[Zope] Son of ZSyncer

Dennis Allison allison@sumeru.stanford.EDU
Mon, 28 Jul 2003 21:56:03 -0700 (PDT)


We are running many independent instances of Zope front-ended by Pound.
Some are Production machines  P0, P1, P2, ...   One is a development
machine E.  We also have a golden master machine S that contains the 
released version of the code that is shared by P0, P1, etc.  Not all 
of the instance code is the same.

Our development workflow is to create and test on E, then sync that
code with the code on S.  (Updates are not localized -- bug fixes 
and feature additions).  On a regular schedule we update the production 
systems, but only a portions of the tree.

And we're from that part of the world that wants everything scripted 
and repeatable without any hand work....   My posting was motivated 
by the prospect of having to examine and hand synchronize several hundred
directories :-(.


-d



On Mon, 28 Jul 2003, Paul Winkler wrote:

> On Mon, Jul 28, 2003 at 01:09:45AM -0700, Dennis Allison wrote:
> > ZSyncer is a cool product but....
> > 
> > It isn't recursive and it's not scriptable.
> 
> Oh, I have a lot more complaints with it than that ;-)
> Check the CVS sources, read the TODO that I've been piling ideas
> into.  
> 
> > I need a product that synchronizes client from a parent site recursively
> > through the folder structure, but configuable so that certain folders and
> > their descendants are not synced.  
> 
> Yes, recursion would be very useful in some situations and is on the
> TODO list. I hadn't thought about configuration such as you suggest.
> Can you elaborate on your requirements a bit? 
> The first idea that occurs to me is to add a text entry to the
> sync view in which the user can enter an expression (TALES?) that needs
> to pass before recursing into a subobject.
> 
> > I also need the difference test to be
> > something a bit more robust than date/time since some clients get hot
> > patched and can possibly get out of sync.  (The fix was done to the master
> > but not distributed, but the secondary got hot patched with a hotfix...)
> 
> yes, date/time comparison is very naive and not at all reliable.
> I've outlined many possible improvements in the TODO. I've had no feedback
> on them yet, comments and better ideas would be very welcome.
>  
> > Suggestions?
> 
> Yes - help us improve zsyncer! :-)
> 
> If you have time to work on any of this stuff, send patches to the
> maintainers listed at Sourceforge. I'd be very interested in testing
> any patches and I can check them in for you.  If that gets too cumbersome,
> we should be able to get you writeable CVS access.
> 
> -- 
> 
> Paul Winkler
> http://www.slinkp.com
> Look! Up in the sky! It's THE SINGLE MOTHER!
> (random hero from isometric.spaceninja.com)
> 
> _______________________________________________
> Zope maillist  -  Zope@zope.org
> http://mail.zope.org/mailman/listinfo/zope
> **   No cross posts or HTML encoding!  **
> (Related lists - 
>  http://mail.zope.org/mailman/listinfo/zope-announce
>  http://mail.zope.org/mailman/listinfo/zope-dev )
>