Ticket #13822 (closed PLIP: wontfix)
Zope 4 for Plone
Reported by: | thet | Owned by: | |
---|---|---|---|
Priority: | minor | Milestone: | 5.0 |
Component: | Unknown | Version: | |
Keywords: | Cc: |
Description (last modified by thet) (diff)
Proposer: Johannes Raggam
Seconder: None as yet
Draft
Motivation
Currently Plone depends on Zope2 2.13.21 and a quite old set of Zope packages. For example, in Plone 5 cordev buildout, zope.component is fixed to version 3.9.5 from 2010-07-09 - thats years and 8 versions behind the latest release 4.1.0. Or zope.interface from 2011-08-20, again 8 version behind the latest release 4.0.5. By switching to Zope 4 and catching up with current versions, Plone could not only gain from that (getting towards Python 3 compatibility, WSGI support, ...) but also become a driver behind Zope development - a role which Plone should take.
Proposal & Implementation
- Depend on Zope 4 and latest ZTK and zopeapp KGS.
- Depend on ZODB (4) instead of ZODB3.
- Have a "Plone next" Jenkins setup which does not have any version fixes to see what's going to break in future releases.
Here is a working buildout: https://github.com/collective/experimental.zope4 To install, read the README.rst
To install Plone in this environment, call http://localhost:8080/@@plone-addsite?site_id=Plone because ZMI is not fully functional yet.
Deliverables
Risks
This PLIP is somehow experimental as we cannot know, how far we can go. Zope 4 is also in Alpha state at the moment.
With this PLIP there would be even more code migration tasks for integrators. We also have to expect some data migration issues if we depend on ZODB 4.
It's quite questionable, if it makes sense to deliver this PLIP for Plone 5. But it makes sense to start working on it to be able to deliver it for maybe Plone 6.
Change History
comment:1 Changed 2 years ago by thet
- Version 4.2 deleted
- Type changed from Bug to PLIP
- Milestone changed from 4.x to 5.0
comment:3 Changed 2 years ago by davisagli
I like the idea of trying to use recent releases of Zope packages, but I'm nervous about depending on Zope 4 since there is no active development on it or apparent commitment to release it.
comment:4 Changed 2 years ago by gforcada
While maybe is too risky, as David points out, ZODB4 I think is a must, as it's a fundamental piece to keep update and get improvements from.
This jenkins job you planned, maybe pinning it to the latest released versions would be a good approximation as to how many tests do fail with this newer versions so to get a rough estimation of the amount of breakage that would imply moving, eventually, to Zope 4.