Draft JEP: Incubating Language and VM Features

[Brian Goetz]

There is nothing in this JEP that is inconsistent with developing an incubating feature in the context of a component JSR.  The choice of a component JSR vs pushing spec changes directly through the platform JSR is an orthogonal one.  

The alternative approach you propose is one we could have taken, but we deliberately chose not to.  While I understand your concerns about pollution, you cite exactly what we’re trying to avoid: 

> - every Java vendor has the possibility to freely decide which JSR he
> wants to support (until the JSR get's merged into the SE specification
> of coarse)

The word for this is ‘fragmentation”; that’s exactly what we don’t want.  The goal here is to eliminate a specific risk: that we get a permanent feature wrong, and we miss the chance to fix it before it becomes permanent legacy.  Incubation is a delicate balance that enables this without encouraging fragmentation.  

So yes, in an alternate universe, we could have proposed this — but we think this is the wrong choice, so we didn’t propose it.