On Thu, 2006-01-19 at 09:48, Henri Yandell wrote:
> On 1/19/06, Allen Gilliland <[EMAIL PROTECTED]> wrote:
> > On Thu, 2006-01-19 at 05:50, David M Johnson wrote:
> > > Allen and I are planning to deploy Roller 2.1 to production sites
> > > next week, so we're going into code freeze mode today. At the same
> > > time, I'd like to try to get an "Apache Roller (incubating)" release
> > > out.
> > >
> > > I propose this plan:
> > >
> > > 1 - Write a what's new in Roller 2.1 page for the wiki
> > > 2 - Create Roller 2.1 release candidate (today or tomorrow)
> > > 3 - Testing through Thursday of next week
> > > 4 - Fix or defer remaining issues here: http://tinyurl.com/d3cwa
> > > 5 - Call for release vote late next week
> >
> > I want to make sure I understand the timing properly.  Do you want to 
> > release Roller 2.1 late next week when we deploy to blogs.sun.com?  or do 
> > you just want to vote then?
> >
> > Personally, I would rather wait a week after deploying to blogs.sun.com 
> > before doing the full Roller 2.1 release to the community.  I just think 
> > it's such a great way to get some real world testing on the release and 
> > usually if there are show stopping bugs we can find them and fix them 
> > within that week.
> 
> How's the codebase work for Sun? Do you have an internal version with
> changes, or do you deploy a pure Roller version?

We try and keep almost everything in the Roller subversion repository, as long 
as it's something that isn't truly unique to Sun.  The only changes to a 
standard Roller build is some additional jsps and some changes to skin that 
makes up the front pages.

> 
> ie: should it be:
> 
> * Build roller-2.1-rc1 and mention on -dev.
> * Deploy that to Sun, IBM, JavaLobby, individuals, wherever.
> * Test, debug, fix.
> * Release 2.1

That is basically the model we have been using for quite a while now, except 
that we don't ask IBM or JavaLobby to deploy to their sites.  I think 
blogs.sun.com itself provides enough of a test case to be quite sure that the 
release is stable enough.  Plus, I believe that IBM and JavaLobby make more 
customizations to the build, which isn't quite as good of a test case.

> 
> ?
> 
> Probably obvious that I'm wanting to emphasize that we make sure we
> don't tie the Sun deploy in as something special and hidden. Language
> can be fickle sometimes, so just being boring and confirming that.

Sure, that makes a lot of sense.  Deploying to blogs.sun.com isn't really a 
necessary step in the release cycle, we just do it because it provides a great 
test case and helps us find last minute bugs.  

-- Allen

> 
> Hen

Reply via email to