[Standards] PEP private nodes

[Pedro Melo]

Hi,

On Apr 5, 2007, at 11:27 PM, Peter Saint-Andre wrote:
> Joe Hildebrand wrote:
>> We've been talking here about how to break out of this discussion  
>> about how to get private nodes correctly in PEP.  Here's another  
>> way to do it other than the two that we have so far.
>> Whenever a client publishes the first item to a node that ends in  
>> "+private", the server creates the node with a default access  
>> model of "whitelist", with an empty list.  The intent is that  
>> these nodes SHOULD stay private, but a client, at the explicit  
>> direction of a user MAY change the access model.  I'm actually  
>> flexible on this last point; i'd be happy to make that SHOULD a  
>> MUST, and take out the MAY.
>
> That works for private nodes but does not address all the worries  
> about difference of opinion regarding other nodes (geolocation or  
> whatever).

I brought the example of geolocation a couple of posts back. right  
now, I believe my use case would only be solved by per-item  
configuration, and I know that this was decided previously not to be  
included, so I'll drop this for now.

I think this (per item configuration) can be revived as an extension  
later on.

right now, per-item configuration seems to me to be the only safe  
(with assured privacy) way to implement private storage.

I though about flushing previous items whenever you change the  
configuration. This might also work, but i'm not sure yet.

Regarding Joe suggestion, the fact that later a client can  
specifically change the configuration and therefore change the access  
model to recently published items invalidates IMHO the "private" in  
private storage. So, I'm OK with the proposal provided that SHOULD  
becomes a MUST.

Thanks,
-- 
Pedro Melo
Blog: http://www.simplicidade.org/notes/
Jabber ID: melo at simplicidade.org
Use Jabber!