Jean-Francois.Doyon@CCRS.NRCan.gc.ca wrote at 2004-2-7 20:34 -0500:
Once again, thanks for the good points. Obviously one wouldn't want to create a catalog index on an attribute that just stored references ... that'd be kind of silly, and I'm sure I'd know better :) Security is pretty obviously a concern. As for acquisition, I guess the traversal would happen from the actual object's (Being referenced) point in the tree
In Zope2, there is no "actual point in the tree". The object has no notion where it is in the tree. All aspects with respect to "location inside the tree" are implemented by context wrappers and determined dynamically on access. That said: there is no distinction whatsoever between a primary reference to an object and secondary references. All references behave identically. Especially, the same object may have different URLs, dependent on how it is accessed.
, not the reference ?
As Zope2 cannot distinquish between a primary reference and a secondary reference and as the object has no notion of "its location", acquisition works with the reference you are currently using (provided the dereferencing process performs automatic acquisition wrapping). -- Dieter