Comment by iLemming
15 hours ago
> What is wrong with markdown mode?
Nothing. It's just not Org-mode, and thus many great features of Org just not possible - sparse tree search; timestamps with built-in calendaring features; outline features like narrow-to-subtree; editing the subtree in an indirect buffer; executable code blocks; embedded LaTeX snippets; real tables with spreadsheet-like capability; checkbox statistics cookies - [/] and [%]; org-sort; column views; tags and tag inheritance; etc.
Anyone who's familiar with even a small subset of things that's possible in Org-mode and unlikely ever be supported in Markdown (just because there's no agreed format standard) have no confusion about it - Org-mode is absolutely superior. But superior doesn't mean universally better for everyone. The biggest drag of Org-mode is that it can never fully exist outside of Emacs. Yet Markdown can't ever fully cover its grounds (even in Emacs), simply because Markdown has to exist and work outside of Emacs as well.
Else where in this thread you can see my argument is that all org-mode features such as calendar dates, todo's, take your pick could be implemented separate to and utilize markdown (or any other format) for actually rending to. I am familiar with org mode but dislike how it sometimes pushes against simply using elisp for things like setting up todo nodes and possible todo values. couple that with the fact that one could in theory just render this all out to markdown instead and come up with some way to render todos into md.
Rendering Org in markdown is a solved problem, but wanting Markdown as the source is nuanced. It's difficult to achieve because there's no single agreement or convention - Org is not just markup syntax, its headings have records with types slots: TODO keyword, priority, tags, properties drawer, clock entries, etc. - Markdown has none of this. A CommonMark heading is a line of text; the spec does not even define a subtree, so folding, refiling and archiving need tree semantics layered on top.
So, you need to invent slots. Logseq did exactly that - all the org-mode specific lines get lifted verbatim into .md. Obsidian did it differently with emoji fields or Dataview inline fields (or whatever they called). Neither produces portable Markdown. GitHub renders those files as literal noise. The portability benefit that motivated the move evaporates the moment you add the metadata you need.
And then you'd have to build a specialized parser - there's no equivalent of `org-element` for Markdown. Sure, tree-sitter helps to some extent, but you're still lacking a semantic layer that turns nodes into records.
In essence - for Emacs to treat Markdown as Org-mode, we need to change Markdown, once that happens (if ever) Emacs can follow. But at this point it is already such a mess, it's crazy. Take front-matter alone - it's not CommonMark, it is a Jekyll convention that each tool does differently, and it is per-file, so your unit of thought is limited to become a file. That is the Obsidian/Denote/zk design, it works - but it moves your outline into the filesystem. You lose refile, subtree archive, narrowing, cut-and-yank of a subtree, and "this TODO belongs under that project heading". Org gives you both: file keywords at the top, property drawers on every heading, with inheritance down the subtree that front-matter structurally cannot express.
The idea was minor a minor mode not to modify markdown mode it's self. You have a point on export from org mode to md or html or whatever but my issue with these has been that MD never rendered the way I wanted it to (simply as possible) and I probably did not try hard enough to play around with the rendering parameters.
The thing I have found more and more is separate from the actual rendering format is wanting org structures that are simpler to manipulate pragmatically I know that one could can simply manipulate as they would any text-buffer but that can be a little convoluted.
All that is to say my original proposal was probably hasty and poor and it sounds like you know org inside and out and I don't so although I use emacs heavily and love it I sort of fell out love with org mode and having this conversation it seems like I probably need to go and read up on it a lot more in light of all of this thank you.
Markdown didn't even exist when org-mode was first developed and released with many of the features you describe. But also, once you start extending markdown to support all these things you start getting something that's not markdown, it's markdown-based but something distinct since you'd rapidly lose compatibility with other markdown tools if you start incorporating custom annotations into it.
I really don't dislike org-mode and respect it and you are right that org-mode was first.
The point is that markdown could have todo's or virtually any other aspect that org mode supports put into it without changing the markdown standard and layer org functionality and various features on top for example todo's could simply be italic's starting with todo: so `# todo:todo hello` could be parsed as a todo entry marked as todo.
1 reply →