On Feb 12, 2007, at 3:19 PM, Ross Patterson wrote:

In the design I've been doing with the CA I keep running into
a... "design pattern" (for lack of a better term) where I want to
retrieve persistent objects using exactly the kind of registry lookup
that the CA provides.  In the past I would have used a catalog to
perform such lookups with sufficient performance.

Registering the objects themselves as adapters where I need to do
these lookups would be tres nifty but I don't know how well the CA
scales.  Doing this through the CA would result in a huge
proliferation of the number of adapter registrations.  Does anyone
know how well this scales?  Does the performance of the CA lookups
depend on the number of registrations being small relative to the
number of content objects or users of an application?

The CA registry wasn't designed for huge numbers of components. In particular, there is extensive caching done that could lead to excessive memory usage if there were a very large number of components. If you are ultimately looking up data, a traditional data-indexing approach is more appropriate.


Jim Fulton                      mailto:[EMAIL PROTECTED]                Python 
CTO                             (540) 361-1714                  
Zope Corporation        http://www.zope.com             http://www.zope.org

Zope3-dev mailing list
Unsub: http://mail.zope.org/mailman/options/zope3-dev/archive%40mail-archive.com

Reply via email to