Jean-Francois.Doyon@CCRS.NRCan.gc.ca wrote:

Brad,

About the branch thing ... That's basically the idea!

The bigger problem here is how to "manage" this both internally and from a user/administrator perspective.

This paradigm only really makes sense in the CMF world anyways, and I want to focus on basic Zope before moving up to the extra layer CMF represents. I'm also worried getting something like this to work with CMF woudl require an ungodly amount of monkey patching :P

The main things with CMF would be:

1) "published" type states vs. other states: how does one tell/define the difference ? How does this information get passed to/from the publishing system or the storage layer ?

Possible solution: Have the subversion/zope interface interact with DCWorkflow. When an object is changed, consult DCWorkflow settings and determine what state the object is in. If the state is deemed "published", copy the object into the publishing branch.

2) If we got this far, then I need to know when to use the branch vs. the trunk. From a user prespective this isn't too big a deal, a ZMI interface, cookies, etc ... That's all ok ... The problem is that by the time the request comes to the storage side of things, that information is most likely gone (Don't know though, I'm still learning). Then there's the whole caching thing that would probably complicate things.

2) Message logging: CMF "comments" when doing a workflow action could go straight into SVN's logging, again how this might be accomplished isn't clear as of yet.

This is all for the future though, I'm nowhere near this yet!!

Also, I'm thinking the cooler features may just have to wait for Zope 3 ... Implementing them in Zope might be too convoluted.

I should probably start a Wiki on this topic !

J.F.

-----Original Message----- From: zope-dev-bounces@zope.org To: zope-dev@zope.org Sent: 13/04/2004 8:40 AM Subject: Re: Zope + Ape + Subversion (was: RE: [Zope-dev] Using a truely revis ion based storage for Zope ?)

On 12 Apr 2004 at 18:03, Jean-Francois.Doyon@CCRS.NRCan.gc.ca wrote:

...

Anyways, I'm just rambling by now ... Comments, thoughts and

constructive

...

criticism welcome !

This sounds wonderful!

Regarding versions. Perhap you can have a control panel setting that selects the branch to publish "by default". The ZMI could offer another mechanism that uses cookies to select the desired "branch or version" That is, a branch name or "PREV, HEAD" etc.

But, I use jEdit and FTP for most of my Zope work, so would really like to have a way to access version/branches via FTP. Maybe that can be done by having pseudo

subdirectories named after revisions or branches. So /myFolder/.Prev/index_html would do what you expect (note you can't save to old versions anyway ..)

François, we did a versioning Product here at redCOR which is about to go into production. It is "served" trough DCWorkflow state changes. It stores a series of objects per versioned object. One for each version. No diff's or such. I would be very much interested to adapt it such that it uses an APE/svn augmented storage for it. Robert