The approach here is a bit novel i.e treat each block as a separate editor instance (contenteditable) and somehow wire-up cursor movements to behave as if it's one big editable area.

But it comes with its downsides as well. For example cross block selection doesn't work.

If the entire page is one big editable area, then it becomes difficult to embed complex blocks like "kanban views" and calendar.

I guess we should think beyond contenteditable at this point and separate rendering layer from input layer[1] - sort of like how Google Docs has built its editor.

But writing a rich text editor is not just a text-editing problem. It essentially needs you to build a layout engine itself (like webkit) that knows how/when to recalculate and draw the affected parts (render objects) when the rich-text changes. Why? because baking in tables and image-wraps and other complex resizable blocks affect other elements around/after it.

Bottomline: we need a custom built layout engine + text renderer + input handler(that respects accessiblity) + selection handler to build an absolutely powerful rich-text editor. I'd like to start a open source project in this direction.

[1] https://twitter.com/jitl/status/1649761057953415174