Benji York wrote:

You know probably Leach & Mealling's discussion of possible solutions:

We are using GUIDs with combined IP address, timestamp, and random

Uwe Oestermeier wrote:

They aren't guaranteed to be unique.  They are statistically
very unlikely to conflict, but that chance of a conflict makes
me nervous.

I'm a bit confused by this conversation. GUIDs (globally-unique identifiers) aren't neccesary for integer IDs. We just need locally-unique identifiers, which are much easier to generate.

Right, but to manage the integer ids, we need to be able to turn objects
into keys, since many objects are not valid keys.  The key reference
framework provides a way to generate keys from objects.  The key reference
adapter for persistent objects has the extremely inconvenient property
that you can't compute a key reference for a persistent object until
it has been added to a database.  Key references based on guids wouldn't
have this negative property.


Jim Fulton           mailto:[EMAIL PROTECTED]       Python Powered!
CTO                  (540) 361-1714  
Zope Corporation
Zope3-dev mailing list

Reply via email to