On Tue, Jan 22, 2008 at 12:19:52PM -0500, Christopher Warner wrote:
> On Tue, 2008-01-22 at 18:16 +0100, Brian Sutherland wrote:
> > On Tue, Jan 22, 2008 at 12:08:10PM -0500, Christopher Warner wrote:
> > > > We do something similar but with our Zope2 instances. With Zope3 we
> > > > don't have a ZODB and do all distribution using Debian packages.
> > >
> > > So how does this answer the question? He's talking about ZODB instances
> > > themselves without having to use Postgres or any other data solution
> > > other than the Data.fs itself.
> > >
> > > My recommendation is to either export all of the data that you want as
> > > object and do a mass import. Or to simply copy around the Data.fs which
> > > would probably not be the best solution but unless someone responds with
> > > a better way :-); the only solution.
> > Perhaps you could have 2 Data.fs in your instance mounted at different
> > points. One for code and the other for data. Then you just copy the code
> > Data.fs around and leave the data Data.fs in place.
> Eventually you'd have to create data to test the code. I guess the real
> issue is once you're complete with all of the code if you should begin a
> move into production to input the data there. The only problem with this
> is that doing so really defeats the purpose of production. I guess it
> would make sense to have a "stage to production" sort of setup where
> before you hit production you have it in staging for data entry. Then
> from there push it into production.
We normally move the data one way and the code the other. Code moves
from staging to live and we also regularly copy the data from live to
staging to make sure our code works with real world data.
The client adds data in live and often tests things in staging.
> Christopher Warner
Zope3-users mailing list