Hello Gennady,
Well, I would say that I have several goals, and they fall under categories 1 and 3.
I have existing code (both Java and Erlang) and I'd like to reuse as much as possible in order to make it work with Che. As a first step, I could live with having the Erlang machinery hidden behind Java APIs (like it is in the existing code) and wrap that Java so that Che can work with it, but in the long term it would be cool to expose the services directly from Erlang. Then even the regular Eclipse code could use those services.
The important part of your answer is at this moment "so far we do not have generic abstractions for it so really java specific things", so I won't lose time trying to find those abstractions. Do you have some plans/ideas about how these abstractions might look like? I tried to define these last year, but I have little experience in this area and the result was too horrible to look at. Where in your time plan would this find a place - 3 months, 6 months, a year?
best regards,
Vlad