Martijn Faassen wrote:
Philipp von Weitershausen wrote: [snip]
Right. Here's what we could do:
1. Copy Five's interface definitions over to Zope 2.8 (mostly to OFS.interfaces, I guess) where they are added as Zope 2 interfaces
2. Keep Five's (redudant) interface definitions. They can stay at their status quo (status Zope 2.7, that is).
3. Add <five:bridge /> calls for every interface so that Five's interfaces are automatically kept up-to-date with the Zope 2.8 ones. The bridges would override the ones defined in the module, potentially updating with newer definitions. The only thing that we need to take care of is fallback for Zope 2.7 where the Zope 2 interfaces don't exist yet.
So you would have the Zope 2.8 interfaces exist in the Five.interfaces module?
Well, no. Five.interfaces would stay as it is; it seems to be pretty accurate for Zope 2.7 (especially with yuppie's fixes, which should be merged to the Five-1.0 branch, btw). Some interfaces were added to Zope 2.8 and it would make sense to manage all of them in the Zope tree for the future, not the Five tree. However, when run within Zope 2.8, we want Five.interfaces to be most accurate, so we would install bridges in Zope 2.8 that bridge the Zope 2.8 interfaces to Five.interfaces. At least that was yuppie's latest idea andI think it's elegant.
If not, we do have a compatibility problem.
I don't think we will. Philipp P.S.: In case you're wondering why I haven't done any work on the Five wrt testing/i18n: My hard drive had a head crash, the laptop is in for repair :( ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.