[Standards] UPDATED: XEP-0237 (Roster Versioning)

Curtis King cking at mumbo.ca
Thu May 14 17:00:30 UTC 2009


On 14-May-09, at 9:32 AM, Peter Saint-Andre wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 5/14/09 10:29 AM, Curtis King wrote:
>>
>> On 13-May-09, at 4:40 PM, Florian Zeitz wrote:
>>
>>> I'm also not really sure why anyone might ever want to use hashes.
>>
>> +1
>
> The Google Talk guys I talked with said they would do hashes. I'd like
> to see broader adoption, so I see no reason to discourage the hash
> approach. Besides, this is all out in implementation land

I think this naive on our part. By "our" I mean people who write,  
influence, edit, and comment on standards. From my experience this is  
where the most interoperability issues arise from, "it's  
implementation defined".

> -- the only
> thing that the spec needs to say is "the identifier is opaque to the
> client".

How many times have seen this ignored and will again in the future?  
The more we can make explicit the better and I find it also opens the  
door for more creative solutions.

The question this long discussion brings to mind is, "Is it really  
stupid developers at fault or us for not accepting how people think  
when implementing a spec?".

ck




More information about the Standards mailing list