[Zope-CMF] Re: CMFUid isn't generating real uids

Dario Lopez-Kästen dario at ita.chalmers.se
Fri Aug 20 10:24:15 EDT 2004


Gregoire Weber wrote:

> Salut Jean-Marc,
> 
> 
>>I think that the confusion comes from the fact being "unique" is not a property that something has intrinsically (as for an object being "red", "blue", ...) even though the expression 'X is unique ' is a one-place predicate. - but in fact it always has to be taken in relation to a policy that needs to be enforced throughout. In the context of a "unique ID" the policy can be enforced inside a same folder, in the case of a universally unique ID the policy must be enforced inside the context that the term "universal" refers too -  So "uniqueness" is more of the nature of laws, regulations, concepts, habits (...) than of the nature of a quality, and this is why the best UID generator is the one that can best enforce a given policy.
> 
> 
> Well stated!
> 
> I'll updating the README.txt this week by something similar to that:
> 
>    The unique ids generated by CMFUid *are unique by CMF instance*.
> 
>    On imports existing CMFUid uids get deleted to avoid doubling
>    uids.
> 
> Gregoire

http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/213761

for a micro-discussion on where to get references to UUID's.

Or we could just use a base counter, and base the UID on an OID, that we 
get ourselves assigned either from IANA, ANSI or BSI,

http://www.alvestrand.no/objectid/

or if we need not be that formal, from a regsitry we create ourselves:

oid.zope.org

or

oid.plone.org

or whatever.  OIDs could effetively be served on a first come, first 
served basis.


Just some random thoughts.

/dario

-- 
-- -------------------------------------------------------------------
Dario Lopez-Kästen, IT Systems & Services Chalmers University of Tech.


More information about the Zope-CMF mailing list