Re: [ELPA] New package: markdown-ts-appear
Stéphane Marks <[email protected]>
| Newsgroups | gmane.emacs.devel |
|---|---|
| Message-ID | <CAN+1HbohZV0qkoHpq17uN07-A2xoV3mOgMxVBq2oPcN7P3d-jQ@mail.gmail.com> |
On Thu, Sep 3, 2026 at 11:52 AM Philip Kaludercic <[email protected]> wrote: > 强生 <[email protected]> writes: > > > Hello, > > > > I would like to submit markdown-ts-appear to NonGNU ELPA. > > > > markdown-ts-appear is a minor mode for Emacs 31's built-in > > markdown-ts-mode. It hides Markdown markup while reading and reveals > > the smallest semantic element at point while editing. It also provides > > optional asynchronous MathJax previews and visual decorations for code > > blocks, callouts, block quotes, links, and tables. > > > > Repository: > > https://github.com/Thysrael/markdown-ts-appear > > As the code is AI generated, I'll just leave a few high level comments: > > - You don't need the :group annotations in user options, they fall back > to the last defgroup. > > - Can you set the default value of > `markdown-ts-appear-enable-math-preview' to be (package-installed-p > 'Mathias)? > > - The logic do draw a box around a code block uses these BOX DRAWINGS > characters that don't always look good with all fonts. I would > recommend re-thinking this, or just plain avoiding using text to draw. > > Otherwise the code seems OK from a skim, albeit very verbose. I cannot > judge just from reading (and cannot test it since I don't build Emacs > with tree sitter) but I have difficulties believing that the entire 2400 > lines of code are all there to match the necessary complexity of the > problem you are trying to solve. Please keep in mind that just because > you use an LLM, doesn't mean that humans don't want to read and change > the code as well without being exhausted, so it valuable to keep the > codebase "human sized". > > Other than that I have CC'ed the markdown-ts-mode maintainers Stephane > and Rahul for further comments. Sadly the features of this package are > currently not candidates for upstreaming, but perhaps they have a better > feeling if the package suggestion makes sense in the current form. > Specifically I would be interested if we could avoid the advice block at > the end of the file. > Thanks. I'll take a deeper look when I have time. At first glance, the name itself probably should be "markdown-ts-reveal" not "appear". I'd probably want to separate showing strict Markdown markup when markup is hidden from the other features like math and evil/meow integration. I agree it appears (pun intended) that there is a lot of code for what it does and not all of it is idiomatic ELisp, but that is separate from whether it works or not. + Andrea Alberti who has a lot more LaTeX experience than I do and he's contributed to the core mode with depth of knowledge. I seem to recall that he may have a similar "reveal" functionality somewhere and has, I think, contributed a to-do list item to improve markup editing while markup is hidden.