[Standards-JIG] block + privacy lists again

[Mridul]

Hi,

  Please see inline.

Regards,
Mridul

Peter Saint-Andre wrote:
> Mridul just had a chat about the interaction between block lists and 
> privacy lists, which resulted in the following conclusions and questions:
>
> 1. Modifications to the block list result in changes only to the 
> default list and only to the jid+deny items in that list. This means 
> that if a privacy list client makes a change to anything except a 
> jid+deny item, that change will not be reflected in the block list. 
> This also means that if a block list client unblocks an item or 
> unblocks all items, that will affect only the jid+deny items in the 
> privacy list. The the concept is that the jid+deny entries must be in 
> sync between the block list and the default privacy list, but other 
> items in the default privacy list are not affected by the block list.
>
> 2. It's not clear what the priority of blocked (jid+deny) items should 
> be in the privacy list. IMHO they should come first in the privacy 
> list, but that's only a recommendation, not a requirement.
>
> 3. Some of this might be easier if we allowed layering of privacy 
> lists, as has been discussed in the past. That is, we could have one 
> privacy list for jid+deny items and then layer other privacy lists on 
> top of that. So that's an item for further discussion.

I was referring to only blocking list layering with privacy list + 
rosters though, not arbitrary set of privacy lists :-)
But I guess the idea could be extended or reduced either way.

>
> 4. If we did invisibility via XEP-0186 and blocking via XEP-0191, how 
> many people would really need privacy lists?
>
> 5. Should we relax the rule in XEP-0016 that says changes to the 
> default list MUST be forbidden if the default list is in use by other 
> resources? Blocking gets around this by sending updates.

Currently, a roster push has the change also sent across, not just that 
the roster was changed : privacy list push on the other hand just say 
that the list was modified - that could be modified and this requirement 
relaxed ?

>
> Mridul, did I miss anything?

I forgot to mention - named block lists ? Just an idea actually :
Different block lists which are 'named' (same semantics as the current 
single list). So client can apply a 'work' or 'home' or 'mobile' block 
list based on name.
This might be mapped to different privacy list as impl.


Thanks,
Mridul

>
> Peter
>