[Standards] LAST CALL: XEP-0216 (XMPP Intermediate IM Server 2008)

Mridul Muralidharan mridul at sun.com
Fri Jun 1 12:49:08 CDT 2007


Peter Saint-Andre wrote:
> Mridul Muralidharan wrote:
>> JD Conley wrote:
>>>> PSA Wrote:
>>>>> Right. So I think you're arguing that we should make JID Escaping
>>>>> required in Basic Server 2008.
>>>> I'm a bit confused. I actually thought this was a client-side XEP.
>>>
>>> Well, the server has to process it as well, as it has to send it to the
>>> external service (LDAP, IMAP, whatever) un-XEP-escaped and then escaped
>>> in whatever format the service needs.
>>>
>>> -JD
>>
>> It mostly affects clients, true.
>> But to identify the backend user, we use the 'node' since in most 
>> cases authorization identity == authentication identity.
>> So what JD mentions is exactly one of the reasons we ended up with 
>> issues.
>>
>> Another being how we look up into the backend store : having '\' in 
>> the uid is not good if you are writing to a file system as a backend, 
>> with node as a key and support windows. That way, JID escaping makes 
>> us require double transformation for those ...
> 
> Well the \ character is allowed by nodeprep. Are you suggesting that 
> nodeprep be changed to disallow the \ character?
> 
> Peter
> 

I am sure as soon as we disallow it, there will be some use of it ;)
Like what JD Conley mentioned, it just results in implementation 
specific processing at server side (same applies for /, *, ?, etc for 
example).
Disallowing might not be a good idea - but server awareness and support 
for jid escaping xep might be a good idea.
Ofcourse, this again applies as an implementation detail - like for 
example, if you use some other backend instead of filesystem (while 
running in a server pool for example), the above does not apply in our case.

Regards,
Mridul


More information about the Standards mailing list