Please stop incrementing the classfile version number when there are no format changes

Brian Goetz brian.goetz at
Mon Oct 14 20:36:11 UTC 2019

> That wasn't my subject line, it was Mike's. I was expressing support for the issues Mike raised, giving other reasons why what he was asking for was entirely reasonable.

You are correct, my apologies.  I meant that, but wrote the wrong name.

> You ask for consensus, but you are the gatekeeper for decisions here, so consensus doesn't actually matter -- your decision does. And I see there is no convincing you of the value of other opinions on this subject.

Obviously, consensus has to include the stewards as well.  But yes, your job, if you want to effect change, is to convince the platform stewards that something is wrong, that there is a reasonable solution, and that it is wrong enough that fixing it should be prioritized over the many other things it competes with for attention and resources.  

Yes, that’s not easy.  Careful analysis is hard.  Persuasion is hard.  But the road to solving this problem lies through making a clear and convincing argument about the problem, not just pushing a simplistic solution.

I actually believe there really is a problem here, but little energy has been spent exposing and clarifying it.  It should be well clear by now that I think the proposed solution is terrible (and you’ve not found a different position from any of the other experts), so you’re right that it is a waste of time to continue to argue _that solution_.  The sooner we stop trying to rehash the policy about class file versioning with the same old arguments, the sooner we have a chance to actually discuss the problem.  

More information about the jdk-dev mailing list