Allowing multiple levels of a tree structure to be pre-fetched would allow more efficient traversal of the tree structure than simply iterating one level at a time. The implementation could use relational vendor specific features, I know Oracle supports such recursive traversals.
David Jordan Object Identity, Inc. -----Original Message----- From: Marco Schulze [mailto:[EMAIL PROTECTED] Sent: Monday, December 19, 2005 7:23 AM To: [email protected] Cc: [EMAIL PROTECTED] Subject: Re: fetch-depth + recursion-depth Jörg von Frantzius wrote: > After having thought about it for a while, I started to doubt whether > we really need a "recursion-depth" number here. Can you think of a > case where you'd need any value greater than 1 here? We have a case in which we need a "recursion-depth" greater than 1: We use a tree of objects (all of the same class, self-referencing via a Collection), where the GUI loads this tree initially with recursion-depth=2 in order to immediately show the first level of the tree expanded and the second level ready - with already knowing whether an expand-icon needs to be visible or not. In other words: we load always one level more in the tree than is visible for avoiding the expand-icon to be visible even though there are no children. > So I'd rather call this a boolean flag "recurse-class-cycles". IMHO, the implementation work for a boolean flag is the same as for a number, so the flexibility of a count should be preferred. > I can't see why the spec should be changed for this rather than > extended. Your requirement for a cycle-detection probably doesn't > invalidate existing requirements for an absolute fetch-depth. With "changed" I meant "extended".
