Question on Slate

[CanRau]

Hey, I,m following portable text basically since it's public and was really excited when I saw the Editor announcement 🥳

We're currently using Lexical for our platforms editing experiences, still in a pretty barebones state and after looking into PTE's docs I think it looks very nice though I'm concerned about Slate as I'm only reading discouraging things especially when mobile and Android support are required.

I never actively used Sanity so I'm curious what your experience is and thoughts are.

Currently ProseMirror (or TipTap) and Lexical seem to be the only viable choices when not building in-house from scratch.

Last thing I just read again on why not Slate
https://news.ycombinator.com/item?id=31814983

[christianhg]

Hi @CanRau!

Thank you for the kind words and apologies for the late response. Really appreciate that you've been following what we are doing with Portable Text!

The Portable Text Editor is almost as old as Sanity and the decision to use Slate was made many years ago when the project received active maintenance. It's only recently we've pulled PTE out of Sanity Studio as a standalone editor and really treating it as such.

Slate does have Android support, but I think you are right that ProseMirror (that both TipTap and Lexical areTipTap is based on) could be the more robust choice today. It's admirable how that project still chugs along. I can't say we have had many, if any, bug reports related to poor mobile support, but that might change once PTE gets in the hands of more people.

I do really appreciate what Slate set out to do, and I don't think any other solution comes close to providing as flexible a framework to build a rich text editor on top of (although I hope PTE can be exactly this for many people). Slate's Transformation engine, for example, is great! And thus it's a shame maintenance has become more sporadic. For reference, I've looked a lot at Lexical and while it looks really good, I don't think we could build PTE or the Behavior API we wanted on top of it.

That being said, we are aware that relying on Slate might come to an end one day, and for the same reason we are actively working on encapsulating the framework and building core PTE functionality on our own APIs. One example is the Behavior API which you can use to hook into editor events and change the editor behavior: https://www.portabletext.org/concepts/behavior/.

The introduction of the Behavior API has meant that we, internally in PTE, have been able to strip the Slate usage way back. We still rely on Slate as our foundation for DOM event handling, but the overall dependency on Slate is shrinking and it's not unfathomable that we'll one day move forward either by forking Slate or choosing a different foundation.

I hope the above was somewhat helpful.

Again, apologies for the late response.

Christian

[CanRau]

Hey Christian, thanks for getting back and don't worries all good 🤝

Uh haha okay yea makes sense, hadn't considered that it's not actually a new project 😅

Lexical is actually not based on ProseMirror, it's a completely new project by Meta to replace their abandoned DraftJS editor. Lexical is still pretty early and maybe even more than ProseMirror a framework to build anything on top, haven't used ProseMirror, but from what I understand and how it seems give's you, Lexical, at least for now, gives you mostly only "lower-level" tools to build on top and then you can get inspiration and/or copy from their playground, also Ghost's Lexical based editor Koenig as well as Payload's are great resources.

Also if you ever consider between ProseMirror (based) or Lexical there's some interesting reads

• Rich Text Editors in Action: Stress Test On Lexical and ProseMirror
• Why I rebuilt ProseMirror’s renderer in React
  

After reading the 2nd article I think personally I'd use react-prosemirror if choosing ProseMirror in a React codebase.

Also there's interesting discussions in Lexical's Discord regarding the comparison articles etc

But yea ProseMirror's schemas are really interesting and compelling, Lexical has a pretty extensive discussion around schemas and ways how you can achieve the same on a lower level.

I'm sometimes thinking if I should've picked TipTap, something else based on ProseMirror like BlockNote or ProseMirror directly, but we're already pretty invested and I'm generally pretty happy, tho community and resources are still more limited and "everyone" seems to use TipTap, at least tech-twitter 😅

From a quick glance I'm pretty sure you could model your Behaviors with Lexical's Listeners and/or Node Transforms possibly keeping you API and "just" changing it's implementation

Also by the way I opened a discussion on Lexical regarding PT as underlying state representation, tho only got a few replies on their Discord

[christianhg]

Lexical is actually not based on ProseMirror

Argh, sorry about that. Of course it's not!

As I said, I see a lot of good things in Lexical, and I'd be more than happy if it has more feature parity than I thought!

Thank you for opening that discussion on PT in Lexical! We noticed, and we appreciate it! ♥️