On Oct 5, 2011, at 11:46 PM, Kim Alvefur wrote:

>> 1. Is this specification needed to fill gaps in the XMPP protocol
>> stack or to clarify an existing protocol?
> Yes.
> 
>> 2. Does the specification solve the problem stated in the introduction
>> and requirements?
> Yes.
> 
>> 3. Do you plan to implement this specification in your code? If not,
>> why not?
> I just pushed some work I did on a plugin for Prosody, updating it to
> the latest spec. Only the catalog serving part though.
> 
>> 4. Do you have any security concerns related to this specification?
> If this counts as security concerns:
> 
> http://xmpp.org/extensions/xep-0258.html#label-catalog
>> If catalog is restrictive, as indicated by the restrict attribute with
>> value of true, the client SHOULD use one of the labels (or no label)
>> offered by the catalog.
> [...]
>> An item may have no security label. Such an item offers a choice of
>> sending a stanza without a label.
> 
> Does that mean that with an non-restrictive catalog, a client may send
> no label even if there's no label-less item? Ie can restrictive=false be
> interpreted as pretending there's a item without label?

The restrict flag was intended to inform a client capable of producing custom 
labels that it should not offer a custom label choice in the context which the 
catalog was provided.  I should clarify the spec that restrict flag just means 
this, removing any semantic overload that might be suggested.

There's also the question of whether a client is or not allowed to send stanzas 
without any label.  This question shouldn't be answered by the restrict 
attribute as that would be a semantic overload.

In initially writing the spec, my view was that a client should always be 
allowed to send a stanza with no label, mostly for backward compatibility with 
clients which don't support XEP 258.

Most XEP258 clients I've seen generally allow generation of messages without 
labels.  I suggest that XEP 258 that this is generally allowed unless someone 
can make a damn good argument that we should add a flag indicating to the 
client they should not generate messages with no label (unless the catalog 
contains an empty item).

> 
>> 5. Is the specification accurate and clearly written?
> It's a bit heavy to read, and could benefit from some formating.
> Currently it feels like a wall of text at some places.

I'll see if I can find a way to make it an easier read.

> For example, attribute names would benefit from some emphasis, like in
> the quoted line above.

I can certainly do that.

-- Kurt

> 
> -- 
> Kim Alvefur <[email protected]>
> 

Reply via email to