On 8/31/11 4:40 AM, Alexander Holler wrote:
> Am 31.08.2011 04:29, schrieb Peter Saint-Andre:
>> On 8/16/11 3:44 PM, Kevin Smith wrote:
>>> On Tue, Aug 16, 2011 at 10:35 PM, Peter
>>> Saint-Andre<[email protected]>  wrote:
>>>> On 8/16/11 2:38 AM, Kevin Smith wrote:
>>>>> On Mon, Aug 15, 2011 at 5:28 PM, Peter
>>>>> Saint-Andre<[email protected]>  wrote:
>>>>>> While completing my edits on XEP-0045, I noticed again the feature
>>>>>> for
>>>>>> requesting a unique room name. This feature seems unnecessary because
>>>>>> the client can simply generate a unique room name on its own
>>>>>> (e.g., by
>>>>>> using a UUID). Shall we remove it?
>>>>>
>>>>> I wasn't convinced there was a compelling reason last time, and I'm
>>>>> still not (I note again that server generated names can achieve more
>>>>> than just uniqueness), to remove this completely (It's an
>>>>> already-deployed feature of a Draft standard),
>>>>
>>>> Is it actually in use?
>>>
>>> ISTR from the discussions previously that Gajim, at least, has
>>> deployed it.
>>
>> Lot of good that does them, if no servers support it. :)
>>
>> In any case, I'm comfortable moving this to a separate spec.
> 
> That feature doesn't make much sense because the service can't guarantee
> that the returned (unique) ID is still unique when the client starts
> using it. If the service want's to guarantee that, the ID has to be
> reserved (maybe for a limited time). So I would suggest too to remove
> the feature.

Right, that's the traditional argument against the feature.

Peter

-- 
Peter Saint-Andre
https://stpeter.im/


Reply via email to