Peter Donald

It looks like there are a few things still need to be addressed in this patch. I would recommend getting Bazel building and then look at the CONTRIBUTING.md document in...

Rather than including `gecko_dom.js`, a better option may be to re-arrange the closure externs so that the non-browser specific are in `w3c_*.js` or `whatwg_*.js` files. For example once google/closure-compiler#3275 lands...

If you touch this in the near future, just be aware I may be trying to migrate the standardized part of the [Selection](https://w3c.github.io/selection-api/#selection-interface) interface to a separate extern :)

Have you considered ‘vue4j’ - the project I am working on is ‘react4j’ for exactly this reason:)

I should note that this is not a blocker (assuming google/closure-compiler#3399 is merged) as the typing of the override method is identical to overridden method and thus I attempted to...

It is perfectly valid closure typing to tighten the types in an override. i.e. you imagine going from `!TrustedHTML|string` in the base class to `!TrustedHTML` in a subclass and this...

The reason is roughly that all native static fields on types go through this `__Constants` generation. See https://github.com/realityforge/jsinterop-generator/blob/7dc62bb37fa0716f8b7fdef13798e094a4f8c818/java/jsinterop/generator/visitor/FieldsConverter.java#L125 If my assumptions above are right and we are willing to let...

Would you consider a pull request that added an option like `--use_constants_overlay_for_interfaces_only` that removed the `__Constants` class unless it was needed to support static fields for an interface. It would...

That would certainly be a better solution but I don't currently have the cycles that I would need to get that right. (Or the inclination as we want to move...

After experimentation, we have decided to move directly to J2CL so no longer see this as a priority. So I will close the issue until plans change.