alt Hacker NewsThere's really no way to decouple slots for shadow roots.
For slots to work you need a container for the slots that the slotted elements do not belong to, and whose slots are separated from other slot containers. Otherwise you can't make an unambiguous relationship between element and slot. This is why a shadow root is a separate tree.
Agreed. The way I explain it is: suppose you have a `<super-table>` element, and you have a child slot called, for example, `<super-row-header>`. Presumably you want to write some JS to transform the slotted content in some way, decorating each row with the header the user provided.
But, if you do that, what happens to the original `<super-row-header>` element that the user provided? Maybe you'd want to delete it…? But how can you tell the difference between the user removing the `<row-header>` and the custom element removing it in the course of its work?
What you'd need is for `<row-header>` to somehow exist and not exist at the same time. Which is to say, you'd have one version of the DOM (the Light DOM) where the slot element exists, and another version of the DOM (the Shadow DOM) where the `<row-header>` element doesn't exist, and the transformed content exists instead.
It's clever, I guess, but the drawbacks far outweigh the benefits.
Client-side components inherently require JS anyway, so just use your favorite JS framework. Frameworks can't really interoperate while preserving fine-grained reactivity (in fact, Shadow DOM makes that harder), so, just pick a framework and use it.