On 22 December 2014 at 13:51, Sam Whited <[email protected]> wrote:
>
>
>
> On 12/22/2014 04:19 AM, Dave Cridland wrote:
> > Slightly confused by this. XEP-0191 is server-side enforced, so the
> > behaviour will be applied and controlled by the server, not the client.
>
> Gajim uses Privacy lists without the XEP-0191 frontend. Sorry, I was
> unclear there.
>
>
Ah, so it's not really doing XEP-0191 at all.


> > This would mean that probes still get sent, which seems inappropriate.
>
> My language probably needs to be tweaked (and updated in several other
> places in the XEP); outgoing probes (from the user to the blocked
> client) should remain the same (dropped so the user appears offline).
> Incoming probes should be handled like they currently are:
>
> From XEP-0191:
> > For presence stanzas (including notifications, subscriptions, and
> > probes), the server MUST NOT respond and MUST NOT return an error.
>
> The server must not respond, but it could still pass notifications on to
> the user.
>
>
As Kurt says, the spec is pretty clear - if a contact is blocked via
XEP-0191, then the user neither sends nor receives any stanzas to/from the
contact. I don't think we want to play spec-lawyer games here.


> > Otherwise we're in the slightly weird situation that we're predicating on
> > remote servers sending presence without a probe - this is quite possible,
> > but could lead to some very odd behaviour when this get out of sync.
> Also,
> > there's the RFC 3921 optimization; that reduces the presence to just
> > online/offline in some cases.
>
> Good point; I hate to potentially leak information by sending probes to
> the server. I'll have to think about this one.
>

Also bear in mind that XEP-0191 was designed to be a simple replacement to
XEP-0016, the observation being that with the exception of some extremely
rare cases, everything people actually used XEP-0016 for could be wrapped
up into XEP-0191 and XEP-0186. I don't really want to make this more
complex than it absolutely has to be.

So overall, I'm bound to entirely agree with Kurt's line of reasoning and
recommended resolutions.

Dave.

Reply via email to