[Zope3-dev] Heads up: zope.app.intid and zope.app.keyreference are now op

[Uwe Oestermeier]

jim at zope.com wrote:
>They aren't guaranteed to be unique.  They are statistically
>very unlikely to conflict, but that chance of a conflict makes
>me nervous.  We tend to create a lot of these, so I think the chances
>for conflict are higher than in many other applications.
>A major component of variablility in guids is Mac or IP
>address, which wil lbe constant on servers that may need to generate
>millions of ids. <shrug>

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

http://www.ietf.org/internet-drafts/draft-mealling-uuid-urn-05.txt
>
>Certainly, it should be possible to implement GUID-nased key
>references, although you will need to store GUIDS on the objects.
>Feel free to try it.  I'd be interested to see how well it works over
>the long run.

We are using GUIDs with combined IP address, timestamp, and random numbers
in our application since three years without problems. But that's no
sufficient evidence. Are there other list members that use GUIDs?
Has anybody ever recognized problems of uniqueness?

Uwe