[Standards] POP/PEP/pubsub

Joe Hildebrand hildjj at gmail.com
Thu May 31 16:11:27 UTC 2007


On May 2, 2007, at 10:59 AM, Ian Paterson wrote:

> There will be many thousands of different use cases, big and small,  
> that are likely to be invented over the lifetime of the new  
> iq:private. IMHO we need to offer the maximum data structure  
> flexibility rather than try and tailor the solution to a few  
> foreseeable use cases. What is sure is that some data will be  
> (borrowing language from PubSub) item-atomic and some will be node- 
> atomic. Some will require collections (heirarchical trees of  
> nodes). Pubsub already addresses this "minimum" set of data  
> structures.

Perhaps what we ought to do is put a recommendation in PEP about what  
item ID to use when you mean "Please store a single instance of  
this".  For example:

<suggested>
For some namespace nodes, it is desirable to have at most one item  
that has been published at any given time.  For example, a particular  
client may want to store its preferences using a node name that is a  
namespace controlled by that client author.  When this pattern is  
desired, all publishers SHOULD use the item ID "UNIQUE" to ensure  
that the publication of a new item will overwrite any old item.
</suggested>

I don't care what the string is.
UNIQUE
SINGLE
ONE
AMO (at most once)

-- 
Joe Hildebrand





More information about the Standards mailing list