text-unicode VS slate

We had to handroll our own OT implementation inhouse (based on https://github.com/ottypes/text-unicode) since we had built the system for it already. I suspect we would've used or forked CodeMirror's collab package if we were starting today.

Also huge fan of yjs, but the implementation is not compatible with codemirror 6, only version 5 (AFAIK)

Well said.

I've had a lot of conversations with javascript engineers over the years who've argued to me that well tuned JS will be nearly as fast as the equivalent C code. I've written plenty of little toy benchmarks over the years, and in my experience they're partly right. Well written JS code in V8 can certainly run fast - sometimes around half the speed of C code. But a massive performance gap opens up when you use nontrivial data structures. Nested fields, non-uniform arrays, trees, and so on will all cripple javascript's performance when compared to C's equivalent of simply embedding nested structs. If you couple clean C data structures with allocation-free hot paths from arenas, the performance of C will easily hit 10-20x the performance of the equivalent JS code.

From memory my plain text based operational transform code does ~800k transforms / second in javascript. The equivalent C code does 20M/second. The C implementation is about twice as many lines of code as the JS version though.

(The code in question: https://github.com/ottypes/text-unicode/blob/master/lib/type... / https://github.com/ottypes/libot )

- it creates a layout based on rows and cells, so it support multi-column layout - each cell can contain a different "cell-plugin", - richt-text editor based on https://github.com/ianstormtaylor/slate is built in and comes with its own plugin system. It can do weight, italic, block-types, alignment and lists and can be extended as you want (even with elements storing data and interactive components) - you can create custom cell plugins based on a schema (or custom control ui) and a component that should be rendered - it stores an object tree that represent it, not html. It therefore can contain any react component, which is great if you want to allow your editors to add interactive components or components that you already built as part of the app - i carefully optimized for SSR and bundle size, so no editor ui is rendered nor loaded. editor ui is only loaded on the client if you disable readOnly. (lazy loading) - it mainly tested with nextjs, since i used it for content-heavy pages. - its not yet tested with react-server components, but it should actually work in read-only mode

Finally there's Slate and Lexical which are super powerful in terms of customizability and extensibility. They're great options for when the editing experience plays a major role in the product.

Slate, as per its documentation, is a completely customizable framework for building rich text editors. Therefore, it doesn't offer a feature-rich text editor but instead provides tools to build one. Let's create a component called Slate and see what the Slate editor looks like.

Link to Repo

Word of warning about Slate: I love the API and the design goals, but it appears to suffer from some fundamental issues. We were experiencing issues similar to this one and a team of multiple 10+ year experienced frontend devs couildn't figure out what was going on. I had to completely rip out a feature we had built with Slate and had to reimplement a new version from scratch with Lexical. So far we have no issues other than those inherent to rich text editing.

We're trying to choose between Lexical and Slate at work. Do you have any examples that would be similar to this? https://github.com/ianstormtaylor/slate/blob/main/site/examp...

If you are going to customise a ton of functionalities and/or implement new functionality I suggest using SlateJS. If not, have a look at Sun editor.

You definitely need to give Slate (https://github.com/ianstormtaylor/slate) a try - the best editor framework I've used.

Slate: Looks very promising, but it's for React. (Someone has floated the idea of making it framework-agnostic, but the maintainers haven't committed to that goal yet.)