Hi - I see that XWiki Enterprise 3.0 Milestone 1 was just released a week ago. I also found the roadmap at http://enterprise.xwiki.org/xwiki/bin/view/Main/Roadmap I can't tell from the announcement however what the intended use is... Is this something that is likely to blow up ("alpha" quality)?

At what point in the roadmap does it make sense for a new install to try a 3.0.x release instead of 2.x release?

In addition to stability of the software, what about stability of any APIs - are those frozen in M1? If I adopt M1, will M2 automatically upgrade from M1, or are automatic upgrades only assured for migration from a stable 2.x release?

When 3.0 final is out, will work on 2.7 basically stop?

Will the current release of XWord work with XE 3.0M1?

Regarding extensions, the listing at http://extensions.xwiki.org/xwiki/bin/view/Main/ says nothing about version compatibility. I'm guessing that because of the velocity upgrade, trying any extension with heavy velocity usage would be a gamble with XE 3.0M1 ? Or is there an automatic "smoke test" report somewhere of what extensions are certain to not work?

Sorry for all the questions, but I seem to have hit xwiki at a cusp in its lifecycle so I'm in a quandary about which version to start with.

On 01/30/2011 07:38 AM, Mark D. Anderson wrote: All our releases are as stable as possible. It is very unlikely that something will "blow up", as we're using Continuous Integration to assure that at almost any point XWiki is stable (https://secure.wikimedia.org/wikipedia/en/wiki/Continuous_Integration) The main difference between a milestone and a final release is that during milestones new features introduced are not yet tested in many different environments, so they might look badly in certain versions of IE, or there might be problems on a different database, but this usually only affects new features introduced in that milestone. Something that was working in the previous stable version should still work in the new milestone. Between a milestone release and the final release, most of the development efforts go into testing and polishing the new features, so that the "stable" 3.0 version is as stable as possible. The "bugfix" releases, like 3.0.1 or 2.6.2, rarely solve critical bugs in their equivalent major release, but usually backport bugs solved during the future release. These bugs aren't blockers, just things that need to be ironed out, so instead of waiting for the 2.7.5 release, better try the 3.0 release, since it is almost as stable, but with even more bugs fixed and some new features. If by "try" you mean installing it in production, then 3.0. If you need really fast something that is only present in the latest milestone, then it should be fairly safe to use a milestone in production, but you should upgrade as soon as the stable version is released. If by "try" you mean start preparing for a production deployment, the answer is, in my opinion, the first release candidate (3.0-rc-1). It's what will be in the final 3.0 release, and it gives you time to prepare the environment and test the wiki before the stable release is out. If by "try" you mean start exploring for an eventual deployment, then you can already start that, although the best point will be the last milestone (in this case it looks like 3.0-milestone-2), since it will have most of the features planned for the final 3.0 release. If by "try" you mean testing it and trying to help the community, then you can take any snapshot (from http://maven.xwiki.org/snapshots/org/xwiki/enterprise/ ) and test it, test the new features introduced, report any bugs encountered, or even things you don't like (usability, interface), things you'd like to be better (better functionality of the features), etc. Scriptable APIs are stable, so most of the Velocity code that you write should work in future versions. The APIs that are broken (as listed on http://www.xwiki.org/xwiki/bin/ReleaseNotes/ReleaseNotesXWikiEnterprise30M1… for example) are usually internal APIs (SPIs), that you should be concerned about only if you're writing custom Java components. 3.0 is an exceptional case, since we're actually removing deprecated public APIs, but those APIs have been deprecated for a long time, and it's highly unlikely that they are still used. Platform upgrades should work pretty well between any two versions, provided you do the upgrade correctly. The closer the two versions, the safer the upgrade. Again, what could break is custom components that rely on internal SPIs. No, the rule is that we maintain two old branches in parallel to the trunk. When 3.0 will be out, the trunk will already be on 3.1-SNAPSHOT, and the previously maintained branches will become 3.0 and 2.7. At that moment in theory the 2.6 branch will be dropped. We sometimes make exceptions from this rule, for example right now we're maintaining 3 older branches (starting with 2.5, see https://svn.xwiki.org/svnroot/xwiki/platform/core/branches/ ). If it works with 2.7, it should work with 3.0 as well, since there have been no changes in the REST or XmlRpc systems. Some extensions list their minimal required version (for example http://extensions.xwiki.org/xwiki/bin/view/Extension/Multipage+Export+Appli… ). There's no maximum version, since we can't predict when will extensions stop working. Most of the extensions there are not officially maintained by the XWiki committers, so we're not testing them for compatibility with each new release. If there's an extension that doesn't work, leave a comment on its page, or send a mail to the users list, or you can try to fix it yourself and upload a new version (it's an open wiki). -- Sergiu Dumitriu http://purl.org/net/sergiu/