What should/shouldn't go into v3-4-test

[Tim Prouty]

Starting a new thread from "Re: Proposal: Add v3-4-test to the build  
farm, and revamp the build farm policy".


On Mar 24, 2009, at 12:53 AM, Andrew Bartlett wrote:

> BTW, what is the difference between 3.4 and master at the
> moment?

v3-4-test is release branch and master is a development branch.  The  
goal is to stabilize v3-4-test so it can be released in the upcoming  
months, while master remains as the current development HEAD.

I don't think the specifics of what this means are documented  
anywhere.  Karolin, maybe you can comment here.  As a start, I propose  
the following:

* Samba4 changes should not go into v3-4-test since v3-4-test is a  
samba3 release.
* Franky/merge-build work should not go into v3-4-test.
* v3-4-test commits should be limited to bug fixes and patches that  
will help stabilize the code.
* New feature development and large destabilizing changes shouldn't go  
into v3-4-test.

This means that v3-4-test and master will diverge, but it also means  
that v3-4-test will become a solid release.  Production quality code  
needs time and testing to stabilize, which is why these release  
branches exist.  We are doing a lot of QA internally on v3-4-test, and  
would like to see it continue to stabilize rather than be a quasi- 
development branch for the next three months.

Any thoughts?

-Tim