On Tue, Mar 15, 2005 at 04:57:00PM +0100, Christian Heimes wrote:
I really *love* to have Five and parts of ZopeX3 in Zope2 but I don't like how it is happening. Zope 2.8 is starting to stablize and still contains some critical bugs like the incompatibility with old style BTrees and you are going to add a heavy bunch of new features to it.
[Could you point me to the issue or mail describing the old-style BTree problem? I may have run into it under another name or something.]
I would prefere this way:
* Get Zope 2.8.0 out ASAP
* Branch of Zope 2.9 from 2.8 when 2.8 is in rc phase. 2.9 should be the target for ZopeX3 and Five only, no other features so you can get 2.9 out some month afte 2.8.0.
"Release often, release early and one step after another". You are going to make the same ugly and bad mistake we made with Plone 2.0.0. :(
I completely understand where you're coming from and I share the worries. Some things mitigate this problem: * integrating Five doesn't actually imply many significant changes to Zope 2.8; it'll just be its own product. * integrating zope 3 only means some integration in the build script for Zope 2. It's just a Python library otherwise. * we're separating the work on Zope 2.8 and Zope 2.8 + Five, so we can back out from this plan if we want to, later, and go the 2.8 and 2.9 route anyway. So, in effect, we're not actually *creating* any new features. We're also not doing any deep integration work (Five already has that work). The features already exist and we're just distributing them along with the standard Zope 2.8 distribution. We're not ambitious about new stuff at all. I hope that is somewhat different from the path Plone took for 2.0. Regards, Martijn