The project leadership certifies that the APIs in this release are "Eclipse Quality".
Hudson 3.1 offers an opportunity to add some significant new features into Hudson aimed at the stated target of supporting large enterprise CI and CD operations. Performance and stability are also an ongoing focus.
Although this release has maintained full compatibility with the previous releases there have been significant internal changes in memory management. This does not have an impact on Hudfson users but it may have an impact on downstream Plug-In writers outside of Eclipse who want to piggyback on the improvements that we've made. Importantly, all existing plug-ins will still be compatible, it's just thay they m,ay not be as effecient as they could be from a memory consumption perspective. We will be publishing guidelines to help plug-in authors with this task after the release.
The one outstanding security issue we know about relates to the use of the "Remember Me" option on the login screen if Hudson is running on plain HTTP rather than HTTPS. This is discussed in: Bug 412488 - SPRING_SECURITY_REMEMBER_ME_COOKIE set on non secure connections.
Fixing this potential problem will actully involve an upgrade to a newer version of the Spring Security library and unfortunately that will in turn require a corresponding upgrade to the rest of the Spring libs. As usch this is a major project needing various IP approvals and extensive testing. This project is targeted for Release 3.2.0
The the meanwhile, Hudson installs in non-secure environments should be run in https mode to prevent any possibility of the cookie being sniffed off of the wire.
During the lifespan of this release we have completed the huge information migration task from the hudson-ci.org wiki to the Eclipse wiki which is a significant milestone. A new book covering Hudson 3.0 is also just about to be published by McGraw Hill: Hudson Continuous Integration in Practice. This will be available in time for JavaOne this year
The major change in this release does involve some possible usage changes in the way that Hudson is used as it adds a whole new management concept in the form of Team support. The good news is that this is an optional new feature that Hudson users can eveolve to use if they need it, and when they are ready. Existing Hudson users can upgrade to this new release and not enable team support unless they want to. To help users uptake this new feature we started to write a series of how-to articles to take them through the process.
Other new features have no impact on the actual use of Hudson.
No features EOLed
During the cycle of this release we have brought Bob Foster of Oracle on board as a new and highly valuable team member and committer. Bob has been responsible for a large chuck of the Team Concept feature in 3.1.0. Additionaly we've received external contributions from Roy Varghese on the new memory management modelf or Hudson. Roy is not yet a committer on the project but his contribution has been significant for this release. Another new face in the community is Latha Amujuri, who did an excellent job on depth testing the Team Management feature for the project.