Java-SDK artifact deployment strategies

[jason810496]

Why

How should user deploy their JARs?
We should know the full user-story ourself.

What

Whether to bind DagBundle with JavaCoordinator config.

[jason810496]

I just realized that user could also setup S3 Dag Bundle to pull the JARs to somewhere (if possible, need double check), then setup the coordinator jars_root properly if they insist in deploying with DagBundle.

[jason810496]

Another approach pointed out by TP in #65958 (comment).

User might compile all the Java Dag as one Jar and point the main_class to the "single big JAR bundle" that contain all the potential Dags.

---

There're also more context in #65958 (comment) and the following reply.

[uranusjr]

We made the decision to not commit to a specific workflow for now. This is OK since the feature is experimental for now. We’ll try to gather community opinions and improve things. (This is also mentioned in user-facing docs.)