At 08:22 AM 7/20/2005, Bruce Snyder wrote:
I'd like to hear some opinions from community members who are not
committers. If you are reading this message and you are not a
committer, PLEASE SPEAK UP! We want to hear your opinion on this
matter.
The more branching and integging you do the less time you have for actually
fixing bugs and developing features.
The two questions users are asking right now is (a) where is a version that
works and (b) where is the version that is CTS compliant. I think for most
people Geronimo is a new thing that they are struggling to understand and
the less obstacles to their understanding the better (and the more user
momentum geronimo will get). People will try the "stable" milestones first
and IMO these MUST work. On the other hand this is still an alpha release,
so people are more forgiving about compatibility and missing features.
I actually don't think a branching model will help you much here (and it
will slow you down), its really down to the discipline of individuals.
Periodically freezing the codebase in order to put out a stable milestone
is IMO a better approach, but it will only work if people really are
disciplined about what they commit during the freeze. Once the feature set
is permanently frozen then that's probably an appropriate time to branch.
To answer the question in the subject, I would push it out to M5 if its
going to be very de-stabilizing, but include it if its just going to be
incomplete. I doubt many users are going to try switching at this stage
(unless there are outages in Jetty that tomcat doesn't have).
My $0.02
andy
