[Grok-dev] Re: letting Grok take over the site

Brandon Craig Rhodes brandon at rhodesmill.org
Mon Sep 17 22:39:45 EDT 2007


Shane Hathaway <shane.list.grok-dev at hathawaymix.org> writes:

> It would probably be much simpler and safer to modify traversal
> instead of replace the ZODB root object.

For the applications we are likely to write in my shop, we will
probably try to design so that the ZODB will not exist.  We already
have instantiations of the objects we need, usually in several places
already, with consequent duplication of information and possible
inconsistency!  All of the information that we could want already
lives in Oracle databases, PostgreSQL databases, LDAP directories, and
even a stray MySQL database here and there.

The last thing we need is to introduce another data store!  And with
another data stores comes something else that has to be backed up and
restored if we want our applications back intact.

We just need to wrap readily-available information and present it for
display and manipulation on the web.

So my personal goal, as soon as I understand enough about Zope 3 (and
it's taking time to figure out how it all works!), is to make every
object "not persistent" in the sense of "not stored in the ZODB".

My goal is for Zope to start up, create a single instance of my
Application object, and for my own Application's __init__ method to
take care of creating on-the-fly any virtual containers or other
objects it needs to run.  Thus installing my application on a new box
would mean only checking out the source code, and perhaps filling out
a configuration file with passwords into the data stores that lie
behind the web app.

-- 
Brandon Craig Rhodes   brandon at rhodesmill.org   http://rhodesmill.org/brandon


More information about the Grok-dev mailing list