On Mon, February 7, 2011 00:08, Doug Lea wrote: > On 02/05/11 07:56, Mark Wielaard wrote: >> So, does OpenJDK implement/define JavaSE or not? > > No; OpenJDK will produce stable releases of projects such as > "jdk8" that might just so happen to implement some other > bodies' definition of "Java SE 8". > > I continue to think that one can and should omit reference > to any such other bodies or corporations in these parts > of the bylaws. So for example, no reference TCK licenses > or Oracle branding agreements pertaining to products.
Agreed. > However, the GB is required to ensure that contributors have > access to "infrastructure": code, docs, tests, reviewers, etc > necessary to successfully contribute. Yes, this is very important. All participants should have equal access to code, specs, tests, etc. And all contributors should make sure their contributions are complete and provide all these. > Which is likely to > include those tests that comprise the TCK But here I disagree. There is no way that the current TCK setup can be integrated with an open community approach. As long as the TCK is proprietary, only available to an arbitrary subset of participants, based on the whims of a commercial corporation, requires signing an NDA that prevents effective exchange of information between all participants, the TCK should be avoided. (If only to shield the governance board members who undoubtedly will become very frustrated like has happened with the JCP.) > as well as > participation in a (hopefully revamped) "compatibility > review" process that occasionally rejects (occasionally for > good reason :-) bug fixes on the grounds of backward > compatibility. And so on. Here I do agree again. There should be clear rules for how anyone can participate in this process and make sure every participant does provide tests for such backward compatibility guarantees we decide to make through our codebase. Cheers, Mark
