This topic (modifying bean meta data at runtime) has been coming and going since CDI 1.0.

As we are moving to Java SE with potentially more dynamic env, I think interesting to see if there is a real need to explore it.

It's a very disruptive feature and obviously costly to add to the spec, so I think we should start gathering proof that it would bring more power and new usage to CDI before discussing all the problems it could raise.

I propose to g that way:

1) Launch a public consultation to ask our user if they are interested and more than all give use real use case where it would solve a problem

2) Analyse these use cases to see if they are real or as some people think are only a lack of knowledge of the spec

3) From there we could either left this aside if nothing relevant comes out or start discussing the constraint around this feature.

Wdyt?

Antoine