On Sun, Mar 16, 2008 at 4:48 PM, Eugeny N Dzhurinsky <[EMAIL PROTECTED]> wrote: > On Sun, Mar 16, 2008 at 05:34:46PM +0200, Alex Lukin wrote: > > 1) User creates some content in his home node and then > > 2) copies it in incoming node for content editing person. Content editor > > checks it, may be modifies and > > 3) moves to some other place where content is published > > > > I think contents 1 and 2 are different but 2 and 3 are the same. UUID > > identifies some unique node. So I expect that UUIDs in 1 and 2 must be > > different, and in 2 and 3 equival. > > > > Questions: > > > > 1. Can I expect that move() operation does not change jcr:uuid but copy() > does? > > Yes, UUID wouldn't be changed on move operations, while copy creates new node > with new UUID, as stated in API docs > > > > 2. Can I use UUID to find my conent node independent of its location > (path)? > > Yes, you can. > > > > 3. How fast is getting node by UUID? > > As we found, accessing node by UUID usually works much faster than accessing > the node by the path, however you may find it useful to create some tests and > give them a try to ensure UUID access is really faster ;)
access by UUID is *the* fastest node retrieval method in jackrabbit ;) cheers stefan > > > > As I told before to Cristophe, path is not a "right thing" for primary key > > equivalent for OCM, and not "right thing" to refer unque objects. This use > > case shows why. So question is: Is UUID good candidate for such unique key? > > I think yes but I do not know what developers think of :) > > We are using a way of finding content by UUID - not paths in our application, > and it works just fine. > > -- > Eugene N Dzhurinsky >
