Bauke Scholtz

+1 but this is not backwards compatible and hence we did not do this during the initial work for 2.3 https://github.com/jakartaee/faces/issues/1355. I'm not sure if we can convince spec committee...

I checked and this specific change is allowed for a major version. - https://jakartaee.github.io/platform/CompatibilityRequirements - https://jakarta.ee/committees/specification/versioning/ And we have now control over old-TCK so we can easily adjust existing tests...

It should be registered on all views. The javadoc is indeed not explicit on this and there's no single word in the spec, below is all there is on this...

We should probably add a TCK to assert this.

It's already in 4.0 TCK? https://github.com/jakartaee/faces/blob/master/tck/faces40/extensionless-mapping/src/test/java/org/javaee8/cdi/dynamic/bean/ExtensionlessMappingIT.java#L107-L121 How did MyFaces pass that? Perhaps worked in WAR only not in Quarkus JAR?

Closing off stale issue. And it's already in TCK.

Interesting idea. What do we do with `UIComponent getFacet(String name)`? Deprecate it? And how about `Map getFacets()`? Should there be an equivalent returning `EnumMap`?

+1

I can't generify `Object eval(Serializable key, Supplier defaultValueSupplier);` as this makes existing impls incompatible. E.g. `ComponentStateHelper` doesn't anymore compile after this change. This thus breaks backwards compatibility. The correct procedure...

Yeah, but it's a public API and therefore the JEE spec committee won't accept this .. The correct procedure is deprecation. Unless there's an alternative I'm not aware of, @arjantijms?