[ejabberd] Result Set Management with PubSub

Yiorgis Gozadinos ggozad at jarn.com
Sat Jan 7 17:35:31 MSK 2012


We are currently developing a web based application that would rely heavily on PubSub. Some cases involve a large number of nodes (and collections) with a large number (>1k) of items that are to be persisted. Our problem is that the web clients need to fetch updates from these nodes each time a user logs-in. These can be large and we are trying to find a way to load them on demand, i.e. fetch the first N items and when necessary fetch some more later.
From what I gathered RSM support exists for MUC but not PubSub. What are the alternatives we have? I thought of:

1) Is it possible to get item ids without payloads? That would allow us to a large number of recent item ids compare to what we have and fetch items later.
2) Using a SQL storage and writing a component with direct access to the DB to do a minimal RSM.

Is there some other solution? Are there any plans to implement RSM for PubSub in the near future? Are there ways for us to help (financially) the development?
Thanks for making such an awesome product!

Yiorgis Gozadinos

More information about the ejabberd mailing list