[Grok-dev] Grok 1.1 the plan

[Martijn Faassen]

Hi there,

Here's a plan for Grok 1.1, getting it to release.

Grok 1.1 release focus:

* use the Zope Toolkit as a base instead of Zope 3.4

Secondary features:

* new grokui.admin

* extracted grokcore.model (has this landed on trunk already? don't see 
a changelog entry or entry in set.py yet)

* getting rid of old style zopectl template (I think this is a change in 
grokproject)

We have two plans on how to get to Grok 1.1 final. Plan A:

* try to release a Grok 1.1 that doesn't depend on most zope.app.* 
packages at all. They're not needed in the path in order to be able to 
start a Grok app. We release a backwards compatibility package that does 
pull them into the path (and we maintain versions for these packages in 
zopeapp.cfg)

* this means we need to detach zope.app.wsgi and zope.app.publication 
from zope.app.testing and zope.app.zcmlfiles. Along with 
zope.app.appsetup (already detached) we're going to keep these zope.app. 
packages in Grok for now. See my previous mails for more details (in the 
"losing momentum" thread, and the recent sprint report).

* We might eventually choose to rename these packages to zope.wsgi and 
zope.appsetup and zope.publication (or whatever) though that doesn't 
appear necessary for us to lift the dependencies. It would just make the 
zope.app.* world clearly a world we don't need to care about anymore.

Plan A depends on how much work we spend on it. We will review the 
situation by february 16, about two weeks from now. If we have made 
unsufficient progress on this topic by then, we'll go for plan B.

Plan B:

* release Grok 1.1 without all dependencies cut. This means we will 
effectively (through zope.app.testing and zope.app.zcmlfiles) pull in 
*all* zope.app.* packages.

* We *will* depend on the ZTK however.

* Grok 1.1 is however still useful as a transition. Developers can start 
to convert their code to change imports everywhere to use non zope.app.* 
as much as possible.

* Grok 1.2 will then be the ZTK-less version. We'll try plan A again.

Plan B is still a good plan. But I hope we'll see some more efforts to 
loosen our connection with zope.app.* in the coming weeks. The further 
we get with plan A, the better.

Grok 1.2 looking forward:

* we need to land the Martian inheritance changes

* we need to land the grokcore.view view registry changes

* Martian testing changes (the Manuel work)

Regards,

Martijn

P.S. Thank you Souheil for making me write this plan down. It's much 
appreciated, and the clarity is really needed.