IMHO, providing an export backend sounds much cleaner (and more > flexible). No argument there. My concern is that a separate export path is a surprise to the user. IOW, I could adjust `org-html-with-latex' to get MathJax or images for math, but I’d have to change how I export to use MathML. It feels a bit inconsistent to me. The `html' option of `org-html-with-latex' seems closest to what I want (using `org-latex-to-html-convert-command'). However, it doesn’t seem to use any caching mechanism, doesn’t act on environments, and sees only a fragment instead of the minimal document the image methods see. > I think that ox-odt can benefit from better mathml export most. IMHO `ox-html' (and even `ox-md') would benefit most since they can’t really do MathML at the moment. I don't have a use for ODT, but I’m happy to improve it along the way. What kind of improvements did you have in mind here? One thing I noticed is that `org-odt-with-latex' can’t use image generation methods from `org-preview-latex-process-alist' like the HTML counterpart. Markdown could also benefit from images and MathML (e.g. a README on sourcehut). Best, -- Jacob S. Gordon jacob.as.gor...@gmail.com Please avoid sending me HTML emails and MS Office documents. https://useplaintext.email/#etiquette
