[Standards-JIG] Historical XEPs

Jean-Louis Seguineau jean-louis.seguineau at laposte.net
Fri Nov 17 07:46:13 UTC 2006


I agree with Remko on the point that a full fledged PubSub approach may
sometime be too much for simple tasks. 

Referring to StPeter recent blog post on scalability, I would say we need to
adapt and provide appropriate solution for small and large tasks. 

Clients and even server components use private storage for different
purposes because it is simple, quick to implement and work in many odd
cases. Obviously as Maciek points out it does not provide all the
flexibility associated with a true DB, but it does the jib at simple things
like user preferences, configs, simple XML doc store, etc... 

Private storage is easy to implement and bodes well for quick and agile
solutions. It allows putting something out quickly and leaving the developer
time to evaluate if a more complex solution is appropriate.
Finally, in the same way "pull" cannot provide a universal solution in all
use cases, pubsub is not always appropriate. We need the flexibility. IMO
private storage and pubsub/PEP address different use cases and are
complementary. 

I am not in favor of deprecating XEP-0049  Private XML Storage.

My $0.02

Jean-Louis

-----Original Message-----
Message: 2
Date: Thu, 16 Nov 2006 23:36:11 +0100
From: Remko Troncon <remko at el-tramo.be>
Subject: [Standards-JIG] Re: Historical XEPs
To: XMPP Extension Discussion List <standards-jig at jabber.org>
Message-ID: <D3166DFC-CA53-4AAD-9CDB-43EF25059A1D at el-tramo.be>
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed


> Old XML Storage fails when you want to store a list of items (like
> bookmarks): you cannot easily add/remove one item. And more fun starts
> when you have more then one connected resource. You don't know if  
> other
> instance changed something (because there are no event notifications).
> Not to mention race conditions.

Other XEPs and RFC parts have the same requirements (adding/removing  
items, notifying connected resources of changes, ...), and don't use  
PEP either ?

Are we suddenly affraid to make protocols tailored to their goal,  
with an eye on making it easy to implement for clients, and are we  
going to start making nails out of every problem for our PubSub and  
Jingle hammers ? Besides, dragging in explicit creation of nodes and  
custom node configurations for PEP (i.e. whitelist nodes) is not much  
fun for a simple IM client to implement just to get private XML storage.

cheers,
Remko




More information about the Standards mailing list