As always, our intent is that JDK 7 will ship concurrently with a Java SE 7 JSR, and likewise for JDK 8 and Java SE 8, and also that there will be JSRs for Lambda and Coin.
For me, this is the real news - and good news - since uncertainties in this regard already have piled up a lot of fears and doubts (hint: TCK discussions). This in turn might also explain better, why Plan B release of JDK 7, even though almost feature complete, should take another nine months or so:
Properly setting up JSRs, and each one going through the JCP process (including drafts, reviews and all the ballots) alone takes time. As someone mentioned "employing 9 pregnant women won't deliver one baby after one month of pregnancy". And the recent difficulties in the relations of Google and Oracle respectively Apache and Sun don't make it easier to pass these JSRs quickly.
I also really like, what gili had to say about it here:
Please take as much time as necessary to deliver a well thought-out, polished solution.
I would love for Oracle to introduce a new community process whereby new features would be marked as "experimental" in JDK7 (such as Project Jigsaw) with explicit warnings that backwards compatibility *will* be broken. Then continue evolving the design until it *naturally* evolves to a steady state (very little modifications over a long period of time). We'd then mark the API as stable. I don't believe that it's technically possible to deliver a perfect design the first time. You'd be surprised how many developers will use experimental APIs in production code and let you learn on their dime. So long as you're up-front about expectations (i.e. this API is unstable) the more conservative customers will simply avoid them (which is fine because most of them are still using JDK5).