[Asterisk-Dev] stable/feature freeze?

[Paul]

Steven wrote:

>On Wed, 2005-05-18 at 11:57 -0400, C. Maj wrote:
>  
>
>>On Wed, 18 May 2005, Russell Bryant waxed:
>>    
>>
>>>If I knew a way to make everyone stop referring to it as the "stable"
>>>branch and just refer to it as the "1.0" branch, then I would happily do
>>>it.  I agree that "stable" is a bad name, and it just causes stress for
>>>me.  I maintain the branch on my own, and pretty much the only time
>>>anyone talks to me about it is when they want to complain about their
>>>specific problem and want to know why it was a problem in the first
>>>place, and then why it wasn't fixed yesterday.
>>>
>>>Russell
>>>      
>>>
>>You could come up with some clever names like Debian does
>>for 'woody', 'sarge', 'sid' (stable,testing,unstable), so
>>maybe for asterisk 'masterflash', 'starlight', 'dogpound',
>>etc.  Regardless, nice work on going thru bug after bug and
>>marking whether the patch gets applied to 1.0, especially
>>since most people are only writing patches against head.
>>    
>>
>
>Debian has different goals to satisfy by using those names though. Sid
>has been deemed always unstable. Woody is the 3.0 branch and Sarge is
>the future release with pending release number. The reason is to
>mitigate the problems when the new stable branch gets released. If you
>track the release name Woody, it won't change when Sarge becomes the
>stable release and some other Toy Story character becomes the testing
>name. 
>
>Of course there are a few experimental trees that lay over unstable.
>  
>
We could get ideas from debian, but obviously things would be different. 
We don't have 15000 different packages comprising an asterisk distro and 
we don't have a lot of different hardware architectures. With debian if 
a package builds and runs okay on the i386 arch but not on sparc it is 
considered a release-critical bug. There is a whole set of rules for 
moving things from unstable to testing and I think we would only need a 
small subset of those rules.

I think the goal should be to have a testing branch that encourages 
people to test. I have spare server hardware. If there were debian 
packages in an asterisk testing branch, I would be installing them on a 
test server and filing bugs(the type that maintainers appreciate). I 
might install testing on a production machine, but would never allow 
that one to be automatically upgraded. We only need to build for debian 
and the most popular rpm-based distros to get a lot more testers. With 
unstable a lot of users never get it built and running so the only thing 
they are testing is the makefile.